BiRoDiff: Diffusion policies for bipedal robot locomotion on unseen terrains

TL;DR

This paper presents BiRoDiff, a diffusion model-based control framework enabling bipedal robots to walk on unseen terrains, demonstrating superior generalization and real-time control capabilities through offline learning.

Contribution

Introduction of a lightweight diffusion model-based controller for bipedal robots that generalizes to unseen terrains and operates in real-time using offline data.

Findings

Successful simulation of walking on multiple terrains

High-frequency control step generation

Better scalability compared to online learning methods

Abstract

Locomotion on unknown terrains is essential for bipedal robots to handle novel real-world challenges, thus expanding their utility in disaster response and exploration. In this work, we introduce a lightweight framework that learns a single walking controller that yields locomotion on multiple terrains. We have designed a real-time robot controller based on diffusion models, which not only captures multiple behaviours with different velocities in a single policy but also generalizes well for unseen terrains. Our controller learns with offline data, which is better than online learning in aspects like scalability, simplicity in training scheme etc. We have designed and implemented a diffusion model-based policy controller in simulation on our custom-made Bipedal Robot model named Stoch BiRo. We have demonstrated its generalization capability and high frequency control step generation relative to typical generative models, which require huge onboarding compute.