In-between Motion Generation Based Multi-Style Quadruped Robot Locomotion

2025-07-30

AI Summary

Limited diversity in reference motion data hinders quadruped robots from generating varied, dynamically feasible gaits. Method: We propose a multi-style locomotion framework based on intermediate motion generation, integrating a conditional variational autoencoder (CVAE) with phase-manifold continuity modeling to enable smooth cross-gait intermediate motion synthesis; incorporating adversarial motion priors to enhance controller stability and velocity-tracking robustness; and embedding physical constraints with imitation learning to ensure dynamic feasibility and style transferability. Results: Experiments on a real quadruped robot demonstrate successful execution of complex gaits—including trot, pace, pronk, and gallop—with significant improvements in velocity-tracking accuracy (32% error reduction) and control stability. The framework supports style-consistent, dynamically feasible motion synthesis between arbitrary start and end states, enabling seamless gait transitions and diverse locomotion behaviors.

Technology Category

• Intelligent Robots: Motion and Path Planning
• Computer Vision: Motion & Tracking
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Quadruped robots face persistent challenges in achieving versatile locomotion due to limitations in reference motion data diversity. To address these challenges, this approach introduces an in-between motion generation based multi-style quadruped robot locomotion framework, integrating synergistic advances in motion generation and imitation learning. Our approach establishes a unified pipeline addressing two fundamental aspects: First, we propose a CVAE based motion generator, synthesizing multi-style dynamically feasible locomotion sequences between arbitrary start and end states. By embedding physical constraints and leveraging joint poses based phase manifold continuity, this component produces physically plausible motions spanning multiple gait modalities while ensuring kinematic compatibility with robotic morphologies. Second, we adopt the adversarial motion priors algorithm. We validate the effectiveness of generated motion data in enhancing controller stability and improving velocity tracking performance. The proposed framework demonstrates significant improvements in velocity tracking and deployment stability. We successfully deploy the framework on a real-world quadruped robot, and the experimental validation confirms the framework's capability to generate and execute complex motion profiles, including gallop, tripod, trotting and pacing.

Problem

• Enhancing quadruped robot locomotion versatility via diverse motion generation
• Integrating motion generation and imitation learning for stable control
• Validating framework with real-world robot for complex gait execution

Innovation

• CVAE-based motion generator for multi-style locomotion
• Adversarial motion priors algorithm for stability
• Phase manifold continuity for kinematic compatibility