Diffusion-based learning of contact plans for agile locomotion
🤖 AI Summary
Agile locomotion of quadrupedal robots on narrow-constrained terrains (e.g., stepping stones) remains challenging due to stringent dynamic feasibility and contact-scheduling requirements.
Method: This paper proposes a synergistic MCTS-NMPC framework that, for the first time, incorporates a conditional diffusion model into contact sequence planning to explicitly capture the multimodal distribution of optimal policies. The approach integrates Monte Carlo Tree Search (MCTS) for global discrete contact planning with nonlinear model predictive control (NMPC) for real-time whole-body trajectory tracking, augmented by supervised policy learning to enhance generalization.
Results: Evaluated on the Solo12 platform, the method generates jumping trajectories over stepping stones in seconds per planning instance and demonstrates adaptive responses to environmental disturbances during deployment. Experiments show substantial improvements in locomotion agility, robustness, and dynamic feasibility on complex constrained terrains.
Application Category
📝 Abstract
Legged robots have become capable of performing highly dynamic maneuvers in the past few years. However, agile locomotion in highly constrained environments such as stepping stones is still a challenge. In this paper, we propose a combination of model-based control, search, and learning to design efficient control policies for agile locomotion on stepping stones. In our framework, we use nonlinear model predictive control (NMPC) to generate whole-body motions for a given contact plan. To efficiently search for an optimal contact plan, we propose to use Monte Carlo tree search (MCTS). While the combination of MCTS and NMPC can quickly find a feasible plan for a given environment (a few seconds), it is not yet suitable to be used as a reactive policy. Hence, we generate a dataset for optimal goal-conditioned policy for a given scene and learn it through supervised learning. In particular, we leverage the power of diffusion models in handling multi-modality in the dataset. We test our proposed framework on a scenario where our quadruped robot Solo12 successfully jumps to different goals in a highly constrained environment (video).
Problem
Research questions and friction points this paper is trying to address.
Agile locomotion in constrained environments like stepping stones
Combining model-based control, search, and learning for efficient policies
Using diffusion models to handle multi-modality in dataset learning
Innovation
Methods, ideas, or system contributions that make the work stand out.
Uses nonlinear model predictive control (NMPC)
Employs Monte Carlo tree search (MCTS)
Leverages diffusion models for learning
🔎 Similar Papers
No similar papers found.
