Guided Riemannian Optimization (GuRO): Bridging Model Predictive Control and Decision Transformers
Industry
Source: arXiv cs.ROPublish time unverified
arXiv:2608.23204v1 Announce Type: new Abstract: Decision-making in high-dimensional, nonlinear systems remains a central challenge in robotics. While model-based methods like Model Predictive Control (MPC) offer sample efficiency and interpretability, their performance degrades when the dynamics model is inaccurate or long-horizon predictions are required.