Revisiting Reward Design and Evaluation for Robust Humanoid Standing and Walking

TL;DR

This paper introduces a systematic benchmarking framework for evaluating humanoid robot standing and walking controllers, focusing on reward function design and real-world performance metrics to facilitate fair comparisons and improvements.

Contribution

It presents a low-cost, quantitative benchmarking method and revisits reward design to improve humanoid locomotion controllers trained with reinforcement learning.

Findings

Benchmarking framework effectively identifies performance gaps.

Minimally constraining reward functions enable robust control.

Comparison shows trade-offs among different controllers.

Abstract

A necessary capability for humanoid robots is the ability to stand and walk while rejecting natural disturbances. Recent progress has been made using sim-to-real reinforcement learning (RL) to train such locomotion controllers, with approaches differing mainly in their reward functions. However, prior works lack a clear method to systematically test new reward functions and compare controller performance through repeatable experiments. This limits our understanding of the trade-offs between approaches and hinders progress. To address this, we propose a low-cost, quantitative benchmarking method to evaluate and compare the real-world performance of standing and walking (SaW) controllers on metrics like command following, disturbance recovery, and energy efficiency. We also revisit reward function design and construct a minimally constraining reward function to train SaW controllers. We experimentally verify that our benchmarking framework can identify areas for improvement, which can be systematically addressed to enhance the policies. We also compare our new controller to state-of-the-art controllers on the Digit humanoid robot. The results provide clear quantitative trade-offs among the controllers and suggest directions for future improvements to the reward functions and expansion of the benchmarks.