What are the algorithmic principles that would allow a robot to run through a rocky terrain, lift a couch while reaching for an object that rolled under it or manipulate a screwdriver while balancing on top of a ladder? By answering these questions, we try to understand the fundamental principles for robot locomotion and manipulation that will endow robots with the robustness and adaptability necessary to efficiently and autonomously act in an unknown and changing environment.
Our research assumption is that understanding how robots should move is a necessary step towards autonomy and a comprehensive theory of robot movement is needed. This theory should have at least three important properties:
- it can be used to control any robot with legs and arms for both manipulation and locomotion tasks,
- it allows robots to constantly improve their performances as they experience the world,
- it is fully automated, (i.e. no need for time-intensive engineering each time a new robot is used or a new task needs to be performed).
With this goal in mind, our research agenda follows several complementary directions that define a consistent research program for the generation of movements in autonomous robots. In particular, we explore problems related to high performance torque control, contact interactions, reactive motion planning and movement learning and we apply our research to both locomotion and manipulation problems.