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

TL;DR

This paper introduces SME-AGOL, an interpretable neural network framework with online learning for efficient robot locomotion, achieving high performance with fewer samples and rapid training on both simulated and real robots.

Contribution

The work presents a novel interpretable neural network architecture combined with an online learning algorithm, enhancing sample efficiency and interpretability in robot locomotion learning.

Findings

Requires 40% fewer samples than state-of-the-art methods.

Achieves 150% higher final reward in simulation.

Learns from scratch in 10 minutes on a physical robot.

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.