LLA-MPPI: Rapidly Adaptive Whole-body Control of Legged Robots with GPU-Accelerated Parallel Simulations

2026-10-07

AI Summary

This study addresses the performance degradation of whole-body control in legged robots under dynamic disturbances and the reliance of existing adaptive methods on offline training. We propose a pretraining-free, real-time adaptive control framework that reformulates dynamics adaptation as a model selection task. By integrating GPU-parallelized batch contact simulation with a Model Predictive Path Integral (MPPI) planner, the method employs a prediction-error-based windowing strategy to identify the optimal dynamics model online. This paradigm offers physical interpretability and substantially lowers deployment barriers. Experimental results demonstrate that our approach achieves a 97.5% success rate in simulation, significantly outperforming baselines. Furthermore, its robust locomotion capabilities are validated on hardware across challenging scenarios, including payload carrying, leg failure, and box pushing.

Abstract

Real-time whole-body controllers for legged robots typically plan through a fixed nominal model and degrade when the deployed dynamics change. Adaptive methods typically require a model structure that contact dynamics do not provide, or they need offline training for each anticipated condition. We present Look-back and Look-ahead Adaptive Model Predictive Path Integral control (LLA-MPPI). The method converts whole-body adaptation into selection over a bank of GPU-batched contact simulators with different physical or structural parameters. Windowed prediction errors select the simulator that best explains recent motion. A whole-body MPPI planner optimizes controls through the selected model. The framework requires no offline training, and its selected hypotheses are physically interpretable. Across four simulated tasks, it achieves 97.5% success while the strongest baseline reaches 74% and an oracle with the true model reaches 98.5%. Hardware validation on a Unitree Go2 shows the robot walking under a payload added mid-run, walking after one leg is disabled, and pushing a box to its goal while increasing its mass on the fly. Code, videos, and project details are available at: https://lla-control.github.io

Problem

• legged robots
• whole-body control
• adaptive control
• contact dynamics
• model mismatch

Innovation

• Adaptive Whole-body Control
• Model Predictive Path Integral (MPPI)
• GPU-Accelerated Simulation
• Legged Robots
• Contact Dynamics