KiRAS: Keyframe Guided Self-Imitation for Robust and Adaptive Skill Learning in Quadruped Robots

2026-03-16

AI Summary

Existing methods struggle to generalize due to the scarcity of diverse multi-skill data in complex unstructured terrains. This work proposes a keyframe-guided self-imitation learning framework that acquires a repertoire of locomotion skills on flat terrain without requiring expert demonstrations. By leveraging proficiency-based skill initialization, the framework enables continual training on rough terrain while effectively mitigating catastrophic forgetting. Central to the approach is the use of keyframes as a universal skill representation, integrated within an end-to-end reinforcement learning architecture to yield a lightweight, switchable multi-skill control policy. Experiments on Solo-8 and Unitree Go1 quadrupedal robots demonstrate robust execution and smooth transitions among multiple skills in challenging terrains, establishing an efficient platform for multi-skill data collection.

Technology Category

• Machine Learning: Imitation Learning & Inverse Reinforcement Learning
• Intelligent Robots: Behavior Learning & Control
• Multiagent Systems: Multiagent Learning

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

With advances in reinforcement learning and imitation learning, quadruped robots can acquire diverse skills within a single policy by imitating multiple skill-specific datasets. However, the lack of datasets on complex terrains limits the ability of such multi-skill policies to generalize effectively in unstructured environments. Inspired by animation, we adopt keyframes as minimal and universal skill representations, relaxing dataset constraints and enabling the integration of terrain adaptability with skill diversity. We propose Keyframe Guided Self-Imitation for Robust and Adaptive Skill Learning (KiRAS), an end-to-end framework for acquiring and transitioning between diverse skill primitives on complex terrains. KiRAS first learns diverse skills on flat terrain through keyframe-guided self-imitation, eliminating the need for expert datasets; then continues training the same policy network on rough terrains to enhance robustness. To eliminate catastrophic forgetting, a proficiency-based Skill Initialization Technique is introduced. Experiments on Solo-8 and Unitree Go1 robots show that KiRAS enables robust skill acquisition and smooth transitions across challenging terrains. This framework demonstrates its potential as a lightweight platform for multi-skill generation and dataset collection. It further enables flexible skill transitions that enhance locomotion on challenging terrains.

Problem

• quadruped robots
• skill generalization
• complex terrains
• multi-skill policy
• dataset limitation

Innovation

• keyframe-guided learning
• self-imitation
• quadruped locomotion
• skill generalization
• catastrophic forgetting mitigation