Using reiforcement learning to build better robots through the co-optimization of robot control and structure


Project Members: Christian Warloe | Gopi Suresh | Prathyush Katukojwala | Jai Srivastav |


We seek to use reinforcement learning to design better robots. Taking advantage of the parameterized designs which can be created in RoCo, we hope to create better controllers tailored to optimized designs. Using simulation learning environments we hope to be able to create a process which automatically determines the best parameters for a robot based on a task and environment specification.