Rapidly Adaptable Legged Robots via Evolutionary Meta-Learning
This work addresses the challenge of rapid adaptation for legged robots in noisy, real-world environments, representing a domain-specific incremental improvement.
The paper tackles the problem of enabling legged robots to quickly adapt to changes in dynamics by introducing a meta-learning method that combines evolutionary strategies with a noise-tolerant adaptation operator, resulting in the robot adapting its policy with less than 3 minutes of real data and outperforming prior gradient-based approaches.
Learning adaptable policies is crucial for robots to operate autonomously in our complex and quickly changing world. In this work, we present a new meta-learning method that allows robots to quickly adapt to changes in dynamics. In contrast to gradient-based meta-learning algorithms that rely on second-order gradient estimation, we introduce a more noise-tolerant Batch Hill-Climbing adaptation operator and combine it with meta-learning based on evolutionary strategies. Our method significantly improves adaptation to changes in dynamics in high noise settings, which are common in robotics applications. We validate our approach on a quadruped robot that learns to walk while subject to changes in dynamics. We observe that our method significantly outperforms prior gradient-based approaches, enabling the robot to adapt its policy to changes based on less than 3 minutes of real data.