Skip to main content

Resultados relacionados

Resultados totales (Incluyendo duplicados): 54
Encontrada(s) 1 página(s)

DIGITAL.CSIC. Repositorio Institucional del CSIC
Publication Contribución de congreso (ConferenceOutput)

Variable impedance control in Cartesian latent space while avoiding obstacles in null space

  • Parent Alonso, David
  • Colomé, Adrià
  • Torras, Carme
Trabajo presentado en la IEEE International Conference on Robotics and Automation (ICRA), conferencia virtual celebrada del 31 de mayo al 31 de agosto de 2020, Human-robot interaction is one of the keys of assistive robots. Robots are expected to be compliant with people but at the same time correctly perform the tasks. In such applications, Cartesian impedance control is preferred over joint control, as the desired interaction and environmental feedback can be described more naturally, and the force to be exerted by the robot can be readily adjusted. This paper addresses the problem of controlling a robot arm in the operational space with variable stiffness so as to continuously adapt the force exerted in each phase of motion according to the precision requirements. Moreover, performing dimensionality reduction we can separate the degrees of freedom (DoF) relevant for the task from the redundant ones. The stiffness of the former can be adjusted constantly to achieve the required accuracy, while task-redundant DoF can be used to achieve other goals such as avoiding obstacles by moving in the directions where accuracy is not critical. The designed method is tested teaching the robot to give water to drink to a model of human head. Our empirical results demonstrate that the robot can learn precision requirements from demonstration. Furthermore, dimensionality reduction is proved to be useful to avoid obstacles., This work was partially developed in the context of the project CLOTHILDE ("CLOTH manIpulation Learning from DEmonstrations"), which has received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (Advanced Grant agreement No 741930). This work is also supported by the Spanish State Research Agency through the María de Maeztu Seal of Excellence to IRI MdM-IP-2018-05., Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publication Contribución de congreso (ConferenceOutput)

Sample-efficient robot motion learning using Gaussian process latent variable models

  • Delgado-Guerrero, Juan Antonio
  • Colomé, Adrià
  • Torras, Carme
Trabajo presentado en la IEEE International Conference on Robotics and Automation (ICRA), conferencia virtual celebrada del 31 de mayo al 31 de agosto de 2020, Robotic manipulators are reaching a state where we could see them in household environments in the following decade. Nevertheless, such robots need to be easy to instruct by lay people. This is why kinesthetic teaching has become very popular in recent years, in which the robot is taught a motion that is encoded as a parametric function - usually a Movement Primitive (MP)-. This approach produces trajectories that are usually suboptimal, and the robot needs to be able to improve them through trial-and-error. Such optimization is often done with Policy Search (PS) reinforcement learning, using a given reward function. PS algorithms can be classified as model-free, where neither the environment nor the reward function are modelled, or model-based, which can use a surrogate model of the reward function and/or a model for the dynamics of the task. However, MPs can become very high-dimensional in terms of parameters, which constitute the search space, so their optimization often requires too many samples. In this paper, we assume we have a robot motion task characterized with an MP of which we cannot model the dynamics. We build a surrogate model for the reward function, that maps an MP parameter latent space (obtained through a Mutual-information-weighted Gaussian Process Latent Variable Model) into a reward. While we do not model the task dynamics, using mutual information to shrink the task space makes it more consistent with the reward and so the policy improvement is faster in terms of sample efficiency., This work was partially developed in the context of the project CLOTHILDE ("CLOTH manIpulation Learning from DEmonstrations"), which has received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (Advanced Grant agreement No 741930). This work is also supported by the Spanish State Research Agency through the María de Maeztu Seal of Excellence to IRI MDM-2016-0656., Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publication Contribución de congreso (ConferenceOutput)

Dynamic Cloth Manipulation with Deep Reinforcement Learning

  • Jangir, Rishabh
  • Alenyà, Guillem
  • Torras, Carme
Trabajo presentado en la IEEE International Conference on Robotics and Automation (ICRA), conferencia virtual celebrada del 31 de mayo al 31 de agosto de 2020, In this paper we present a Deep Reinforcement Learning approach to solve dynamic cloth manipulation tasks. Differing from the case of rigid objects, we stress that the followed trajectory (including speed and acceleration) has a decisive influence on the final state of cloth, which can greatly vary even if the positions reached by the grasped points are the same. We explore how goal positions for non-grasped points can be attained through learning adequate trajectories for the grasped points. Our approach uses few demonstrations to improve control policy learning, and a sparse reward approach to avoid engineering complex reward functions. Since perception of textiles is challenging, we also study different state representations to assess the minimum observation space required for learning to succeed. Finally, we compare different combinations of control policy encodings, demonstrations, and sparse reward learning techniques, and show that our proposed approach can learn dynamic cloth manipulation in an efficient way, i.e., using a reduced observation space, a few demonstrations, and a sparse reward., This work has been supported by the ERC project Clothilde (ERC-2016- ADG-741930), the HuMoUR project TIN2017-90086-R (AEI/FEDER, UE) and by the AEI through the María de Maeztu Seal of Excellence to IRI (MDM-2016-0656), Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publication Contribución de congreso (ConferenceOutput)

Automatic learning of cognitive exercises for socially assistive robotics

  • Suárez Hernández, Alejandro
  • Andriella, Antonio
  • Taranović, Aleksandar
  • Segovia-Aguas, Javier
  • Torras, Carme
  • Alenyà, Guillem
Trabajo presentado en el 30th IEEE International Conference on Robot and Human Interactive Communication (RO-MAN), celebrada de forma virtual del 8 al 12 de agosto de 2021, In this paper, we present a learning approach to facilitate the teaching of new board exercises to assistive robotic systems. We formulate the problem as the learning of action models using Boolean predicates, disjunctive preconditions, and existential quantifiers from demonstrations of successful exercise executions. To be able to cope with exercises whose rules depend on a set of features that are initialized at the beginning of each play-out, we introduce the concept of dynamic context. Furthermore, we show how the learnt knowledge can be represented intuitively in a graphical interface that helps the caregiver understand what the system has learnt. As validation, we conducted a user study in which we evaluated whether and to which extent different types of feedback can affect the subjects¿ performance while teaching three types of exercises: (1) sorting numbers; (2) arranging letters; and (3) reproducing shapes sequences in reversed order. The results suggest that textual and graphical feedback are beneficial., A. Andriella, C. Torras and A. Suarez-Hernandez were partially funded ´ by the European Union´s Horizon 2020 under ERC Advanced Grant CLOTHILDE (no. 741930), G. Alenya by the EU H2020 research and ` innovation programme IMAGINE (no. 731761) and J. Segovia-Aguas by the programme TAILOR (no. 952215). The work was partially supported by the Spanish State Research Agency through the Mar´ıa de Maeztu Seal of Excellence to IRI (MDM-2016-0656).
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publication Contribución de congreso (ConferenceOutput)

Ethics for social robotics: A critical analysis

  • Pareto Boada, Júlia
  • Román Maestre, Begoña
  • Torras, Carme
Trabajo presentado en el The Road to a successful HRI: AI, Trust and ethicS - TRAITS Workshop, celebrado online y presencial el 11 de marzo de 2022. Celebrado conjuntamente con ACM/IEEE International Conference on Human-Robot Interaction (HRI 2022), Social robotics development for the practice of care and European prospects to incorporate these AI-based systems in institutional healthcare contexts call for an urgent ethical reflection to (re)configurate our practical life according to human values and rights. Despite the growing attention to the ethical implications of social robotics, the current debate on one of its central branches, social assistive robotics (SAR), rests upon an impoverished ethical approach. This paper presents and examines some tendencies of this prevailing approach, which have been identified as a result of a critical literature review. Based on this analysis of a representative case of how ethical reflection is being led towards social robotics, some future research lines are outlined, which may help reframe and deepen in its ethical implications., This work has been partially supported by the Spanish Ministry of Science and Innovation under a FPI scholarship for predoctoral contracts for the training of doctors (PRE2018-084286), and by the European Union Horizon 2020 Programme under grant agreement no. 741930 (CLOTHILDE).
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publication Contribución de congreso (ConferenceOutput)

Knowledge representation to enable high-level planning in cloth manipulation tasks

  • García-Camacho, Irene
  • Borràs, Julia
  • Alenyà, Guillem
Trabajo presentado en la ICAPS Workshop on Knowledge Engineering for Planning and Scheduling, celebrada en Singapore, el 15 de junio de 2022, Cloth manipulation is very relevant for domestic robotic tasks, but it presents many challenges due to the complexity of representing, recognizing and predicting the behaviour of cloth under manipulation. In this work, we propose a generic, compact and simplified representation of the states of cloth manipulation that allows for representing tasks as sequences of states and transitions semantically. We also define a Cloth Manipulation Graph that encodes all the strategies to accomplish a task. Our novel representation is used to encode two different cloth manipulation tasks, learned from an experiment with human subjects manipulating clothes with video data. We show how our simplified representation allows to obtain a map of meaningful steps that can serve to describe cloth manipulation tasks as domain models in PDDL, enabling high-level planning. Finally, we discuss on the existing skills that could enable the sensory motor grounding and the low-level execution of the plan., The research leading to these results receives funding from the European Research Council (ERC) from the European Union Horizon 2020 Programme under grant agreement no. 741930 (CLOTHILDE: CLOTH manIpulation Learning from DEmonstrations); by MCIN/ AEI /10.13039/501100011033, Spain, under the project CHLOE-GRAPH (PID2020-119244GB-I00); by MCIN/ AEI /10.13039/501100011033, Spain, and by the “European Union NextGenerationEU/PRTR, Spain, under the project COHERENT (PCI2020-120718-2); and by the European Commission- NextGenerationEU, through CSIC’s Thematic Platforms (PTI+ Neuro-Aging).
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publication Contribución de congreso (ConferenceOutput)

Proactive learning of cognitive exercises with a social robot

  • Suárez Hernández, Alejandro
  • Andriella, Antonio
  • Alenyà, Guillem
  • Torras, Carme
Trabajo presentado en el ICRA Workshop on Prediction and Anticipation Reasoning in Human Robot Interaction (PAR-HRI) (PAR-HRI), celebrado en Philadelphia, PA (Estados Unidos), el 27 de mayo de 2022, We introduce INtuitive PROgramming 2 (IN-PRO2), an improvement over our previous INPRO framework for learning board exercises via demonstrations. INPRO2 makes use of our Online Action Recognition through Unification (OARU) algorithm, which maintains and extends as needed a library of STRIPS action schemata that represent the dynamics, rules and goal of the exercise. OARU operates on a sequence of states shown by the user. Each state transition is either used to learn a new action, or is recognized as an instance of one action currently present in the library, possibly refining it. We have extended OARU to support negative examples (i.e. invalid moves that show forbidden state transitions) in order to increase the complexity of the exercises that can be learned. This new OARU¿s feature is exploited through another crucial element of INPRO2: its ability to proactively ask for the legality of certain moves to the user in critical situations, and fix overly permissive actions. We show an example of a typical INPRO2 learning session. We also outline a plan for a user study that will serve to assess the proactive behavior of the robot., The research leading to these results receives funding from the EU H2020 Programme under grant agreement ERC–2016–ADG–741930 (CLOTHILDE); ASH acknowledges the financial support of the Apadrina la Ciencia association and of the Ford Spanish branch
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publication Contribución de congreso (ConferenceOutput)

Reconstruction of sampled surfaces with boundary via Morse theory

  • Coltraro, Franco
  • Amorós, Jaume
  • Alberich-Carramiñana, Maria
  • Torras, Carme
Trabajo presentado en el Congreso Español de Informática Gráfica (CEIG), celebrado en Palma de Mallorca (España), del 3 al 6 de julio de 2023, We study the perception problem for garments (e.g. a pair of pants) using tools from computational topology: the identification of their geometry and position from point-cloud samples, as obtained e.g. with 3D scanners. We present a reconstruction algorithm based on Morse theory that proceeds directly from the point-cloud to obtain a cellular decomposition of the surface derived via a Morse function. No intermediate triangulation or local implicit equations are used, avoiding reconstruction-induced artifices. The results are a piecewise parametrization of the surface as a union of Morse cells, suitable for tasks such as noise-filtering or mesh-independent reparametrization, and a cell complex of small rank determining the surface topology. This algorithm can be applied to smooth surfaces with or without boundary, embedded in an ambient space of any dimension., This work was developed in the context of the project CLOTHILDE which has received funding from European Union’s Horizon 2020 research and innovation programme (grant agreement No. 741930). M. Alberich-Carramiñana is partially supported by the MCIN / AEI grant PID2019-103849GB-I00.
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

Quadratic dynamic matrix control for fast cloth manipulation

  • Caldarelli, Edoardo
  • Colomé Figueras, Adrià|||0000-0001-9715-4062
  • Ocampo-Martínez, Carlos|||0000-0001-9251-6044
  • Torras, Carme|||0000-0002-2933-398X
© 2023 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting /republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works, Robotic cloth manipulation is an increasingly relevant area of research, challenging classic control algorithms due to the deformable nature of cloth. While it is possible to apply linear model predictive control to make the robot move the cloth according to a given reference, this approach suffers from a large dimensionality of the state-space representation of the cloth models. To address this issue, in this work we study the application of an input-output model predictive control strategy, based on quadratic dynamic matrix control, to robotic cloth manipulation. To account for uncertain disturbances on the cloth's motion, we further extend the algorithm with suitable chance constraints. In extensive simulated experiments, involving disturbances and obstacle avoidance, we show that quadratic dynamic matrix control can be successfully applied in different cloth manipulation scenarios, with significant gains in optimization speed compared to standard model predictive control strategies. The experiments further demonstrate that the closed-loop model used by quadratic dynamic matrix control can be beneficial to the tracking accuracy, leading to improvements over the standard predictive control strategy. Moreover, a preliminary experiment on a real robot shows that quadratic dynamic matrix control can indeed be employed in real settings., This work was supported by the project CLOTHILDE (“CLOTH manIpulation Learning from DEmonstrations”), funded by the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (Advanced Grant agreement No 741930)., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

A virtual reality framework for fast dataset creation applied to cloth manipulation with automatic semantic labelling

  • Borràs Sol, Júlia|||0000-0002-1662-2037
  • Boix Granell, Arnau
  • Foix Salmerón, Sergi|||0000-0001-9249-6696
  • Torras, Carme|||0000-0002-2933-398X
© 2023 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works., Teaching complex manipulation skills, such as folding garments, to a bi-manual robot is a very challenging task, which is often tackled through learning from demonstration. The few datasets of garment-folding demonstrations available nowadays to the robotics research community have been either gathered from human demonstrations or generated through simulation. The former have the great difficulty of perceiving both cloth state and human action as well as transferring them to the dynamic control of the robot, while the latter require coding human motion into the simulator in open loop, i.e., without incorporating the visual feedback naturally used by people, resulting in far-from-realistic movements. In this article, we present an accurate dataset of human cloth folding demonstrations. The dataset is collected through our novel virtual reality (VR) framework, based on Unity’s 3D platform and the use of an HTC Vive Pro system. The framework is capable of simulating realistic garments while allowing users to interact with them in real time through handheld controllers. By doing so, and thanks to the immersive experience, our framework permits exploiting human visual feedback in the demonstrations while at the same time getting rid of the difficulties of capturing the state of cloth, thus simplifying data acquisition and resulting in more realistic demonstrations. We create and make public a dataset of cloth manipulation sequences, whose cloth states are semantically labeled in an automatic way by using a novel low-dimensional cloth representation that yields a very good separation between different cloth configurations., The research leading to these results receives funding from the European Research Council (ERC) from the European Union Horizon 2020 Programme under grant agreement no. 741930 (CLOTHILDE: CLOTH manIpulation Learning from DEmonstrations) and project SoftEnable (HORIZONCL4-2021-DIGITAL-EMERGING-01-101070600). Authors also received funding from project CHLOE-GRAPH (PID2020-118649RB-I00) funded by MCIN/ AEI /10.13039/501100011033 and COHERENT (PCI2020-120718-2) funded by MCIN/ AEI /10.13039/501100011033 and cofunded by the ”European Union NextGenerationEU/PRTR”., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

Fitting parameters of linear dynamical systems to regularize forcing terms in dynamical movement primitives

  • Stulp, Freek
  • Colomé Figueras, Adrià|||0000-0001-9715-4062
  • Torras, Carme|||0000-0002-2933-398X
© 2024 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works., Due to their flexibility and ease of use, Dynamical Movement Primitives (DMPs) are widely used in robotics applications and research. DMPs combine linear dynamical systems to achieve robustness to perturbations and adaptation to moving targets with non-linear function approximators to fit a wide range of demonstrated trajectories. We propose a novel DMP formulation with a generalized logistic function as a delayed goal system. This formulation inherently has low initial jerk, and generates the bell-shaped velocity profiles that are typical of human movement. As the novel formulation is more expressive, it is able to fit a wide range of human demonstrations well, also without a non-linear forcing term. We exploit this increased expressiveness by automating the fitting of the dynamical system parameters through optimization. Our experimental evaluation demonstrates that this optimization regularizes the forcing term, and improves the interpolation accuracy of parametric DMPs., This work was partially funded by the European ERC/EIC/Horizon projects with grant numbers 741930 (CLOTHILDE), 951992 (VeriDream), and 101070596 (euROBIN), as well the CSIC project 202350E080 (ClothIRI), and the MCIN/AEI project PID2020-118649RBI00 (CHLOE-GRAPH)., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

Evaluating the effect of theory of mind on people’s trust in a faulty robot

  • Rossi, Alessandra
  • Andriella, Antonio|||0000-0002-6641-6450
  • Rossi, Silvia
  • Torras, Carme|||0000-0002-2933-398X
  • Alenyà Ribas, Guillem|||0000-0002-6018-154X
© 2022 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes,creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works., The success of human-robot interaction is strongly affected by the people's ability to infer others' intentions and behaviours, and the level of people's trust that others will abide by their same principles and social conventions to achieve a common goal. The ability of understanding and reasoning about other agents' mental states is known as Theory of Mind (ToM). ToM and trust, therefore, are key factors in the positive outcome of human-robot interaction. We believe that a robot endowed with a ToM is able to gain people's trust, even when this may occasionally make errors. In this work, we present a user study in the field in which participants (N=123) interacted with a robot that may or may not have a ToM, and may or may not exhibit erroneous behaviour. Our findings indicate that a robot with ToM is perceived as more reliable, and they trusted it more than a robot without a ToM even when the robot made errors. Finally, ToM results to be a key driver for tuning people's trust in the robot even when the initial condition of the interaction changed (i.e., loss and regain of trust in a longer relationship)., This work has been supported by Italian PON R&I 2014-2020 - REACT-EU (CUP E65F21002920003); by the European Union’s Horizon 2020 under ERC Advanced Grant CLOTHILDE (no. 741930); by MCIN/ AEI /10.13039/501100011033 and by the ”European Union NextGenerationEU/PRTR” under the project ROBIN (PLEC2021-007859) and the project COHERENT (PCI2020-120718-2); by the ”European Union NextGenerationEU/PRTR” through CSIC’s Thematic Platforms (PTI+ Neuro-Aging) and by the European Union’s Horizon 2020 research and innovation programme PRO-CARED (no.801342) under Marie Skłodowska-Curie grant agreement (Tecniospring INDUSTRY) and the Government of Catalonia’s Agency for Business Competitiveness (ACCIO)., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

Reconstruction of sampled surfaces with boundary via Morse theory

  • Coltraro, Franco|||0000-0002-9149-950X
  • Amorós Torrent, Jaume|||0000-0002-3133-9985
  • Alberich Carramiñana, Maria|||0000-0003-2749-4875
  • Torras, Carme|||0000-0002-2933-398X
We study the perception problem for garments (e.g. a pair of pants) using tools from computational topology: the identification of their geometry and position from point-cloud samples, as obtained e.g. with 3D scanners. We present a reconstruction algorithm based on Morse theory that proceeds directly from the point-cloud to obtain a cellular decomposition of the surface derived via a Morse function. No intermediate triangulation or local implicit equations are used, avoiding reconstruction-induced artifices. The results are a piecewise parametrization of the surface as a union of Morse cells, suitable for tasks such as noise-filtering or mesh-independent reparametrization, and a cell complex of small rank determining the surface topology. This algorithm can be applied to smooth surfaces with or without boundary, embedded in an ambient space of any dimension., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

Ordinal inverse reinforcement learning applied to robot learning with small data

  • Colomé Figueras, Adrià|||0000-0001-9715-4062
  • Torras, Carme|||0000-0002-2933-398X
© 2022 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works., Over the last decade, the ability to teach actions to robots in a user-friendly way has gained relevance, and a practical way of teaching robots a new task is to use Inverse Reinforcement Learning (IRL). In IRL, an expert teacher shows the robot a desired behaviour and an agent builds a model of the reward. The agent can also infer a policy that performs in an optimal way within the limitations of the knowledge provided to it. However, most IRL approaches assume an (almost) optimal performance of the teaching agent, which might become unpractical if the teacher is not actually an expert. In addition, most IRL focus on discrete state-action spaces that limit their applicability to certain real-world problems such as within the context of direct Policy Search (PS) reinforcement learning. Therefore, in this paper we introduce Ordinal Inverse Reinforcement Learning (OrdIRL) for continuous state variables, in which the teacher can qualitatively evaluate robot performance by selecting one among the predefined performance levels (e.g. {bad, medium, good} for three tiers of performance). Once the OrdIRL has fit an ordinal distribution to the data, we propose to use Bayesian Optimization (BO) to either gain knowledge on the inferred model (exploration) or find a policy or action that maximizes the expected reward given the prior knowledge on the reward (exploitation). In the case of large-dimensional state-action spaces, we use Dimensionality Reduction (DR) techniques and perform the BO in the latent space. Experimental results on simulation and with a robot arm show how this approach allows for learning the reward function with small data., This work was developed in the context of the project CLOTHILDE ("CLOTH manIpulation Learning from DEmonstrations"), which has received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (Advanced Grant agreement No 741930)., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

Knowledge representation to enable high-level planning in cloth manipulation tasks

  • Garcia Camacho, Irene|||0000-0002-1210-7345
  • Borràs Sol, Júlia|||0000-0002-1662-2037
  • Alenyà Ribas, Guillem|||0000-0002-6018-154X
Cloth manipulation is very relevant for domestic robotic tasks, but it presents many challenges due to the complexity of representing, recognizing and predicting the behaviour of cloth under manipulation. In this work, we propose a generic, compact and simplified representation of the states of cloth manipulation that allows for representing tasks as sequences of states and transitions semantically. We also define a Cloth Manipulation Graph that encodes all the strategies to accomplish a task. Our novel representation is used to encode two different cloth manipulation tasks, learned from an experiment with human subjects manipulating clothes with video data. We show how our simplified representation allows to obtain a map of meaningful steps that can serve to describe cloth manipulation tasks as domain models in PDDL, enabling high-level planning. Finally, we discuss on the existing skills that could enable the sensory motor grounding and the low-level execution of the plan., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

Discovering SOCIABLE: using a conceptual model to evaluate the legibility and effectiveness of backchannel cues in an entertainment scenario

  • Andriella, Antonio|||0000-0002-6641-6450
  • Huertas García, Ruben
  • Forgas Coll, Santiago
  • Torras, Carme|||0000-0002-2933-398X
  • Alenyà Ribas, Guillem|||0000-0002-6018-154X
© 2020 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works., Robots are expected to become part of everyday life. However, while there have been important breakthroughs during the recent decades in terms of technological advances, the ability of robots to interact with humans intuitively and effectively is still an open challenge. In this paper, we aim to evaluate how humans interpret and leverage backchannel cues exhibited by a robot which interacts with them in an entertainment context. To do so, a conceptual model was designed to investigate the legibility and the effectiveness of a designed social cue, called SOCial ImmediAcy BackchanneL cuE (SOCIABLE), on participant’s performance. In addition, user’s attitude and cognitive capability were integrated into the model as an estimator of participants’ motivation and ability to process the cue. In working toward such a goal, we conducted a two-day long user study (N=114) at an international event with untrained participants who were not aware of the social cue the robot was able to provide. The results showed that participants were able to perceive the social signal generated from SOCIABLE and thus, they benefited from it. Our findings provide some important insights for the design of effective and instantaneous backchannel cues and the methodology for evaluating them in social robots., This work has been partially funded by the EU project SOCRATESH2020-MSCA-ITN-721619, by the European Union ́s Horizon 2020 underERC Advanced Grant CLOTHILDE (no. 741930), by the Spanish Ministryof Science and Innovation HuMoUR TIN2017-90086-R, and by the StateResearch Agency through the Mar ́ıa de Maeztu Seal of Excellence to IRI(MDM-2016-0656)., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

Automatic learning of cognitive exercises for socially assistive robotics

  • Suárez Hernández, Alejandro
  • Andriella, Antonio|||0000-0002-6641-6450
  • Taranovic, Aleksandar
  • Segovia Aguas, Javier
  • Torras, Carme|||0000-0002-2933-398X
  • Alenyà Ribas, Guillem|||0000-0002-6018-154X
© 2021 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting /republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works, In this paper, we present a learning approach to facilitate the teaching of new board exercises to assistive robotic systems. We formulate the problem as the learning of action models using Boolean predicates, disjunctive preconditions, and existential quantifiers from demonstrations of successful exercise executions. To be able to cope with exercises whose rules depend on a set of features that are initialized at the beginning of each play-out, we introduce the concept of dynamic context. Furthermore, we show how the learnt knowledge can be represented intuitively in a graphical interface that helps the caregiver understand what the system has learnt. As validation, we conducted a user study in which we evaluated whether and to which extent different types of feedback can affect the subjects’ performance while teaching three types of exercises: (1) sorting numbers; (2) arranging letters; and (3) reproducing shapes sequences in reversed order. The results suggest that textual and graphical feedback are beneficial., A. Andriella, C. Torras and A. Suarez-Hern ´ andez were partially funded ´ by the European Union´s Horizon 2020 under ERC Advanced Grant CLOTHILDE (no. 741930), G. Alenya by the EU H2020 research and ìnnovation programme IMAGINE (no. 731761) and J. Segovia-Aguas by the programme TAILOR (no. 952215). The work was partially supported by the Spanish State Research Agency through the María de Maeztu Seal of Excellence to IRI (MDM-2016-0656), Peer Reviewed, Award-winning
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

Contextual policy search for micro-data robot motion learning through covariate Gaussian process latent variable models

  • Delgado Guerrero, Juan Antonio
  • Colomé Figueras, Adrià|||0000-0001-9715-4062
  • Torras, Carme|||0000-0002-2933-398X
In the next few years, the amount and variety of context-aware robotic manipulator applications is expected to increase significantly, especially in household environments. In such spaces, thanks to programming by demonstration, non-expert people will be able to teach robots how to perform specific tasks, for which the adaptation to the environment is imperative, for the sake of effectiveness and users safety. These robot motion learning procedures allow the encoding of such tasks by means of parameterized trajectory generators, usually a Movement Primitive (MP) conditioned on contextual variables. However, naively sampled solutions from these MPs are generally suboptimal/inefficient, according to a given reward function. Hence, Policy Search (PS) algorithms leverage the information of the experienced rewards to improve the robot performance over executions, even for new context configurations. Given the complexity of the aforementioned tasks, PS methods face the challenge of exploring in high-dimensional parameter search spaces. In this work, a solution combining Bayesian Optimization, a data-efficient PS algorithm, with covariate Gaussian Process Latent Variable Models, a recent Dimensionality Reduction technique, is presented. It enables reducing dimensionality and exploiting prior demonstrations to converge in few iterations, while also being compliant with context requirements. Thus, contextual variables are considered in the latent search space, from which a surrogate model for the reward function is built. Then, samples are generated in a low-dimensional latent space, and mapped to a context-dependent trajectory. This allows us to drastically reduce the search space with the covariate GPLVM, e.g. from 105 to 2 parameters, plus a few contextual features. Experimentation in two different scenarios proves the data-efficiency and the power of dimensionality reduction of our approach., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

Mixtures of controlled Gaussian processes for dynamical modeling of deformable objects

  • Xu Zheng, Ce
  • Colomé Figueras, Adrià|||0000-0001-9715-4062
  • Sentis, Luis
  • Torras, Carme|||0000-0002-2933-398X
Control and manipulation of objects is a highly relevant topic in Robotics research. Although significant advances have been made over the manipulation of rigid bodies, the manipulation of non-rigid objects is still challenging and an open problem. Due to the uncertainty of the outcome when applying physical actions to non-rigid objects, using prior knowledge on objects’ dynamics can greatly improve the control performance. However, fitting such models is a challenging task for materials such as clothing, where the state is represented by points in a mesh, resulting in very large dimensionality that makes models difficult to learn, process and predict based on measured data. In this paper, we expand previous work on Controlled Gaussian Process Dynamical Models (CGPDM), a method that uses a non-linear projection of the state space onto a much smaller dimensional latent space, and learns the object dynamics in the latent space. We take advantage of the variability in training data by employing Mixture of Experts (MoE), and we devise theory and experimental validations that demonstrate significant improvements in training and prediction times, plus robustness and error stability when predicting deformable objects exposed to disparate movement ranges., This work was partially developed in the context of the project CLOTHILDE (”CLOTH manIpulation Learning from DEmonstrations”), which has received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (Advanced Grant agreement No 741930). We would like to thank the members of the HCRL Lab and the Department of Aerospace Engineering and Engineering Mechanics at The University of Texas at Austin for their feedback during the development of this work., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

Interaction identification through tactile sensing during cloth manipulation using a 3-axis touch sensor

  • Geer Couste, Idril-tadzio|||0000-0002-3517-1032
  • Maceira Duch, Marc
  • Borràs Sol, Júlia|||0000-0002-1662-2037
  • Torras, Carme|||0000-0002-2933-398X
  • Alenyà Ribas, Guillem|||0000-0002-6018-154X
Tactile feedback during cloth manipulation could be crucial in addressing the huge challenges involved in closing the loop during execution, complementing vision. However, up to our knowledge, tactile sensing has only been successfully used in cloth manipulation to classify type of fabrics, detect how many layers were grasped, and estimate the grasping force. In this work, we want to explore its potential to also provide information about whether the task is executed as expected. Two types of experiments are performed, in which a robot carries out 4 simple tasks, all involving a single finger manipulating a flat cloth on a table. Firstly, we analyze the sensor’s signals once the cloth manipulation has finished using Dynamic Time Warping (DTW) to see if they are informative enough to classify the tasks. Our results show that tactile feedback depends highly on the type of manipulated fabric. Secondly, we analyze the tactile feedback during the manipulation of the cloth using a recurrent neural network (RNN). For each sensor measurement, the RNN recognizes if the finger slides over the cloth, pulls it, flattens a fold in it, or if it’s about to lose contact, with 95.7% accuracy. These are promising results that show how tactile sensing has the potential of providing crucial information that would be very difficult to obtain with vision only., This work receives funding from the European Union Horizon 2020 Pro-gram under grant agreement no. 741930 (CLOTHILDE), by the HuMoURproject TIN2017-90086-R (AEI/FEDER, EU) and by AEI through the Mariade Maeztu Seal of Excellence to IRI (MDM-2016-0656)., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

Development and pilot evaluation of the ArmTracker: a wearable system to monitor arm kinematics during daily life

  • Carmona Ortiz, Víctor A.
  • Lobo Prat, Joan
  • van Ruysevelt, Jeremy
  • Torras, Carme|||0000-0002-2933-398X
  • Font Llagunes, Josep Maria|||0000-0002-7192-2980
© 2020 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works., Wearable sensing technology is proving useful for promoting health and fitness for the general public and athletes, yet few devices are tailored to people with movement impairments. For devices targeting home-based rehabilitation, it is crucial to have robust and non-obtrusive sensors capable of measuring activity for long periods of time outside of a laboratory environment. Studies focusing on continuous monitoring of arm activity during daily life over weeks or months only use IMU sensors or accelerometers at the wrist, and do not capture multi-segment kinematics. In this paper we present the development of the ArmTracker, a fully portable, non-obtrusive and wearable IMU-based motion capture system that can measure arm and torso kinematics for long-periods of time during daily life. We also present the results of a preliminary evaluation of the prototype, carried out with an unimpaired subject and a subject with Becker muscular dystrophy (BMD). Both subjects were asked to wear the ArmTracker device during daily life for 7 hours. We carried out an exploratory graphical analysis with the measured data using three types of movement quality metrics: 1) Range of Motion, 2) Functional Workspace Distribution and, 3) Accelerometry. Results provided an insightful view on the motor function capabilities and limitations of the two subjects. Arm activity of the subject with BMD showed low variability in terms of joint angles and hand positions over the workspace with a clear preference of using his hands in front and below shoulder height. Arm movements of the BMD subject were also slower compared with the unimpaired subject and with a slight preference for using the dominant arm. We plan to extend this study with a larger sample of subjects, and with measurements over several days and weeks to capture representative data on arm activity., This work has been partiallysupported by the Juan de la Cierva - Formaci ́on Postdoctoral Fellowshipof the Ministerio de Ciencia e Innovaci ́on (Ref: FJCI-2017-31754), and bythe European Research Council (ERC) under the Horizon 2020 AdvancedGrant agreement no. 741930 (project CLOTHILDE)., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

Sample-efficient robot motion learning using Gaussian process latent variable models

  • Delgado Guerrero, Juan Antonio
  • Colomé Figueras, Adrià|||0000-0001-9715-4062
  • Torras, Carme|||0000-0002-2933-398X
Robotic manipulators are reaching a state where we could see them in household environments in the following decade. Nevertheless, such robots need to be easy to instruct by lay people. This is why kinesthetic teaching has become very popular in recent years, in which the robot is taught a motion that is encoded as a parametric function - usually a Movement Primitive (MP)-. This approach produces trajectories that are usually suboptimal, and the robot needs to be able to improve them through trial-and-error. Such optimization is often done with Policy Search (PS) reinforcement learning, using a given reward function. PS algorithms can be classified as model-free, where neither the environment nor the reward function are modelled, or model-based, which can use a surrogate model of the reward function and/or a model for the dynamics of the task. However, MPs can become very high-dimensional in terms of parameters, which constitute the search space, so their optimization often requires too many samples. In this paper, we assume we have a robot motion task characterized with an MP of which we cannot model the dynamics. We build a surrogate model for the reward function, that maps an MP parameter latent space (obtained through a Mutual-information-weighted Gaussian Process Latent Variable Model) into a reward. While we do not model the task dynamics, using mutual information to shrink the task space makes it more consistent with the reward and so the policy improvement is faster in terms of sample efficiency., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

Probabilistic planning for robotics with ROSPlan

  • Canal Camprodon, Gerard|||0000-0002-6718-1198
  • Cashmmore, Michael
  • Krivic, Senka
  • Alenyà Ribas, Guillem|||0000-0002-6018-154X
  • Magazzeni, Daniele
  • Torras, Carme|||0000-0002-2933-398X
The final publication is available at link.springer.com, Probabilistic planning is very useful for handling uncertainty in planning tasks to be carried out by robots. ROSPlan is a framework for task planning in the Robot Operating System (ROS), but until now it has not been possible to use probabilistic planners within the framework. This systems paper presents a standardized integration of probabilistic planners into ROSPlan that allows for reasoning with non-deterministic effects and is agnostic to the probabilistic planner used. We instantiate the framework in a system for the case of a mobile robot performing tasks indoors, where probabilistic plans are generated and executed by the PROST planner. We evaluate the effectiveness of the proposed approach in a real-world robotic scenario., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

User interactions and negative examples to improve the learning of semantic rules in a cognitive exercise scenario

  • Suárez Hernández, Alejandro
  • Andriella, Antonio|||0000-0002-6641-6450
  • Torras, Carme|||0000-0002-2933-398X
  • Alenyà Ribas, Guillem|||0000-0002-6018-154X
© 2023 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works., Enabling a robot to perform new tasks is a complex endeavor, usually beyond the reach of non-technical users. For this reason, research efforts that aim at empowering end-users to teach robots new abilities using intuitive modes of interaction are valuable. In this article, we present INtuitive PROgramming 2 (INPRO2), a learning framework that allows inferring planning actions from demonstrations given by a human teacher. INPRO2 operates in an assistive scenario, in which the robot may learn from a healthcare professional (a therapist or caregiver) new cognitive exercises that can be later administered to patients with cognitive impairment. INPRO2 features significant improvements over previous work, namely: (1) exploitation of negative examples; (2) proactive interaction with the teacher to ask questions about the legality of certain movements; and (3) learning goals in addition to legal actions. Through simulations, we show the performance of different proactive strategies for gathering negative examples. Real-world experiments with human teachers and a TIAGo robot are also presented to qualitatively illustrate INPRO2., The research leading to these results receives funding from the EU H2020 Programme under grant agreement ERC–2016–ADG–741930 (CLOTHILDE); by MCIN/ AEI /10.13039/501100011033 under the project CHLOE-GRAPH (PID2020-119244GB-I00); by MCIN/ AEI /10.13039/501100011033 and by the ”European Union NextGenerationEU/PRTR under the project COHERENT (PCI2020-120718-2); AA acknowledges the European Union’s Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No 801342 (Tecniospring INDUSTRY); ASH acknowledges the financial support of the Apadrina la Ciencia association and of the Ford Spanish branch., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

Implications of robot backchannelling in cognitive therapy

  • Andriella, Antonio|||0000-0002-6641-6450
  • Torras, Carme|||0000-0002-2933-398X
  • Alenyà Ribas, Guillem|||0000-0002-6018-154X
The social ability of humans to provide active feedback during conversations is known as backchannelling. Recent work has recognised the importance of endowing robots with such social behaviour to make interactions more natural. Nonetheless, very little is known about how backchannelling should be designed in order to be detected and whether it can have an impact on users' behaviour and performance in cooperative tasks. In this article, we aim at evaluating the legibility of robot's backchannelling behaviour on Persons with Dementia (PwDs) and its effect on their performance when playing cognitive training exercises. Aiming to do so, a TIAGo robot was endowed with backchannelling behaviour generated by combining verbal and non-verbal cues. To evaluate our system, two user studies were carried out, in which the social signal was provided first by a human therapist and later on by a robot. Results indicate that patients were capable of identifying such kind of feedback. Nonetheless, our findings pointed out a significant difference in terms of performance between the two studies. They reveal how patients in the study with the robot overused the feedback to obtain the correct answer, putting in place a cheating mechanism that has led them to significantly worsen their performance. We conclude our work by discussing the implications of our findings when deploying robots in sensitive roles and possible solutions to address such unexpected behaviours., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

Anticipatory science fiction to foster ethical debates on AI and robotics

  • Torras, Carme|||0000-0002-2933-398X
The influence that hyperconnectivity and our increasing interaction with machines will have on the evolution of society and on people’s daily lives is difficult to predict. Thus, when trying to anticipate the potential benefits and risks of information technologies, not just lay people resort to science fiction (SF), but some scholars and even companies do so. Several universities, particularly in the US, include in their Computer Science and Engineering degrees a course on ethics and human values relative to technology. In this context of university education, my novel The Vestigial Heart (MIT Press, 2018) has been published together with online materials to teach a course on Ethics in Social Robotics and AI. Anticipatory literature has always taken science seriously and has tried to project its accomplishments into the future. It seems that science is also starting to take this literature seriously and find inspiration therein. This confluence could be extremely productive and is very good news, opening up interesting perspectives for the coming years., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

Garment manipulation dataset for robot learning by demonstration through a virtual reality framework

  • Boix Granell, Arnau
  • Foix Salmerón, Sergi|||0000-0001-9249-6696
  • Torras, Carme|||0000-0002-2933-398X
., Being able to teach complex capabilities, such as folding garments, to a bi-manual robot is a very challenging task, which is often tackled using learning from demonstration datasets. The few garment folding datasets available nowadays to the robotics research community are either gathered from human demonstrations or generated through simulation. The former have the huge problem of perceiving human action and transferring it to the dynamic control of the robot, while the latter requires coding human motion into the simulator in open loop, resulting in far-from-realistic movements. In this article, we present a reduced but very accurate dataset of human cloth folding demonstrations. The dataset is collected through a novel virtual reality (VR) framework we propose, based on Unity’s 3D platform and the use of a HTC Vive Pro system. The framework is capable of simulating very realistic garments while allowing users to interact with them, in real time, through handheld controllers. By doing so, and thanks to the immersive experience, our framework gets rid of the gap between the human and robot perception-action loop, while simplifying data capture and resulting in more realistic sample, This work was developed in the context of the project CLOTHILDE (”CLOTH manIpulation Learning from DEmonstrations”) which has received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation program (grant agreement No. 741930) and is also supported by the BURG project PCI2019-103447 funded by MCIN/ AEI /10.13039/501100011033 and by the ”European Union”., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

Dynamic cloth manipulation with deep reinforcement learning

  • Jangir, Rishabh
  • Alenyà Ribas, Guillem|||0000-0002-6018-154X
  • Torras, Carme|||0000-0002-2933-398X
© 2020 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works., In this paper we present a Deep Reinforcement Learning approach to solve dynamic cloth manipulation tasks. Differing from the case of rigid objects, we stress that the followed trajectory (including speed and acceleration) has a decisive in¿uence on the ¿nal state of cloth, which can greatly vary even if the positions reached by the grasped points are the same. We explore how goal positions for non-grasped points can be attained through learning adequate trajectories for the grasped points. Our approach uses few demonstrations to improve control policy learning, and a sparse reward approach to avoid engineering complex reward functions. Since perception of textiles is challenging, we also study different state representations to assess the minimum observation space required for learning to succeed. Finally, we compare different combinations of control policy encodings, demonstrations, and sparse reward learning techniques, and show that our proposed approach can learn dynamic cloth manipulation in an ef¿cient way, i.e., using a reduced observation space, a few demonstrations, and a sparse reward., his work has been supported by the ERC project Clothilde (ERC-2016-ADG-741930), the HuMoUR project TIN2017-90086-R (AEI/FEDER, UE)and by the AEI through the María de Maeztu Seal of Excellence to IRI(MDM-2016-0656), Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

Heteroscedastic Gaussian processes and random features: scalable motion primitives with guarantees

  • Caldarelli, Edoardo
  • Chatalic, Antoine
  • Colomé Figueras, Adrià|||0000-0001-9715-4062
  • Rosasco, Lorenzo
  • Torras, Carme|||0000-0002-2933-398X
Heteroscedastic Gaussian processes (HGPs) are kernel-based, non-parametric models that can be used to infer nonlinear functions with time-varying noise. In robotics, they can be employed for learning from demonstration as motion primitives, i.e. as a model of the trajectories to be executed by the robot. HGPs provide variance estimates around the reference signal modeling the trajectory, capturing both the predictive uncertainty and the motion variability. However, similarly to standard Gaussian processes they suffer from a cubic complexity in the number of training points, due to the inversion of the kernel matrix. The uncertainty can be leveraged for more complex learning tasks, such as inferring the variable impedance profile required from a robotic manipulator. However, suitable approximations are needed to make HGPs scalable, at the price of potentially worsening the posterior mean and variance profiles. Motivated by these observations, we study the combination of HGPs and random features, which are a popular, data-independent approximation strategy of kernel functions. In a theoretical analysis, we provide novel guarantees on the approximation error of the HGP posterior due to random features. Moreover, we validate this scalable motion primitive on real robot data, related to the problem of variable impedance learning. In this way, we show that random features offer a viable and theoretically sound alternative for speeding up the trajectory processing, without sacrificing accuracy., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

The KIT swiss knife gripper for disassembly tasks: a multi-functional gripper for bimanual manipulation with a single arm

  • Borràs Sol, Júlia|||0000-0002-1662-2037
  • Heudorfer, Raphael
  • Rader, Samuel
  • Asfour, Tamim
© 20xx IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works., This work presents the concept of a robotic gripper designed for the disassembly of electromechanical devices that comprises several innovative ideas. Novel concepts include the ability to interchange built-in tools without the need to grasp them, the ability to reposition grasped objects in-hand, the capability of performing classic dual arm manipulation within the gripper and the utilization of classic industrial robotic arms kinematics within a robotic gripper. We analyze state of the art grippers and robotic hands designed for dexterous in-hand manipulation and extract common characteristics and weak points. The presented concept is obtained from the task requirements for disassembly of electromechanical devices and it is then evaluated for general purpose grasping, in-hand manipulation and operations with tools. We further present the CAD design for a first prototype., Peer Reviewed
UPCommons. Portal del coneixement obert de la UPC
Publication Contribución de congreso (ConferenceOutput)

Variable impedance control in cartesian latent space while avoiding obstacles in null space

  • Parent Alonso, David
  • Colomé Figueras, Adrià|||0000-0001-9715-4062
  • Torras, Carme|||0000-0002-2933-398X
© 2020 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works., Human-robot interaction is one of the keys of assistive robots. Robots are expected to be compliant with people but at the same time correctly perform the tasks. In such applications, Cartesian impedance control is preferred over joint control, as the desired interaction and environmental feedback can be described more naturally, and the force to be exerted by the robot can be readily adjusted. This paper addresses the problem of controlling a robot arm in the operational space with variable stiffness so as to continuously adapt the force exerted in each phase of motion according to the precision requirements. Moreover, performing dimensionality reduction we can separate the degrees of freedom (DoF) relevant for the task from the redundant ones. The stiffness of the former can be adjusted constantly to achieve the required accuracy, while task-redundant DoF can be used to achieve other goals such as avoiding obstacles by moving in the directions where accuracy is not critical. The designed method is tested teaching the robot to give water to drink to a model of human head. Our empirical results demonstrate that the robot can learn precision requirements from demonstration. Furthermore, dimensionality reduction is proved to be useful to avoid obstacles., This work was partially developed in the context of the project CLOTHILDE ("CLOTH manIpulation Learning from DEmonstrations"),which has received funding from the European Research Council (ERC) underthe European Union’s Horizon 2020 research and innovation programme(Advanced Grant agreement No 741930). This work is also supported bythe Spanish State Research Agency through the María de Maeztu Seal ofExcellence to IRI MdM-IP-2018-05., Peer Reviewed
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

Evaluating the effect of theory of mind on people’s trust in a faulty robot

  • Rossi, Alessandra
  • Andriella, Antonio
  • Rossi, Silvia
  • Torras, Carme
  • Alenyà Ribas, Guillem
© 2022 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes,creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works., The success of human-robot interaction is strongly affected by the people's ability to infer others' intentions and behaviours, and the level of people's trust that others will abide by their same principles and social conventions to achieve a common goal. The ability of understanding and reasoning about other agents' mental states is known as Theory of Mind (ToM). ToM and trust, therefore, are key factors in the positive outcome of human-robot interaction. We believe that a robot endowed with a ToM is able to gain people's trust, even when this may occasionally make errors. In this work, we present a user study in the field in which participants (N=123) interacted with a robot that may or may not have a ToM, and may or may not exhibit erroneous behaviour. Our findings indicate that a robot with ToM is perceived as more reliable, and they trusted it more than a robot without a ToM even when the robot made errors. Finally, ToM results to be a key driver for tuning people's trust in the robot even when the initial condition of the interaction changed (i.e., loss and regain of trust in a longer relationship)., This work has been supported by Italian PON R&I 2014-2020 - REACT-EU (CUP E65F21002920003); by the European Union’s Horizon 2020 under ERC Advanced Grant CLOTHILDE (no. 741930); by MCIN/ AEI /10.13039/501100011033 and by the ”European Union NextGenerationEU/PRTR” under the project ROBIN (PLEC2021-007859) and the project COHERENT (PCI2020-120718-2); by the ”European Union NextGenerationEU/PRTR” through CSIC’s Thematic Platforms (PTI+ Neuro-Aging) and by the European Union’s Horizon 2020 research and innovation programme PRO-CARED (no.801342) under Marie Skłodowska-Curie grant agreement (Tecniospring INDUSTRY) and the Government of Catalonia’s Agency for Business Competitiveness (ACCIO)., Peer Reviewed, Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

Garment manipulation dataset for robot learning by demonstration through a virtual reality framework

  • Boix Granell, Arnau
  • Foix Salmerón, Sergi
  • Torras, Carme
., Being able to teach complex capabilities, such as folding garments, to a bi-manual robot is a very challenging task, which is often tackled using learning from demonstration datasets. The few garment folding datasets available nowadays to the robotics research community are either gathered from human demonstrations or generated through simulation. The former have the huge problem of perceiving human action and transferring it to the dynamic control of the robot, while the latter requires coding human motion into the simulator in open loop, resulting in far-from-realistic movements. In this article, we present a reduced but very accurate dataset of human cloth folding demonstrations. The dataset is collected through a novel virtual reality (VR) framework we propose, based on Unity’s 3D platform and the use of a HTC Vive Pro system. The framework is capable of simulating very realistic garments while allowing users to interact with them, in real time, through handheld controllers. By doing so, and thanks to the immersive experience, our framework gets rid of the gap between the human and robot perception-action loop, while simplifying data capture and resulting in more realistic sample, This work was developed in the context of the project CLOTHILDE (”CLOTH manIpulation Learning from DEmonstrations”) which has received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation program (grant agreement No. 741930) and is also supported by the BURG project PCI2019-103447 funded by MCIN/ AEI /10.13039/501100011033 and by the ”European Union”., Peer Reviewed, Postprint (published version)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

Dynamic cloth manipulation with deep reinforcement learning

  • Jangir, Rishabh
  • Alenyà Ribas, Guillem
  • Torras, Carme
© 2020 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works., In this paper we present a Deep Reinforcement Learning approach to solve dynamic cloth manipulation tasks. Differing from the case of rigid objects, we stress that the followed trajectory (including speed and acceleration) has a decisive in¿uence on the ¿nal state of cloth, which can greatly vary even if the positions reached by the grasped points are the same. We explore how goal positions for non-grasped points can be attained through learning adequate trajectories for the grasped points. Our approach uses few demonstrations to improve control policy learning, and a sparse reward approach to avoid engineering complex reward functions. Since perception of textiles is challenging, we also study different state representations to assess the minimum observation space required for learning to succeed. Finally, we compare different combinations of control policy encodings, demonstrations, and sparse reward learning techniques, and show that our proposed approach can learn dynamic cloth manipulation in an ef¿cient way, i.e., using a reduced observation space, a few demonstrations, and a sparse reward., his work has been supported by the ERC project Clothilde (ERC-2016-ADG-741930), the HuMoUR project TIN2017-90086-R (AEI/FEDER, UE)and by the AEI through the María de Maeztu Seal of Excellence to IRI(MDM-2016-0656), Peer Reviewed, Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

Contextual policy search for micro-data robot motion learning through covariate Gaussian process latent variable models

  • Delgado Guerrero, Juan Antonio
  • Colomé Figueras, Adrià
  • Torras, Carme
In the next few years, the amount and variety of context-aware robotic manipulator applications is expected to increase significantly, especially in household environments. In such spaces, thanks to programming by demonstration, non-expert people will be able to teach robots how to perform specific tasks, for which the adaptation to the environment is imperative, for the sake of effectiveness and users safety. These robot motion learning procedures allow the encoding of such tasks by means of parameterized trajectory generators, usually a Movement Primitive (MP) conditioned on contextual variables. However, naively sampled solutions from these MPs are generally suboptimal/inefficient, according to a given reward function. Hence, Policy Search (PS) algorithms leverage the information of the experienced rewards to improve the robot performance over executions, even for new context configurations. Given the complexity of the aforementioned tasks, PS methods face the challenge of exploring in high-dimensional parameter search spaces. In this work, a solution combining Bayesian Optimization, a data-efficient PS algorithm, with covariate Gaussian Process Latent Variable Models, a recent Dimensionality Reduction technique, is presented. It enables reducing dimensionality and exploiting prior demonstrations to converge in few iterations, while also being compliant with context requirements. Thus, contextual variables are considered in the latent search space, from which a surrogate model for the reward function is built. Then, samples are generated in a low-dimensional latent space, and mapped to a context-dependent trajectory. This allows us to drastically reduce the search space with the covariate GPLVM, e.g. from 105 to 2 parameters, plus a few contextual features. Experimentation in two different scenarios proves the data-efficiency and the power of dimensionality reduction of our approach., Peer Reviewed, Postprint (published version)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

A virtual reality framework for fast dataset creation applied to cloth manipulation with automatic semantic labelling

  • Borràs Sol, Júlia
  • Boix Granell, Arnau
  • Foix Salmerón, Sergi
  • Torras, Carme
© 2023 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works., Teaching complex manipulation skills, such as folding garments, to a bi-manual robot is a very challenging task, which is often tackled through learning from demonstration. The few datasets of garment-folding demonstrations available nowadays to the robotics research community have been either gathered from human demonstrations or generated through simulation. The former have the great difficulty of perceiving both cloth state and human action as well as transferring them to the dynamic control of the robot, while the latter require coding human motion into the simulator in open loop, i.e., without incorporating the visual feedback naturally used by people, resulting in far-from-realistic movements. In this article, we present an accurate dataset of human cloth folding demonstrations. The dataset is collected through our novel virtual reality (VR) framework, based on Unity’s 3D platform and the use of an HTC Vive Pro system. The framework is capable of simulating realistic garments while allowing users to interact with them in real time through handheld controllers. By doing so, and thanks to the immersive experience, our framework permits exploiting human visual feedback in the demonstrations while at the same time getting rid of the difficulties of capturing the state of cloth, thus simplifying data acquisition and resulting in more realistic demonstrations. We create and make public a dataset of cloth manipulation sequences, whose cloth states are semantically labeled in an automatic way by using a novel low-dimensional cloth representation that yields a very good separation between different cloth configurations., The research leading to these results receives funding from the European Research Council (ERC) from the European Union Horizon 2020 Programme under grant agreement no. 741930 (CLOTHILDE: CLOTH manIpulation Learning from DEmonstrations) and project SoftEnable (HORIZONCL4-2021-DIGITAL-EMERGING-01-101070600). Authors also received funding from project CHLOE-GRAPH (PID2020-118649RB-I00) funded by MCIN/ AEI /10.13039/501100011033 and COHERENT (PCI2020-120718-2) funded by MCIN/ AEI /10.13039/501100011033 and cofunded by the ”European Union NextGenerationEU/PRTR”., Peer Reviewed, Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

Fitting parameters of linear dynamical systems to regularize forcing terms in dynamical movement primitives

  • Stulp, Freek
  • Colomé Figueras, Adrià
  • Torras, Carme
© 2024 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works., Due to their flexibility and ease of use, Dynamical Movement Primitives (DMPs) are widely used in robotics applications and research. DMPs combine linear dynamical systems to achieve robustness to perturbations and adaptation to moving targets with non-linear function approximators to fit a wide range of demonstrated trajectories. We propose a novel DMP formulation with a generalized logistic function as a delayed goal system. This formulation inherently has low initial jerk, and generates the bell-shaped velocity profiles that are typical of human movement. As the novel formulation is more expressive, it is able to fit a wide range of human demonstrations well, also without a non-linear forcing term. We exploit this increased expressiveness by automating the fitting of the dynamical system parameters through optimization. Our experimental evaluation demonstrates that this optimization regularizes the forcing term, and improves the interpolation accuracy of parametric DMPs., This work was partially funded by the European ERC/EIC/Horizon projects with grant numbers 741930 (CLOTHILDE), 951992 (VeriDream), and 101070596 (euROBIN), as well the CSIC project 202350E080 (ClothIRI), and the MCIN/AEI project PID2020-118649RBI00 (CHLOE-GRAPH)., Peer Reviewed, Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

Anticipatory science fiction to foster ethical debates on AI and robotics

  • Torras, Carme
The influence that hyperconnectivity and our increasing interaction with machines will have on the evolution of society and on people’s daily lives is difficult to predict. Thus, when trying to anticipate the potential benefits and risks of information technologies, not just lay people resort to science fiction (SF), but some scholars and even companies do so. Several universities, particularly in the US, include in their Computer Science and Engineering degrees a course on ethics and human values relative to technology. In this context of university education, my novel The Vestigial Heart (MIT Press, 2018) has been published together with online materials to teach a course on Ethics in Social Robotics and AI. Anticipatory literature has always taken science seriously and has tried to project its accomplishments into the future. It seems that science is also starting to take this literature seriously and find inspiration therein. This confluence could be extremely productive and is very good news, opening up interesting perspectives for the coming years., Peer Reviewed, Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

Discovering SOCIABLE: using a conceptual model to evaluate the legibility and effectiveness of backchannel cues in an entertainment scenario

  • Andriella, Antonio
  • Huertas García, Ruben
  • Forgas Coll, Santiago
  • Torras, Carme
  • Alenyà Ribas, Guillem
© 2020 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works., Robots are expected to become part of everyday life. However, while there have been important breakthroughs during the recent decades in terms of technological advances, the ability of robots to interact with humans intuitively and effectively is still an open challenge. In this paper, we aim to evaluate how humans interpret and leverage backchannel cues exhibited by a robot which interacts with them in an entertainment context. To do so, a conceptual model was designed to investigate the legibility and the effectiveness of a designed social cue, called SOCial ImmediAcy BackchanneL cuE (SOCIABLE), on participant’s performance. In addition, user’s attitude and cognitive capability were integrated into the model as an estimator of participants’ motivation and ability to process the cue. In working toward such a goal, we conducted a two-day long user study (N=114) at an international event with untrained participants who were not aware of the social cue the robot was able to provide. The results showed that participants were able to perceive the social signal generated from SOCIABLE and thus, they benefited from it. Our findings provide some important insights for the design of effective and instantaneous backchannel cues and the methodology for evaluating them in social robots., This work has been partially funded by the EU project SOCRATESH2020-MSCA-ITN-721619, by the European Union ́s Horizon 2020 underERC Advanced Grant CLOTHILDE (no. 741930), by the Spanish Ministryof Science and Innovation HuMoUR TIN2017-90086-R, and by the StateResearch Agency through the Mar ́ıa de Maeztu Seal of Excellence to IRI(MDM-2016-0656)., Peer Reviewed, Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

Ordinal inverse reinforcement learning applied to robot learning with small data

  • Colomé Figueras, Adrià
  • Torras, Carme
© 2022 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works., Over the last decade, the ability to teach actions to robots in a user-friendly way has gained relevance, and a practical way of teaching robots a new task is to use Inverse Reinforcement Learning (IRL). In IRL, an expert teacher shows the robot a desired behaviour and an agent builds a model of the reward. The agent can also infer a policy that performs in an optimal way within the limitations of the knowledge provided to it. However, most IRL approaches assume an (almost) optimal performance of the teaching agent, which might become unpractical if the teacher is not actually an expert. In addition, most IRL focus on discrete state-action spaces that limit their applicability to certain real-world problems such as within the context of direct Policy Search (PS) reinforcement learning. Therefore, in this paper we introduce Ordinal Inverse Reinforcement Learning (OrdIRL) for continuous state variables, in which the teacher can qualitatively evaluate robot performance by selecting one among the predefined performance levels (e.g. {bad, medium, good} for three tiers of performance). Once the OrdIRL has fit an ordinal distribution to the data, we propose to use Bayesian Optimization (BO) to either gain knowledge on the inferred model (exploration) or find a policy or action that maximizes the expected reward given the prior knowledge on the reward (exploitation). In the case of large-dimensional state-action spaces, we use Dimensionality Reduction (DR) techniques and perform the BO in the latent space. Experimental results on simulation and with a robot arm show how this approach allows for learning the reward function with small data., This work was developed in the context of the project CLOTHILDE ("CLOTH manIpulation Learning from DEmonstrations"), which has received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (Advanced Grant agreement No 741930)., Peer Reviewed, Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

Variable impedance control in cartesian latent space while avoiding obstacles in null space

  • Parent Alonso, David
  • Colomé Figueras, Adrià
  • Torras, Carme
© 2020 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works., Human-robot interaction is one of the keys of assistive robots. Robots are expected to be compliant with people but at the same time correctly perform the tasks. In such applications, Cartesian impedance control is preferred over joint control, as the desired interaction and environmental feedback can be described more naturally, and the force to be exerted by the robot can be readily adjusted. This paper addresses the problem of controlling a robot arm in the operational space with variable stiffness so as to continuously adapt the force exerted in each phase of motion according to the precision requirements. Moreover, performing dimensionality reduction we can separate the degrees of freedom (DoF) relevant for the task from the redundant ones. The stiffness of the former can be adjusted constantly to achieve the required accuracy, while task-redundant DoF can be used to achieve other goals such as avoiding obstacles by moving in the directions where accuracy is not critical. The designed method is tested teaching the robot to give water to drink to a model of human head. Our empirical results demonstrate that the robot can learn precision requirements from demonstration. Furthermore, dimensionality reduction is proved to be useful to avoid obstacles., This work was partially developed in the context of the project CLOTHILDE ("CLOTH manIpulation Learning from DEmonstrations"),which has received funding from the European Research Council (ERC) underthe European Union’s Horizon 2020 research and innovation programme(Advanced Grant agreement No 741930). This work is also supported bythe Spanish State Research Agency through the María de Maeztu Seal ofExcellence to IRI MdM-IP-2018-05., Peer Reviewed, Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

Heteroscedastic Gaussian processes and random features: scalable motion primitives with guarantees

  • Caldarelli, Edoardo
  • Chatalic, Antoine
  • Colomé Figueras, Adrià
  • Rosasco, Lorenzo
  • Torras, Carme
Heteroscedastic Gaussian processes (HGPs) are kernel-based, non-parametric models that can be used to infer nonlinear functions with time-varying noise. In robotics, they can be employed for learning from demonstration as motion primitives, i.e. as a model of the trajectories to be executed by the robot. HGPs provide variance estimates around the reference signal modeling the trajectory, capturing both the predictive uncertainty and the motion variability. However, similarly to standard Gaussian processes they suffer from a cubic complexity in the number of training points, due to the inversion of the kernel matrix. The uncertainty can be leveraged for more complex learning tasks, such as inferring the variable impedance profile required from a robotic manipulator. However, suitable approximations are needed to make HGPs scalable, at the price of potentially worsening the posterior mean and variance profiles. Motivated by these observations, we study the combination of HGPs and random features, which are a popular, data-independent approximation strategy of kernel functions. In a theoretical analysis, we provide novel guarantees on the approximation error of the HGP posterior due to random features. Moreover, we validate this scalable motion primitive on real robot data, related to the problem of variable impedance learning. In this way, we show that random features offer a viable and theoretically sound alternative for speeding up the trajectory processing, without sacrificing accuracy., Peer Reviewed, Postprint (published version)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

Knowledge representation to enable high-level planning in cloth manipulation tasks

  • Garcia Camacho, Irene
  • Borràs Sol, Júlia
  • Alenyà Ribas, Guillem
Cloth manipulation is very relevant for domestic robotic tasks, but it presents many challenges due to the complexity of representing, recognizing and predicting the behaviour of cloth under manipulation. In this work, we propose a generic, compact and simplified representation of the states of cloth manipulation that allows for representing tasks as sequences of states and transitions semantically. We also define a Cloth Manipulation Graph that encodes all the strategies to accomplish a task. Our novel representation is used to encode two different cloth manipulation tasks, learned from an experiment with human subjects manipulating clothes with video data. We show how our simplified representation allows to obtain a map of meaningful steps that can serve to describe cloth manipulation tasks as domain models in PDDL, enabling high-level planning. Finally, we discuss on the existing skills that could enable the sensory motor grounding and the low-level execution of the plan., Peer Reviewed, Postprint (published version)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

Automatic learning of cognitive exercises for socially assistive robotics

  • Suárez Hernández, Alejandro
  • Andriella, Antonio
  • Taranovic, Aleksandar
  • Segovia Aguas, Javier
  • Torras, Carme
  • Alenyà Ribas, Guillem
© 2021 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting /republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works, In this paper, we present a learning approach to facilitate the teaching of new board exercises to assistive robotic systems. We formulate the problem as the learning of action models using Boolean predicates, disjunctive preconditions, and existential quantifiers from demonstrations of successful exercise executions. To be able to cope with exercises whose rules depend on a set of features that are initialized at the beginning of each play-out, we introduce the concept of dynamic context. Furthermore, we show how the learnt knowledge can be represented intuitively in a graphical interface that helps the caregiver understand what the system has learnt. As validation, we conducted a user study in which we evaluated whether and to which extent different types of feedback can affect the subjects’ performance while teaching three types of exercises: (1) sorting numbers; (2) arranging letters; and (3) reproducing shapes sequences in reversed order. The results suggest that textual and graphical feedback are beneficial., A. Andriella, C. Torras and A. Suarez-Hern ´ andez were partially funded ´ by the European Union´s Horizon 2020 under ERC Advanced Grant CLOTHILDE (no. 741930), G. Alenya by the EU H2020 research and ìnnovation programme IMAGINE (no. 731761) and J. Segovia-Aguas by the programme TAILOR (no. 952215). The work was partially supported by the Spanish State Research Agency through the María de Maeztu Seal of Excellence to IRI (MDM-2016-0656), Peer Reviewed, Award-winning, Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

Development and pilot evaluation of the ArmTracker: a wearable system to monitor arm kinematics during daily life

  • Carmona Ortiz, Víctor A.
  • Lobo Prat, Joan
  • van Ruysevelt, Jeremy
  • Torras, Carme
  • Font Llagunes, Josep Maria
© 2020 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works., Wearable sensing technology is proving useful for promoting health and fitness for the general public and athletes, yet few devices are tailored to people with movement impairments. For devices targeting home-based rehabilitation, it is crucial to have robust and non-obtrusive sensors capable of measuring activity for long periods of time outside of a laboratory environment. Studies focusing on continuous monitoring of arm activity during daily life over weeks or months only use IMU sensors or accelerometers at the wrist, and do not capture multi-segment kinematics. In this paper we present the development of the ArmTracker, a fully portable, non-obtrusive and wearable IMU-based motion capture system that can measure arm and torso kinematics for long-periods of time during daily life. We also present the results of a preliminary evaluation of the prototype, carried out with an unimpaired subject and a subject with Becker muscular dystrophy (BMD). Both subjects were asked to wear the ArmTracker device during daily life for 7 hours. We carried out an exploratory graphical analysis with the measured data using three types of movement quality metrics: 1) Range of Motion, 2) Functional Workspace Distribution and, 3) Accelerometry. Results provided an insightful view on the motor function capabilities and limitations of the two subjects. Arm activity of the subject with BMD showed low variability in terms of joint angles and hand positions over the workspace with a clear preference of using his hands in front and below shoulder height. Arm movements of the BMD subject were also slower compared with the unimpaired subject and with a slight preference for using the dominant arm. We plan to extend this study with a larger sample of subjects, and with measurements over several days and weeks to capture representative data on arm activity., This work has been partiallysupported by the Juan de la Cierva - Formaci ́on Postdoctoral Fellowshipof the Ministerio de Ciencia e Innovaci ́on (Ref: FJCI-2017-31754), and bythe European Research Council (ERC) under the Horizon 2020 AdvancedGrant agreement no. 741930 (project CLOTHILDE)., Peer Reviewed, Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

Interaction identification through tactile sensing during cloth manipulation using a 3-axis touch sensor

  • Geer Couste, Idril-tadzio
  • Maceira Duch, Marc
  • Borràs Sol, Júlia
  • Torras, Carme
  • Alenyà Ribas, Guillem
Tactile feedback during cloth manipulation could be crucial in addressing the huge challenges involved in closing the loop during execution, complementing vision. However, up to our knowledge, tactile sensing has only been successfully used in cloth manipulation to classify type of fabrics, detect how many layers were grasped, and estimate the grasping force. In this work, we want to explore its potential to also provide information about whether the task is executed as expected. Two types of experiments are performed, in which a robot carries out 4 simple tasks, all involving a single finger manipulating a flat cloth on a table. Firstly, we analyze the sensor’s signals once the cloth manipulation has finished using Dynamic Time Warping (DTW) to see if they are informative enough to classify the tasks. Our results show that tactile feedback depends highly on the type of manipulated fabric. Secondly, we analyze the tactile feedback during the manipulation of the cloth using a recurrent neural network (RNN). For each sensor measurement, the RNN recognizes if the finger slides over the cloth, pulls it, flattens a fold in it, or if it’s about to lose contact, with 95.7% accuracy. These are promising results that show how tactile sensing has the potential of providing crucial information that would be very difficult to obtain with vision only., This work receives funding from the European Union Horizon 2020 Pro-gram under grant agreement no. 741930 (CLOTHILDE), by the HuMoURproject TIN2017-90086-R (AEI/FEDER, EU) and by AEI through the Mariade Maeztu Seal of Excellence to IRI (MDM-2016-0656)., Peer Reviewed, Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

The KIT swiss knife gripper for disassembly tasks: a multi-functional gripper for bimanual manipulation with a single arm

  • Borràs Sol, Júlia
  • Heudorfer, Raphael
  • Rader, Samuel
  • Asfour, Tamim
© 20xx IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works., This work presents the concept of a robotic gripper designed for the disassembly of electromechanical devices that comprises several innovative ideas. Novel concepts include the ability to interchange built-in tools without the need to grasp them, the ability to reposition grasped objects in-hand, the capability of performing classic dual arm manipulation within the gripper and the utilization of classic industrial robotic arms kinematics within a robotic gripper. We analyze state of the art grippers and robotic hands designed for dexterous in-hand manipulation and extract common characteristics and weak points. The presented concept is obtained from the task requirements for disassembly of electromechanical devices and it is then evaluated for general purpose grasping, in-hand manipulation and operations with tools. We further present the CAD design for a first prototype., Peer Reviewed, Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

Quadratic dynamic matrix control for fast cloth manipulation

  • Caldarelli, Edoardo
  • Colomé Figueras, Adrià
  • Ocampo-Martínez, Carlos
  • Torras, Carme
© 2023 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting /republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works, Robotic cloth manipulation is an increasingly relevant area of research, challenging classic control algorithms due to the deformable nature of cloth. While it is possible to apply linear model predictive control to make the robot move the cloth according to a given reference, this approach suffers from a large dimensionality of the state-space representation of the cloth models. To address this issue, in this work we study the application of an input-output model predictive control strategy, based on quadratic dynamic matrix control, to robotic cloth manipulation. To account for uncertain disturbances on the cloth's motion, we further extend the algorithm with suitable chance constraints. In extensive simulated experiments, involving disturbances and obstacle avoidance, we show that quadratic dynamic matrix control can be successfully applied in different cloth manipulation scenarios, with significant gains in optimization speed compared to standard model predictive control strategies. The experiments further demonstrate that the closed-loop model used by quadratic dynamic matrix control can be beneficial to the tracking accuracy, leading to improvements over the standard predictive control strategy. Moreover, a preliminary experiment on a real robot shows that quadratic dynamic matrix control can indeed be employed in real settings., This work was supported by the project CLOTHILDE (“CLOTH manIpulation Learning from DEmonstrations”), funded by the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (Advanced Grant agreement No 741930)., Peer Reviewed, Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

Probabilistic planning for robotics with ROSPlan

  • Canal Camprodon, Gerard
  • Cashmmore, Michael
  • Krivic, Senka
  • Alenyà Ribas, Guillem
  • Magazzeni, Daniele
  • Torras, Carme
The final publication is available at link.springer.com, Probabilistic planning is very useful for handling uncertainty in planning tasks to be carried out by robots. ROSPlan is a framework for task planning in the Robot Operating System (ROS), but until now it has not been possible to use probabilistic planners within the framework. This systems paper presents a standardized integration of probabilistic planners into ROSPlan that allows for reasoning with non-deterministic effects and is agnostic to the probabilistic planner used. We instantiate the framework in a system for the case of a mobile robot performing tasks indoors, where probabilistic plans are generated and executed by the PROST planner. We evaluate the effectiveness of the proposed approach in a real-world robotic scenario., Peer Reviewed, Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

Mixtures of controlled Gaussian processes for dynamical modeling of deformable objects

  • Xu Zheng, Ce
  • Colomé Figueras, Adrià
  • Sentis, Luis
  • Torras, Carme
Control and manipulation of objects is a highly relevant topic in Robotics research. Although significant advances have been made over the manipulation of rigid bodies, the manipulation of non-rigid objects is still challenging and an open problem. Due to the uncertainty of the outcome when applying physical actions to non-rigid objects, using prior knowledge on objects’ dynamics can greatly improve the control performance. However, fitting such models is a challenging task for materials such as clothing, where the state is represented by points in a mesh, resulting in very large dimensionality that makes models difficult to learn, process and predict based on measured data. In this paper, we expand previous work on Controlled Gaussian Process Dynamical Models (CGPDM), a method that uses a non-linear projection of the state space onto a much smaller dimensional latent space, and learns the object dynamics in the latent space. We take advantage of the variability in training data by employing Mixture of Experts (MoE), and we devise theory and experimental validations that demonstrate significant improvements in training and prediction times, plus robustness and error stability when predicting deformable objects exposed to disparate movement ranges., This work was partially developed in the context of the project CLOTHILDE (”CLOTH manIpulation Learning from DEmonstrations”), which has received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (Advanced Grant agreement No 741930). We would like to thank the members of the HCRL Lab and the Department of Aerospace Engineering and Engineering Mechanics at The University of Texas at Austin for their feedback during the development of this work., Peer Reviewed, Postprint (published version)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

Reconstruction of sampled surfaces with boundary via Morse theory

  • Coltraro, Franco
  • Amorós Torrent, Jaume
  • Alberich Carramiñana, Maria
  • Torras, Carme
We study the perception problem for garments (e.g. a pair of pants) using tools from computational topology: the identification of their geometry and position from point-cloud samples, as obtained e.g. with 3D scanners. We present a reconstruction algorithm based on Morse theory that proceeds directly from the point-cloud to obtain a cellular decomposition of the surface derived via a Morse function. No intermediate triangulation or local implicit equations are used, avoiding reconstruction-induced artifices. The results are a piecewise parametrization of the surface as a union of Morse cells, suitable for tasks such as noise-filtering or mesh-independent reparametrization, and a cell complex of small rank determining the surface topology. This algorithm can be applied to smooth surfaces with or without boundary, embedded in an ambient space of any dimension., Peer Reviewed, Postprint (published version)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

Sample-efficient robot motion learning using Gaussian process latent variable models

  • Delgado Guerrero, Juan Antonio
  • Colomé Figueras, Adrià
  • Torras, Carme
Robotic manipulators are reaching a state where we could see them in household environments in the following decade. Nevertheless, such robots need to be easy to instruct by lay people. This is why kinesthetic teaching has become very popular in recent years, in which the robot is taught a motion that is encoded as a parametric function - usually a Movement Primitive (MP)-. This approach produces trajectories that are usually suboptimal, and the robot needs to be able to improve them through trial-and-error. Such optimization is often done with Policy Search (PS) reinforcement learning, using a given reward function. PS algorithms can be classified as model-free, where neither the environment nor the reward function are modelled, or model-based, which can use a surrogate model of the reward function and/or a model for the dynamics of the task. However, MPs can become very high-dimensional in terms of parameters, which constitute the search space, so their optimization often requires too many samples. In this paper, we assume we have a robot motion task characterized with an MP of which we cannot model the dynamics. We build a surrogate model for the reward function, that maps an MP parameter latent space (obtained through a Mutual-information-weighted Gaussian Process Latent Variable Model) into a reward. While we do not model the task dynamics, using mutual information to shrink the task space makes it more consistent with the reward and so the policy improvement is faster in terms of sample efficiency., Peer Reviewed, Postprint (published version)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

Implications of robot backchannelling in cognitive therapy

  • Andriella, Antonio
  • Torras, Carme
  • Alenyà Ribas, Guillem
The social ability of humans to provide active feedback during conversations is known as backchannelling. Recent work has recognised the importance of endowing robots with such social behaviour to make interactions more natural. Nonetheless, very little is known about how backchannelling should be designed in order to be detected and whether it can have an impact on users' behaviour and performance in cooperative tasks. In this article, we aim at evaluating the legibility of robot's backchannelling behaviour on Persons with Dementia (PwDs) and its effect on their performance when playing cognitive training exercises. Aiming to do so, a TIAGo robot was endowed with backchannelling behaviour generated by combining verbal and non-verbal cues. To evaluate our system, two user studies were carried out, in which the social signal was provided first by a human therapist and later on by a robot. Results indicate that patients were capable of identifying such kind of feedback. Nonetheless, our findings pointed out a significant difference in terms of performance between the two studies. They reveal how patients in the study with the robot overused the feedback to obtain the correct answer, putting in place a cheating mechanism that has led them to significantly worsen their performance. We conclude our work by discussing the implications of our findings when deploying robots in sensitive roles and possible solutions to address such unexpected behaviours., Peer Reviewed, Postprint (author's final draft)
Recercat. Dipósit de la Recerca de Catalunya
Publication Contribución de congreso (ConferenceOutput)

User interactions and negative examples to improve the learning of semantic rules in a cognitive exercise scenario

  • Suárez Hernández, Alejandro
  • Andriella, Antonio
  • Torras, Carme
  • Alenyà Ribas, Guillem
© 2023 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works., Enabling a robot to perform new tasks is a complex endeavor, usually beyond the reach of non-technical users. For this reason, research efforts that aim at empowering end-users to teach robots new abilities using intuitive modes of interaction are valuable. In this article, we present INtuitive PROgramming 2 (INPRO2), a learning framework that allows inferring planning actions from demonstrations given by a human teacher. INPRO2 operates in an assistive scenario, in which the robot may learn from a healthcare professional (a therapist or caregiver) new cognitive exercises that can be later administered to patients with cognitive impairment. INPRO2 features significant improvements over previous work, namely: (1) exploitation of negative examples; (2) proactive interaction with the teacher to ask questions about the legality of certain movements; and (3) learning goals in addition to legal actions. Through simulations, we show the performance of different proactive strategies for gathering negative examples. Real-world experiments with human teachers and a TIAGo robot are also presented to qualitatively illustrate INPRO2., The research leading to these results receives funding from the EU H2020 Programme under grant agreement ERC–2016–ADG–741930 (CLOTHILDE); by MCIN/ AEI /10.13039/501100011033 under the project CHLOE-GRAPH (PID2020-119244GB-I00); by MCIN/ AEI /10.13039/501100011033 and by the ”European Union NextGenerationEU/PRTR under the project COHERENT (PCI2020-120718-2); AA acknowledges the European Union’s Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No 801342 (Tecniospring INDUSTRY); ASH acknowledges the financial support of the Apadrina la Ciencia association and of the Ford Spanish branch., Peer Reviewed, Postprint (author's final draft)