Offline Adaptation of Quadruped Locomotion using Diffusion Models

2024-11-13

arXiv.org

AI Summary

This work addresses two key challenges in quadrupedal locomotion control: weak multi-skills interpolation capability and difficulty in offline adaptation to novel gaits post-training. We propose an offline gait adaptation method based on classifier-free guided diffusion, enabling direct extraction of goal-conditioned behavioral policies from unlabeled motion data. The approach supports zero-shot gait generation, millisecond-level online interpolation, and real-time adaptation to new terrains or targets—without retraining. To our knowledge, this is the first application of classifier-free guided diffusion to quadrupedal motion control, achieving offline gait transfer with no fine-tuning. The policy is fully deployed on-board a resource-constrained CPU, exhibiting minimal computational overhead. Experiments on the ANYmal platform validate both effectiveness and embedded real-time performance, demonstrating sub-10 ms inference latency and robust gait generalization across diverse locomotion tasks and environments.

Technology Category

• Intelligent Robots: Behavior Learning & Control
• Humans and AI: Human-Aware Planning and Behavior Prediction
• Machine Learning: Imitation Learning & Inverse Reinforcement Learning

Application Category

• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

We present a diffusion-based approach to quadrupedal locomotion that simultaneously addresses the limitations of learning and interpolating between multiple skills and of (modes) offline adapting to new locomotion behaviours after training. This is the first framework to apply classifier-free guided diffusion to quadruped locomotion and demonstrate its efficacy by extracting goal-conditioned behaviour from an originally unlabelled dataset. We show that these capabilities are compatible with a multi-skill policy and can be applied with little modification and minimal compute overhead, i.e., running entirely on the robots onboard CPU. We verify the validity of our approach with hardware experiments on the ANYmal quadruped platform.

Problem

• Addresses limitations in learning and interpolating multiple locomotion skills.
• Enables offline adaptation to new locomotion behaviors post-training.
• Demonstrates goal-conditioned behavior extraction from unlabelled datasets.

Innovation

• Diffusion models for quadruped locomotion adaptation
• Classifier-free guided diffusion for behavior extraction
• Onboard CPU execution with minimal compute overhead