Convex Controller Synthesis for Robot Contact

1 minute read

By Hung Pham and Quang-Cuong Pham, in IEEE Robotics and Automation Letters, 2020. Also presented at ICRA 2020. [paper (arXiv)]

Introducing CCS

In many applications of robotics, it is essential that robots response to external forces properly when it makes contact with the external environment. This class of problems is almost as old as robotics itself: The first seminal works date back to the 80s. And there have been many research since then, most notably Admittance/Impedance control, Hybrid Force control.

What one can achieve with these classical approaches is quite limited. Consider admittance controller, there are three parameters: mass, spring and damper. It’s hard to believe that the best controller is realized by a selection of just these three parameters. And yet, finding a good set of parameters is not at all easy. Should you increase the mass or the damping coefficients. And which direction?

These are hard questions. It would be ideal to have an algorithm that can perform the search for the best controller automatically using a more generic structure, without the limitation of a fixed controller structure. This is exactly what I propose in this work.

Our Experiments

Two experiments we perform in preparation for the manuscript: Hand-guiding and force control using the same.

Code

github repository

Paper Abstract

Controlling contacts is truly challenging, and this has been a major hurdle to deploying industrial robots into unstructured/human-centric environments. More specifically, the main challenges are: (i) how to ensure stability at all times; (ii) how to satisfy task-specific performance specifications; (iii) how to achieve (i) and (ii) under environment uncertainty, robot parameters uncertainty, sensor and actuator time delays, external perturbations, etc. Here, we propose a new approach – Convex Controller Synthesis (CCS) – to tackle the above challenges based on robust control theory and convex optimization. In two physical interaction tasks – robot hand guiding and sliding on surfaces with different and unknown stiffnesses – we show that CCS controllers outperform their classical counterparts in an essential way.