Many current quadrupedal robots use rigid feet for motion, which allows model-based controllers to simulate interactions with the ground as point contacts. However, the ability to control the force profile and geometry on the robot through the use of compliant feet can improve robot performance and agility. The introduction of larger contact areas and higher controller precision can increase the variety of gaits, top speed and ability to traverse complex terrains. This will diversify the set of possible tasks quadrupedal robots can perform, especially in cluttered and dynamic environments. My role will be to develop suitable simulation environments for this task and then use a learning-based approach to adjust previous model-based controllers to perform well in the new system.

In my recent coursework and internships, I have started to learn about the power of machine learning and controls algorithms. In pursuing this SuperUROP, I would like to explore the research aspect of these fields and learn what it takes to make novel developments. Furthermore, the physical aspect of this project and its applications to robotics interests me very much and makes me excited to work on the cutting edge of this technology.