Diffusion-ES: Gradient-Free Planning with Diffusion for Autonomous and Instruction-Guided Driving

This paper proposes Diffusion-ES, a method that combines gradient-free optimization with trajectory denoising to optimize black-box non-differentiable objectives while staying in the data manifold and shows that Diffusion-ES outperforms existing sampling-based planners, reactive deterministic or diffusion-based policies, and reward-gradient guidance.