An Interpretable Neural Control Network with Adaptable Online Learning for Sample Efficient Robot Locomotion Learning

2025-01-18

AI Summary

To address the dual challenges of low sample efficiency and poor policy interpretability in robotic gait learning, this paper proposes SME-AGOL: a novel framework comprising SME—a three-layer interpretable neural controller that innovatively decouples sequential latent states from trigonometric basis functions to explicitly encode key postures—and AGOL—an adaptive gradient-correlation weighting mechanism enabling online dynamic optimization of parameter updates. SME-AGOL is the first approach to jointly integrate interpretable structural modeling with gradient-aware learning. Evaluated on a simulated hexapod robot, it reduces sample requirements by 40% and increases final reward by 150% over baselines. On a physical platform, it achieves end-to-end gait acquisition from scratch in just 10 minutes—significantly outperforming state-of-the-art methods in both data efficiency and real-world deployability.

Technology Category

• Intelligent Robots: Learning & Optimization for ROB
• Multiagent Systems: Multiagent Learning
• Search and Optimization: Learning to Search

Application Category

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

Abstract

Robot locomotion learning using reinforcement learning suffers from training sample inefficiency and exhibits the non-understandable/black-box nature. Thus, this work presents a novel SME-AGOL to address such problems. Firstly, Sequential Motion Executor (SME) is a three-layer interpretable neural network, where the first produces the sequentially propagating hidden states, the second constructs the corresponding triangular bases with minor non-neighbor interference, and the third maps the bases to the motor commands. Secondly, the Adaptable Gradient-weighting Online Learning (AGOL) algorithm prioritizes the update of the parameters with high relevance score, allowing the learning to focus more on the highly relevant ones. Thus, these two components lead to an analyzable framework, where each sequential hidden state/basis represents the learned key poses/robot configuration. Compared to state-of-the-art methods, the SME-AGOL requires 40% fewer samples and receives 150% higher final reward/locomotion performance on a simulated hexapod robot, while taking merely 10 minutes of learning time from scratch on a physical hexapod robot. Taken together, this work not only proposes the SME-AGOL for sample efficient and understandable locomotion learning but also emphasizes the potential exploitation of interpretability for improving sample efficiency and learning performance.

Problem

• Reinforcement Learning
• Robot Locomotion
• Learning Efficiency

Innovation

• SME-AGOL
• Visualization of Learning Process
• Efficiency Enhancement in Robot Learning