NavOL: Navigation Policy with Online Imitation Learning

TL;DR

NavOL introduces an online imitation learning framework for robotic navigation that interacts with simulators, leveraging expert demonstrations and a pretrained diffusion policy to improve efficiency and robustness without reward engineering.

Contribution

The paper presents NavOL, a novel online imitation learning paradigm that trains navigation policies through interactive rollouts and expert guidance, reducing distribution shift and eliminating reward design.

Findings

NavOL achieves consistent performance improvements in simulation benchmarks.

It scales efficiently across multiple scenes with high throughput.

Real-world experiments validate its effectiveness in practical navigation tasks.

Abstract

Learning robust navigation policies remains a core challenge in robotics. Offline imitation learning suffers from distribution shift and compounding errors at rollout, while reinforcement learning requires reward engineering and learns inefficiently. In this paper, we propose NavOL, an online imitation learning paradigm that interacts with a simulator and updates itself using expert demonstrations gathered online. Built upon a pretrained navigation diffusion policy that maps local observations to future waypoints, NavOL trains in a rollout update loop: during rollout, the policy acts in the simulator and queries a global planner which has privileged access to the global environment for the optimal path segment as ground truth trajectory labels; during update, the policy is trained on the online collected observation trajectory pairs. This online imitation loop removes the need for reward design, improves learning efficiency, and mitigates distribution shift by training on the policy own explored rollouts. Built on IsaacLab with fast, high-fidelity parallel rendering and domain randomization of camera pose and start-goal pairs, our system scales across 50 scenes on 8 RTX 4090 GPUs, collecting over 2,000 new trajectories per hour, each averaging more than 400 steps. We also introduce an indoor visual navigation benchmark with predefined start and goal positions for zero-shot generalization. Extensive evaluations on simulation benchmarks, including the NavDP benchmark and our proposed benchmark, as well as carefully designed real-world experiments, demonstrate the effectiveness of NavOL, showing consistent performance gains in online imitation learning.