GROQLoco: Generalist and RObot-agnostic Quadruped Locomotion Control using Offline Datasets

2025-05-16

AI Summary

To address the challenge of general locomotion control for quadrupedal robots across diverse morphologies and terrains, this paper proposes a fully robot-agnostic offline imitation learning framework. Methodologically, it leverages proprioceptive-only sensory inputs and employs a Transformer-based attention architecture to jointly distill expert demonstrations from multiple robots (Go1 and Stoch 5) operating on flat ground and stairs—enabling cross-morphology behavioral distillation and unified modeling of both periodic and non-periodic gaits. Its key innovation lies in eliminating robot-specific encoding and online optimization, thereby supporting zero-shot transfer and real-time end-to-end deployment. Experimental results demonstrate that the learned policy generalizes directly—without fine-tuning—to significantly different hardware platforms and complex unseen terrains (e.g., outdoor environments and staircases). The system runs with low latency on an Intel i7 NUC, validating its practicality for embedded deployment.

Technology Category

• Intelligent Robots: Behavior Learning & Control
• Machine Learning: Imitation Learning & Inverse Reinforcement Learning
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Experiences and lessons learnt from Web-based algorithms and system deployments
• Search and Retrieval-Augmented AI: Agentic search
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Recent advancements in large-scale offline training have demonstrated the potential of generalist policy learning for complex robotic tasks. However, applying these principles to legged locomotion remains a challenge due to continuous dynamics and the need for real-time adaptation across diverse terrains and robot morphologies. In this work, we propose GROQLoco, a scalable, attention-based framework that learns a single generalist locomotion policy across multiple quadruped robots and terrains, relying solely on offline datasets. Our approach leverages expert demonstrations from two distinct locomotion behaviors - stair traversal (non-periodic gaits) and flat terrain traversal (periodic gaits) - collected across multiple quadruped robots, to train a generalist model that enables behavior fusion for both behaviors. Crucially, our framework operates directly on proprioceptive data from all robots without incorporating any robot-specific encodings. The policy is directly deployable on an Intel i7 nuc, producing low-latency control outputs without any test-time optimization. Our extensive experiments demonstrate strong zero-shot transfer across highly diverse quadruped robots and terrains, including hardware deployment on the Unitree Go1, a commercially available 12kg robot. Notably, we evaluate challenging cross-robot training setups where different locomotion skills are unevenly distributed across robots, yet observe successful transfer of both flat walking and stair traversal behaviors to all robots at test time. We also show preliminary walking on Stoch 5, a 70kg quadruped, on flat and outdoor terrains without requiring any fine tuning. These results highlight the potential for robust generalist locomotion across diverse robots and terrains.

Problem

• Learning generalist locomotion policy for multiple quadruped robots
• Adapting to diverse terrains without robot-specific encodings
• Achieving zero-shot transfer across robots and challenging environments

Innovation

• Attention-based framework for generalist locomotion policy
• Offline training with diverse quadruped robots and terrains
• Direct deployment without test-time optimization