SurGE: Surrogate Gradient-guided Evolution for Co-design of Legged Robots with Parallel Elasticity

Name: SurGE: Surrogate Gradient-guided Evolution for Co-design of Legged Robots with Parallel Elasticity
Uploaded: 2026-10-08T13:05:14.734900+00:00
Duration: 602 s
Description: SurGE is a framework for co-designing legged robots with elastic elements by overcoming the non-differentiability of contact dynamics. It uses a differentiable surrogate model and an RL policy to compute gradients, which are then injected into CMA-ES. This hybrid approach significantly reduces optimization variance and improves convergence speed compared to standard methods.

SurGE is a framework for co-designing legged robots with elastic elements by overcoming the non-differentiability of contact dynamics. It uses a differentiable surrogate model and an RL policy to compute gradients, which are then injected into CMA-ES. This hybrid approach significantly reduces optimization variance and improves convergence speed compared to standard methods.

Open the full episode