RobotSL · Robot skill learning: imitation, exploitation and control
Horizon 2020 — Marie Skłodowska-Curie Actions
- Duration
- 2021-04-01 → 2023-03-31
- EU contribution
- €212,934
- Participants
- 1
- Scheme
- MSCA-IF
Lines connect the coordinator with its partners.
Results in brief
Robot skill learning: imitation, exploitation and control
Imitation learning has emerged as an important research direction in the community of robot learning. Due to its nature and user-friendly features, imitation learning has achieved great success in a myriad of scenarios, e.g., surgical robots, agricultural robots, and human-robot collaboration. It is well-known that robots often encounter various constraints, e.g., the robot’s end-effector must comply with the plane constraint when wiping a table or writing letters on a board. Moreover, robots must always meet their intrinsic constraints, e.g., motion range, joint and torque limits. In addition, environmental priors and multi-modal features will help to learn skills in complicated and unstructured scenarios. The main goal of the project is to provide an imitation learning framework that can handle various external and internal constraints, and explore environment priors and multi-modal features. The success of the project can bring direct benefits to society, such as equipping a robot with the friendly human-robot interaction ability to take care of the elderly and cope with daily tasks including cleaning floors and carrying heavy objects.
Data: CORDIS, © European Union
Project objective
In this project, I will develop an imitation learning framework for robot skill learning and optimization, aiming at endowing robots with versatile skills and thus allowing robots to work in broad application domains. This framework will handle various constraints (e.g., robot joint limit, trajectory smoothness, obstacle avoidance) that robots encounter in practice, exploit environmental priors and multi-modal properties underlying human demonstrations, as well as design a low-level optimal controller so as to drive robots to execute human-like motions and resist external perturbations. The project objectives and associated concepts are original and novel. This project will provide the first solution for the problem of imitation learning with various constraints (including linear and non-linear, convex and non-convex constraints) and a novel concept of semi-imitation learning by exploring environmental priors. Moreover, it will provide a solution to multi-modal imitation learning from few demonstrations, which can be readily combined with constrained learning and environmental priors. In addition, from a control perspective, this project will study a new concept of control-inspired imitation learning to mimic both human skills and human reactions under perturbations. This project is challenging in the sense that it involves robotics, imitation learning, probability theory, optimization, semi-supervised learning, clustering techniques and optimal control. I will work closely with Prof. Cohn, who is an expert in knowledge representation and reasoning. This fellowship will sharpen my research skills and extend my research network in Leeds and Europe. Specifically, this fellowship will enable me to dive deeper into the challenging but essential problems in robot imitation learning, which will provide new insights and research topics to the community of robot learning, positioning me as a competitive researcher in the community.
Original text from CORDIS.
Participants
- UNIVERSITY OF LEEDS · LeedsCoordinatorUnited Kingdom
Links
Data: CORDIS, © European Union