Bio-inspired & Neuromorphic Control of Soft Robots
We are exploring how learning-based, bio-inspired, and neuromorphic control techniques can be combined with the compliant physics of a soft robot to improve control. A soft robot’s infinite degrees of freedom often frustrate traditional control techniques. Use of learning-based techniques such as reinforcement learning allow us to avoid having to explicitly model these compliant dynamics with the RL algorithm instead implicitly learning a control policy. Bio-inspired and neuromorphic control extends this by adopting strategies evolved by biological creates to solve this challenge.