Header logo is dlg
3 results


Thumb xl lcsslip
Beyond Basins of Attraction: Evaluating Robustness of Natural Dynamics

Steve Heim, , Sproewitz, A.

arXiv preprint arXiv:1806.08081, 2018 (unpublished)

It is commonly accepted that properly designing a system to exhibit favorable natural dynamics can greatly simplify designing or learning the control policy. It is however still unclear what constitutes favorable natural dynamics, and how to quantify its effect. Most studies of simple walking and running models have focused on the basins of attraction of passive limit-cycles, and the notion of self-stability. We emphasize instead the importance of stepping beyond basins of attraction. We show an approach based on viability theory to quantify robustness, valid for the family of all robust control policies. This allows us to evaluate the robustness inherent to the natural dynamics before designing the control policy or specifying a control objective. We illustrate this approach on simple spring mass models of running and show previously unexplored advantages to using a nonlinear leg stiffness. We believe designing robots with robust natural dynamics is particularly important for enabling learning control policies directly in hardware.

arXiv preprint arXiv:1806.08081 link (url) Project Page [BibTex]

Thumb xl screen shot 2018 04 18 at 11.01.27 am
Learning from Outside the Viability Kernel: Why we Should Build Robots that can Fail with Grace

Heim, S., Spröwitz, A.

Proceedings of SIMPAR 2018, pages: 55-61, IEEE, 2018 IEEE International Conference on Simulation, Modeling, and Programming for Autonomous Robots (SIMPAR), May 2018 (conference)

link (url) DOI Project Page [BibTex]

link (url) DOI Project Page [BibTex]

Thumb xl screen shot 2018 02 03 at 9.09.06 am
Shaping in Practice: Training Wheels to Learn Fast Hopping Directly in Hardware

Heim, S., Ruppert, F., Sarvestani, A., Spröwitz, A.

In Proceedings of the IEEE International Conference on Robotics and Automation (ICRA) 2018, pages: 5076-5081, IEEE, International Conference on Robotics and Automation, May 2018 (inproceedings)

Learning instead of designing robot controllers can greatly reduce engineering effort required, while also emphasizing robustness. Despite considerable progress in simulation, applying learning directly in hardware is still challenging, in part due to the necessity to explore potentially unstable parameters. We explore the of concept shaping the reward landscape with training wheels; temporary modifications of the physical hardware that facilitate learning. We demonstrate the concept with a robot leg mounted on a boom learning to hop fast. This proof of concept embodies typical challenges such as instability and contact, while being simple enough to empirically map out and visualize the reward landscape. Based on our results we propose three criteria for designing effective training wheels for learning in robotics.

Video Youtube link (url) Project Page [BibTex]

Video Youtube link (url) Project Page [BibTex]