Breaking Task Impasses Quickly: Adaptive Neuro-Symbolic Learning for Open-World Robotics
Pierrick Lorang
TL;DR
The paper tackles open-world robotics by addressing the shortcomings of pure reinforcement learning and pure symbolic planning through a neuro-symbolic framework that blends hierarchical action abstractions, symbolic goal-oriented learning, and curiosity-driven imagination. By connecting planning and learning via executors and world models, the approach enables rapid adaptation to unforeseen novelties in dynamic environments. Key contributions include hierarchical nested action abstraction, symbolic goal-conditioned learning with HER, curiosity-driven symbolic imagination, and object-oriented skill learning, validated in robotic manipulation and autonomous driving with strong performance gains and faster convergence. The results indicate improved sample efficiency, robustness, and adaptability, highlighting the potential for real-world deployment of neuro-symbolic learning in open-world robotics.
Abstract
Adapting to unforeseen novelties in open-world environments remains a major challenge for autonomous systems. While hybrid planning and reinforcement learning (RL) approaches show promise, they often suffer from sample inefficiency, slow adaptation, and catastrophic forgetting. We present a neuro-symbolic framework integrating hierarchical abstractions, task and motion planning (TAMP), and reinforcement learning to enable rapid adaptation in robotics. Our architecture combines symbolic goal-oriented learning and world model-based exploration to facilitate rapid adaptation to environmental changes. Validated in robotic manipulation and autonomous driving, our approach achieves faster convergence, improved sample efficiency, and superior robustness over state-of-the-art hybrid methods, demonstrating its potential for real-world deployment.