Leonel Rozo
Research Subjects
Leonel Rozo has a total of 32 co-authors. Here are the top co-authors by number of shared publications:
- Darwin Gordon Caldwell
Instituto Italiano di Tecnologia
5 shared publications - Sylvain Calinon
Idiap Research Institute
4 shared publications Georgios Arvanitidis
Technical University of Denmark (DTU)
2 shared publicationsGerhard Neumann
Karlsruhe Institute of Technology (KIT)
2 shared publications- Hadi Beik-Mohammadi
Bosch Center for Artificial Intelligence (BCAI)
2 shared publications - João Silvério
German Aerospace Center (DLR), Robotics and Mechatronics Center (RMC), Weling, Germany
2 shared publications
Latest Publications
- 5 September 2025
- Published by SAGE Publications in The International Journal of Robotics Research
- Vol. 45 (5) , 714-745
Extended neural contractive dynamical systems: On multiple tasks and Riemannian safety regions
Stability guarantees are crucial when ensuring that a fully autonomous robot does not take undesirable or potentially harmful actions. We recently proposed the Neural Contractive Dynamical Systems (NCDS) , which is a neural network architecture that guarantees contractive stability. With this, learning-from-demonstrations approaches can trivially provide stability guarantees. However, our early work left several unanswered questions, which we here address. Beyond providing an in-depth explanation of NCDS, this paper extends the framework with more careful regularization, a conditional variant of the framework for handling multiple tasks, and an uncertainty-driven approach to latent obstacle avoidance. Experiments verify that the developed system has the flexibility of ordinary neural networks while providing the stability guarantees needed for autonomous robotics.
- 5 September 2025
- Published by SAGE Publications in The International Journal of Robotics Research
- Vol. 45 (5) , 714-745
Extended neural contractive dynamical systems: On multiple tasks and Riemannian safety regions
Stability guarantees are crucial when ensuring that a fully autonomous robot does not take undesirable or potentially harmful actions. We recently proposed the Neural Contractive Dynamical Systems (NCDS) , which is a neural network architecture that guarantees contractive stability. With this, learning-from-demonstrations approaches can trivially provide stability guarantees. However, our early work left several unanswered questions, which we here address. Beyond providing an in-depth explanation of NCDS, this paper extends the framework with more careful regularization, a conditional variant of the framework for handling multiple tasks, and an uncertainty-driven approach to latent obstacle avoidance. Experiments verify that the developed system has the flexibility of ordinary neural networks while providing the stability guarantees needed for autonomous robotics.
- 21 August 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Robotics
- Vol. 41 (15523098) , 5327-5343
Fast and Robust Visuomotor Riemannian Flow Matching Policy
Diffusion-based visuomotor policies excel at learning complex robotic tasks by effectively combining visual data with high-dimensional, multi-modal action distributions. However, diffusion models often suffer from slow inference due to costly denoising processes or require complex sequential training arising from recent distilling approaches. This paper introduces Riemannian Flow Matching Policy (RFMP), a model that inherits the easy training and fast inference capabilities of flow matching (FM). Moreover, RFMP inherently incorporates geometric constraints commonly found in realistic robotic applications, as the robot state resides on a Riemannian manifold. To enhance the robustness of RFMP, we propose Stable RFMP (SRFMP), which leverages LaSalle's invariance principle to equip the dynamics of FM with stability to the support of a target Riemannian distribution. Rigorous evaluation on ten simulated and real-world tasks show that RFMP successfully learns and synthesizes complex sensorimotor policies on Euclidean and Riemannian spaces with efficient training and inference phases, outperforming Diffusion Policies and Consistency Policies.
- 21 August 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Robotics
- Vol. 41 (15523098) , 5327-5343
Fast and Robust Visuomotor Riemannian Flow Matching Policy
Diffusion-based visuomotor policies excel at learning complex robotic tasks by effectively combining visual data with high-dimensional, multi-modal action distributions. However, diffusion models often suffer from slow inference due to costly denoising processes or require complex sequential training arising from recent distilling approaches. This paper introduces Riemannian Flow Matching Policy (RFMP), a model that inherits the easy training and fast inference capabilities of flow matching (FM). Moreover, RFMP inherently incorporates geometric constraints commonly found in realistic robotic applications, as the robot state resides on a Riemannian manifold. To enhance the robustness of RFMP, we propose Stable RFMP (SRFMP), which leverages LaSalle's invariance principle to equip the dynamics of FM with stability to the support of a target Riemannian distribution. Rigorous evaluation on ten simulated and real-world tasks show that RFMP successfully learns and synthesizes complex sensorimotor policies on Euclidean and Riemannian spaces with efficient training and inference phases, outperforming Diffusion Policies and Consistency Policies.
- 5 September 2023
- Published by Elsevier in Robotics and Computer-Integrated Manufacturing
- Vol. 85
The e-Bike motor assembly: Towards advanced robotic manipulation for flexible manufacturing
- 5 September 2023
- Published by Elsevier in Robotics and Computer-Integrated Manufacturing
- Vol. 85
The e-Bike motor assembly: Towards advanced robotic manipulation for flexible manufacturing
- 28 August 2023
- Published by SAGE Publications in The International Journal of Robotics Research
- Vol. 42 (10) , 729-754
Reactive motion generation on learned Riemannian manifolds
In recent decades, advancements in motion learning have enabled robots to acquire new skills and adapt to unseen conditions in both structured and unstructured environments. In practice, motion learning methods capture relevant patterns and adjust them to new conditions such as dynamic obstacle avoidance or variable targets. In this paper, we investigate the robot motion learning paradigm from a Riemannian-manifold perspective. We argue that Riemannian manifolds may be learned via human demonstrations in which geodesics are natural motion skills. The geodesics are generated using a learned Riemannian metric produced by our novel variational autoencoder (VAE), which is intended to recover full-pose end-effector states and joint-space configurations. In addition, we propose a technique for facilitating on-the-fly end-effector/multiple-limb obstacle avoidance by reshaping the learned manifold using an obstacle-aware ambient metric. The motion generated using these geodesics may naturally result in multiple-solution tasks that have not been explicitly demonstrated previously. We extensively tested our approach in task-space and joint-space scenarios using a 7-DoF robotic manipulator. We demonstrate that our method is capable of learning and generating motion skills based on complicated motion patterns demonstrated by a human operator. Additionally, we assess several obstacle-avoidance strategies and generate trajectories in multiple-mode settings.
- 28 August 2023
- Published by SAGE Publications in The International Journal of Robotics Research
- Vol. 42 (10) , 729-754
Reactive motion generation on learned Riemannian manifolds
In recent decades, advancements in motion learning have enabled robots to acquire new skills and adapt to unseen conditions in both structured and unstructured environments. In practice, motion learning methods capture relevant patterns and adjust them to new conditions such as dynamic obstacle avoidance or variable targets. In this paper, we investigate the robot motion learning paradigm from a Riemannian-manifold perspective. We argue that Riemannian manifolds may be learned via human demonstrations in which geodesics are natural motion skills. The geodesics are generated using a learned Riemannian metric produced by our novel variational autoencoder (VAE), which is intended to recover full-pose end-effector states and joint-space configurations. In addition, we propose a technique for facilitating on-the-fly end-effector/multiple-limb obstacle avoidance by reshaping the learned manifold using an obstacle-aware ambient metric. The motion generated using these geodesics may naturally result in multiple-solution tasks that have not been explicitly demonstrated previously. We extensively tested our approach in task-space and joint-space scenarios using a 7-DoF robotic manipulator. We demonstrate that our method is capable of learning and generating motion skills based on complicated motion patterns demonstrated by a human operator. Additionally, we assess several obstacle-avoidance strategies and generate trajectories in multiple-mode settings.
- Open Access
Learning Task Priorities from Demonstrations
Bimanual operations in humanoids offer the possibility to carry out more than one manipulation task at the same time, which in turn introduces the problem of task prioritization. We address this problem from a learning from demonstration perspective, by extending the Task-Parameterized Gaussian Mixture Model (TP-GMM) to Jacobian and null space structures. The proposed approach is tested on bimanual skills but can be applied in any scenario where the prioritization between potentially conflicting tasks needs to be learned. We evaluate the proposed framework in: two different tasks with humanoids requiring the learning of priorities and a loco-manipulation scenario, showing that the approach can be exploited to learn the prioritization of multiple tasks in parallel.
Open AccessLearning Task Priorities from Demonstrations
Bimanual operations in humanoids offer the possibility to carry out more than one manipulation task at the same time, which in turn introduces the problem of task prioritization. We address this problem from a learning from demonstration perspective, by extending the Task-Parameterized Gaussian Mixture Model (TP-GMM) to Jacobian and null space structures. The proposed approach is tested on bimanual skills but can be applied in any scenario where the prioritization between potentially conflicting tasks needs to be learned. We evaluate the proposed framework in: two different tasks with humanoids requiring the learning of priorities and a loco-manipulation scenario, showing that the approach can be exploited to learn the prioritization of multiple tasks in parallel.
- 14 November 2018
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Robotics
- Vol. 35 (1) , 78-94
Learning Task Priorities from Demonstrations
- 14 November 2018
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Robotics
- Vol. 35 (1) , 78-94
Learning Task Priorities from Demonstrations
- Open Access
- 1 November 2018
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 153-159
Bimanual Skill Learning with Pose and Joint Space Constraints
As humanoid robots become commonplace, learning and control algorithms must take into account the new challenges imposed by this morphology, in order to fully exploit their potential. One of the most prominent characteristics of such robots is their bimanual structure. Most research on learning bimanual skills has focused on the coordination between end-effectors, exploiting operational space formulations. However, motion patterns in bimanual scenarios are not exclusive to operational space, also occurring at joint level. Moreover, in addition to position, the end-effector orientation is also essential for bimanual operation. Here, we propose a framework for simultaneously learning constraints in configuration and operational spaces, while considering end-effector orientations, commonly overlooked in previous works. In particular, we extend the Task-Parameterized Gaussian Mixture Model (TP-GMM) with novel Jacobian-based operators that address the foregoing problem. The proposed framework is evaluated in a bimanual task with the COMAN humanoid that requires the consideration of operational and configuration space motions.
- 1 November 2018
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 153-159
Open AccessBimanual Skill Learning with Pose and Joint Space Constraints
As humanoid robots become commonplace, learning and control algorithms must take into account the new challenges imposed by this morphology, in order to fully exploit their potential. One of the most prominent characteristics of such robots is their bimanual structure. Most research on learning bimanual skills has focused on the coordination between end-effectors, exploiting operational space formulations. However, motion patterns in bimanual scenarios are not exclusive to operational space, also occurring at joint level. Moreover, in addition to position, the end-effector orientation is also essential for bimanual operation. Here, we propose a framework for simultaneously learning constraints in configuration and operational spaces, while considering end-effector orientations, commonly overlooked in previous works. In particular, we extend the Task-Parameterized Gaussian Mixture Model (TP-GMM) with novel Jacobian-based operators that address the foregoing problem. The proposed framework is evaluated in a bimanual task with the COMAN humanoid that requires the consideration of operational and configuration space motions.
- 1 November 2018
- Published by Elsevier in Robotics and Autonomous Systems
- Vol. 109, 156-167
Force-based variable impedance learning for robotic manipulation
- 1 November 2018
- Published by Elsevier in Robotics and Autonomous Systems
- Vol. 109, 156-167
Force-based variable impedance learning for robotic manipulation
- 26 April 2018
- Published by Springer Nature in Autonomous Robots
- Vol. 42 (5) , 953-956
Special issue on learning for human–robot collaboration
- 26 April 2018
- Published by Springer Nature in Autonomous Robots
- Vol. 42 (5) , 953-956
Special issue on learning for human–robot collaboration
- 10 January 2017
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 2 (2) , 719-726
A Method for Derivation of Robot Task-Frame Control Authority from Repeated Sensory Observations
In this letter, we propose a novel method that enables the robot to autonomously devise an appropriate control strategy from human demonstrations without a prior knowledge of the demonstrated task. The method is primarily based on observing the patterns and consistency in the observed dataset. This is obtained through a demonstration setting that uses a motion capture system, a force sensor, and muscle activity measurements. The variables (position and force) in the collected dataset are then segmented and analysed for each axis of the observed task frame separately. While checking several conditions based on the consistency, value range, and magnitude of repeated observations, the appropriate controller (i.e., position or force) is delegated to each axis of the task frame. In the final stage, the method also checks for a correlation between variables and muscle activity patterns to determine the desired stiffness behaviour. The robot then uses the derived control strategies in autonomous operation through a hybrid force/impedance controller. To validate the proposed method, we performed experiments on real-life tasks involving physical interaction with the environment, where we considered surface wiping, material sawing, and drilling.
- 10 January 2017
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 2 (2) , 719-726
A Method for Derivation of Robot Task-Frame Control Authority from Repeated Sensory Observations
In this letter, we propose a novel method that enables the robot to autonomously devise an appropriate control strategy from human demonstrations without a prior knowledge of the demonstrated task. The method is primarily based on observing the patterns and consistency in the observed dataset. This is obtained through a demonstration setting that uses a motion capture system, a force sensor, and muscle activity measurements. The variables (position and force) in the collected dataset are then segmented and analysed for each axis of the observed task frame separately. While checking several conditions based on the consistency, value range, and magnitude of repeated observations, the appropriate controller (i.e., position or force) is delegated to each axis of the task frame. In the final stage, the method also checks for a correlation between variables and muscle activity patterns to determine the desired stiffness behaviour. The robot then uses the derived control strategies in autonomous operation through a hybrid force/impedance controller. To validate the proposed method, we performed experiments on real-life tasks involving physical interaction with the environment, where we considered surface wiping, material sawing, and drilling.