| old_uid | 14443 |
|---|
| title | Reference Spreading Hybrid Control – Exploiting Dynamic Contact Transitions in Robotics Applications |
|---|
| start_date | 2017/10/04 |
|---|
| schedule | 15h |
|---|
| online | no |
|---|
| summary | Many of the tasks that we, as humans, perform on a daily basis require physical inter- action with our surrounding. Example tasks include object grasping and manipulation, walking, jumping, and climbing in environments with complex spatial geometry. One of the ambitions in robotics is creating machines that can easily and reliably perform such tasks, with the aim of alleviating mankind from performing necessary activities that are, however, either boring or dangerous for our life and health. While advances in the robotics field have been substantial in the latest fifty years, performing locomotion and manipulation tasks that involve contact transitions at relative high speed still poses many challenges. Besides the limitations and challenges in perception and actuation design, witnessed in current technology, a key challenge is to create a suitable control framework to reach human-like performance when dealing with dynamic contact transitions. The behavior of a robotic system performing locomotion and manipulation tasks can be described employing the theoretical framework of hybrid dynamical systems with state-triggered jumps. In the context of robotics system, state jumps correspond to the fast velocity changes occurring in making and braking contacts with the environment. Control and trajectory tracking for dynamical systems with state-triggered jumps is complicated due to the inevitable time mismatch between desired and closed-loop contact transitions that calls for a redefinition of the concept of tracking error. The study of the trajectory tracking problem is a relatively recent and active field of research in the control community, with possible impact on other research fields, beyond robotics. This talk will review the fundamentals of a control approach for the stabilization of hybrid systems with state-triggered jumps that goes under the name of reference spreading hybrid control [1, 2]. The talk will also discuss the application of reference spreading hybrid control for the closed-loop motion stabilization of mechanical and robotic systems undergoing dynamic contact transitions, by means of numerical simulations and real-world experiments [3, 4, 5].
References :
[1] A. Saccon, N. van de Wouw, H. Nijmeijer, Sensitivity analysis of hybrid systems with state jumps with application to trajectory tracking. IEEE Conference on Decision and Control (CDC), 3065–3070, 2014.
[2] M. Rijnen, A. Saccon, H. Nijmeijer, On optimal trajectory tracking for mechanical systems with unilateral constraints. IEEE Conference on Decision and Control (CDC), 2561-2566, 2015.
[3] G.P. Incremona, A. Saccon, A. Ferrara, H Nijmeijer, Trajectory tracking of mechanical systems with unilateral constraints: Experimental results of a recently introduced hybrid PD feedback controller. IEEE Conference on Decision and Control (CDC), 920–925, 2015.
[4] M.W.L.M. Rijnen, A.T. van Rijn, H. Dallali, A. Saccon, H. Nijmeijer, Hybrid Trajectory Tracking for a Hopping Robotic Leg. IFAC PSYCO, pp. 107–112, 2016.
[5] M. Rijnen, E. de Mooij, S. Traversaro, F. Nori, N. van de Wouw, A. Saccon, H. Nijmeijer, Control of Humanoid Robot Motions with Impacts: Numerical Experiments with Reference Spreading Control. IEEE ICRA 2017. |
|---|
| responsibles | Baumard |
|---|
| |