Domingo Esteban
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Domingo Esteban has a total of 11 co-authors. Here are the top co-authors by number of shared publications:
Leonel Rozo
Bosch Center for Artificial Intelligence,Renningen,Germany,71272
3 shared publicationsDarwin G. Caldwell
Istituto Italiano di Tecnologia,Department of Advanced Robotics,Genova,Italy,16163
2 shared publications- Alberto Jardón
Universidad Carlos III de Madrid
1 shared publication Brian Delhaisse
Department of Advanced Robotics, Istituto Italiano di Tecnologia Via Morego 30, 16163 Genova, Italy
1 shared publication- Carlos Balaguer
Universidad Carlos III de Madrid
1 shared publication - Claudio Semini
Istituto Italiano di Tecnologia
1 shared publication
Latest Publications
- Open Access
- 29 November 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Robotics
- Vol. 39 (2) , 885-904
ViTAL: Vision-Based Terrain-Aware Locomotion for Legged Robots
This article focuses on vision-based planning strategies for legged robots that separate locomotion planning into foothold selection and pose adaptation. Current pose adaptation strategies optimize the robot's body pose relative to given footholds. If these footholds are not reached, the robot may end up in a state with no reachable safe footholds. Therefore, we present a vision-based terrain-aware locomotion (ViTAL) strategy that consists of novel pose adaptation and foothold selection algorithms. ViTAL introduces a different paradigm in pose adaptation that does not optimize the body pose relative to given footholds, but the body pose that maximizes the chances of the legs in reaching safe footholds. ViTAL plans footholds and poses based on skills that characterize the robot's capabilities and its terrain awareness. We use the 90-kg HyQ and 140-kg HyQReal quadruped robots to validate ViTAL and show that they are able to climb various obstacles, including stairs, gaps, and rough terrains, at different speeds and gaits. We compare ViTAL with a baseline strategy that selects the robot pose based on given selected footholds and show that ViTAL outperforms the baseline.
- 29 November 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Robotics
- Vol. 39 (2) , 885-904
Open AccessViTAL: Vision-Based Terrain-Aware Locomotion for Legged Robots
This article focuses on vision-based planning strategies for legged robots that separate locomotion planning into foothold selection and pose adaptation. Current pose adaptation strategies optimize the robot's body pose relative to given footholds. If these footholds are not reached, the robot may end up in a state with no reachable safe footholds. Therefore, we present a vision-based terrain-aware locomotion (ViTAL) strategy that consists of novel pose adaptation and foothold selection algorithms. ViTAL introduces a different paradigm in pose adaptation that does not optimize the body pose relative to given footholds, but the body pose that maximizes the chances of the legs in reaching safe footholds. ViTAL plans footholds and poses based on skills that characterize the robot's capabilities and its terrain awareness. We use the 90-kg HyQ and 140-kg HyQReal quadruped robots to validate ViTAL and show that they are able to climb various obstacles, including stairs, gaps, and rough terrains, at different speeds and gaits. We compare ViTAL with a baseline strategy that selects the robot pose based on given selected footholds and show that ViTAL outperforms the baseline.
- 1 November 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1818-1825
Hierarchical Reinforcement Learning for Concurrent Discovery of Compound and Composable Policies
A common strategy to deal with the expensive reinforcement learning (RL) of complex tasks is to decompose them into a collection of subtasks that are usually simpler to learn as well as reusable for new problems. However, when a robot learns the policies for these subtasks, common approaches treat every policy learning process separately. Therefore, all these individual (composable) policies need to be learned before tackling the learning process of the complex task through policies composition. Moreover, such composition of individual policies is usually performed sequentially, which is not suitable for tasks that require to perform the subtasks concurrently. In this paper, we propose to combine a set of composable Gaussian policies corresponding to these subtasks using a set of activation vectors, resulting in a complex Gaussian policy that is a function of the means and covariances matrices of the composable policies. Moreover, we propose an algorithm for learning both compound and composable policies within the same learning process by exploiting the off-policy data generated from the compound policy. The algorithm is built on a maximum entropy RL approach to favor exploration during the learning process. The results of the experiments show that the experience collected with the compound policy permits not only to solve the complex task but also to obtain useful composable policies that successfully perform in their corresponding subtasks.
- 1 November 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1818-1825
Hierarchical Reinforcement Learning for Concurrent Discovery of Compound and Composable Policies
A common strategy to deal with the expensive reinforcement learning (RL) of complex tasks is to decompose them into a collection of subtasks that are usually simpler to learn as well as reusable for new problems. However, when a robot learns the policies for these subtasks, common approaches treat every policy learning process separately. Therefore, all these individual (composable) policies need to be learned before tackling the learning process of the complex task through policies composition. Moreover, such composition of individual policies is usually performed sequentially, which is not suitable for tasks that require to perform the subtasks concurrently. In this paper, we propose to combine a set of composable Gaussian policies corresponding to these subtasks using a set of activation vectors, resulting in a complex Gaussian policy that is a function of the means and covariances matrices of the composable policies. Moreover, we propose an algorithm for learning both compound and composable policies within the same learning process by exploiting the off-policy data generated from the compound policy. The algorithm is built on a maximum entropy RL approach to favor exploration during the learning process. The results of the experiments show that the experience collected with the compound policy permits not only to solve the complex task but also to obtain useful composable policies that successfully perform in their corresponding subtasks.
- 1 November 2018
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-9
Learning Deep Robot Controllers by Exploiting Successful and Failed Executions
The prohibitively amount of data required when learning complex nonlinear policies, such as deep neural networks, has been significantly reduced with guided policy search (GPS)algorithms. However, while learning the control policy, the robot might fail and therefore generate unacceptable guiding samples. Failures may arise, for example, as a consequence of modeling or environmental uncertainties, and thus unsuccessful interactions should be explicitly considered while learning a complex policy. Currently, GPS methods update the robot policy discarding or giving low probability to unsuccessful trials. In other words, these methods overlook the existence of poorly performing executions, and therefore tend to underestimate the information of these interactions in next iterations. In this paper we propose to learn deep neural network controllers with an extension of G PS that considers trajectories optimized with dualist constraints. These constraints are aimed at assisting the policy learning so that the trajectory distributions updated at each iteration are similar to good trajectory distributions (e.g., sucessful executions)while differing from bad trajectory distributions (e.g. failures). We show that neural network policies guided by trajectories optimized with our method reduce the failures during the policy exploration phase, and therefore encourage safer interactions. This may have a relevant impact in tasks that involve physical contact with the environment or human partners.
- 1 November 2018
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-9
Learning Deep Robot Controllers by Exploiting Successful and Failed Executions
The prohibitively amount of data required when learning complex nonlinear policies, such as deep neural networks, has been significantly reduced with guided policy search (GPS)algorithms. However, while learning the control policy, the robot might fail and therefore generate unacceptable guiding samples. Failures may arise, for example, as a consequence of modeling or environmental uncertainties, and thus unsuccessful interactions should be explicitly considered while learning a complex policy. Currently, GPS methods update the robot policy discarding or giving low probability to unsuccessful trials. In other words, these methods overlook the existence of poorly performing executions, and therefore tend to underestimate the information of these interactions in next iterations. In this paper we propose to learn deep neural network controllers with an extension of G PS that considers trajectories optimized with dualist constraints. These constraints are aimed at assisting the policy learning so that the trajectory distributions updated at each iteration are similar to good trajectory distributions (e.g., sucessful executions)while differing from bad trajectory distributions (e.g. failures). We show that neural network policies guided by trajectories optimized with our method reduce the failures during the policy exploration phase, and therefore encourage safer interactions. This may have a relevant impact in tasks that involve physical contact with the environment or human partners.
- 1 May 2017
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 4142-4149
Transfer learning of shared latent spaces between robots with similar kinematic structure
Learning complex manipulation tasks often requires to collect a large training dataset to obtain a model of a specific skill. This process may become laborious when dealing with high-DoF robots, and even more tiresome if the skill needs to be learned by multiple robots. In this paper, we investigate how this learning process can be accelerated by using shared latent variable models for knowledge transfer among similar robots in an imitation setting. For this purpose, we take advantage of a shared Gaussian process latent variable model to learn a common latent representation of robot skills. Such representation is then reused as prior information to train new robots by reducing the learning process to a latent-to-output mapping. We show that our framework exhibits faster training convergence and similar performance when compared to single- and multi-robot models. All experiments were conducted in simulation on three different robotic platforms: WALK-MAN, COMAN and CENTAURO robots.
- 1 May 2017
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 4142-4149
Transfer learning of shared latent spaces between robots with similar kinematic structure
Learning complex manipulation tasks often requires to collect a large training dataset to obtain a model of a specific skill. This process may become laborious when dealing with high-DoF robots, and even more tiresome if the skill needs to be learned by multiple robots. In this paper, we investigate how this learning process can be accelerated by using shared latent variable models for knowledge transfer among similar robots in an imitation setting. For this purpose, we take advantage of a shared Gaussian process latent variable model to learn a common latent representation of robot skills. Such representation is then reused as prior information to train new robots by reducing the learning process to a latent-to-output mapping. We show that our framework exhibits faster training convergence and similar performance when compared to single- and multi-robot models. All experiments were conducted in simulation on three different robotic platforms: WALK-MAN, COMAN and CENTAURO robots.
- 1 November 2014
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- No. 21640572,p. 146-151
Anticipative humanoid postural control system for locomotive tasks
Postural control of humanoid robots during locomotion tasks has been typically focused in keeping balance by means of classic feedback control systems. The study of human behaviour in postural control reveals the existence of an anticipative component for postural control introduced by means of a feedforward system. This anticipative subsystem is in charge of preparing the human or humanoid system to react against perturbations during locomotion. In this way, the reactions can be more complex and higher perturbations levels can be overcome. The anticipative system is supported by the evaluation of sensory inputs by means of applying concepts extracted from the Surprise Theory. The data captured from sensory sources are transformed into perceptions and then compared with pre-established parameters. The comparison mismatch causes the elicitation of unexpected events or surprises. Then, these events are evaluated to generate a parameter that can be used to fulfil motor reaction patters called synergies. Finally, the new anticipative system is enabled by the use of neuro-fuzzy human style reasoning computing modules.
- 1 November 2014
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- No. 21640572,p. 146-151
Anticipative humanoid postural control system for locomotive tasks
Postural control of humanoid robots during locomotion tasks has been typically focused in keeping balance by means of classic feedback control systems. The study of human behaviour in postural control reveals the existence of an anticipative component for postural control introduced by means of a feedforward system. This anticipative subsystem is in charge of preparing the human or humanoid system to react against perturbations during locomotion. In this way, the reactions can be more complex and higher perturbations levels can be overcome. The anticipative system is supported by the evaluation of sensory inputs by means of applying concepts extracted from the Surprise Theory. The data captured from sensory sources are transformed into perceptions and then compared with pre-established parameters. The comparison mismatch causes the elicitation of unexpected events or surprises. Then, these events are evaluated to generate a parameter that can be used to fulfil motor reaction patters called synergies. Finally, the new anticipative system is enabled by the use of neuro-fuzzy human style reasoning computing modules.