Neuromorphic Reinforcement Learning for Quadruped Locomotion Control on Uneven Terrain

TL;DR

This paper introduces an equilibrium-propagation-based PPO framework for quadruped locomotion that enables energy-efficient, adaptive control on uneven terrain by replacing traditional backpropagation with local learning.

Contribution

It develops a novel EP-compatible PPO method with a bio-inspired CPG policy and residual adjustments, facilitating stable, efficient on-robot adaptation for complex locomotion tasks.

Findings

Achieves stable policy convergence on uneven terrain in experiments.

Matches baseline performance in success rate, velocity, and stability.

Reduces GPU memory usage by 4.3 times compared to BPTT.

Abstract

Reinforcement learning (RL) has enabled robust quadruped locomotion over complex terrain, but most learned controllers are trained offline with backpropagation in massively parallel simulation and deployed as fixed policies, limiting adaptation to terrain variation, payload changes, actuator wear, and other real-world conditions under onboard power constraints. Local learning provides a potential path toward energy-aware on-robot adaptation by replacing global backpropagation graphs with updates driven by local neural states, making the learning rule more compatible with neuromorphic and in-memory computing substrates. This work proposes an equilibrium-propagation (EP)-based proximal policy optimization (PPO) framework for uneven-terrain quadruped locomotion. The controller combines a bio-inspired central pattern generator (CPG) policy with a residual postural adjustment policy, while replacing conventional backpropagation-trained policy and value networks with EP-enabled local learning. To train stochastic continuous-control policies with EP, we derive an EP-compatible PPO output-nudging signal and introduce a two-sided ratio clipping mechanism that stabilizes policy updates during relaxation. Experiments on a 12-DoF A1 quadruped show that the proposed controller achieves stable policy convergence in a two-stage uneven terrain locomotion task. Its locomotion performance is comparable to a backpropagation-trained PPO baseline in success rate, velocity tracking, actuator power, and body stability, while improving GPU memory efficiency by 4.3\(\times\) compared with backpropagation through time (BPTT). These results suggest that local equilibrium-based learning can support high-dimensional embodied locomotion and provide an algorithmic foundation for low-power on-robot adaptation and fine-tuning.