CityWalker: Learning Embodied Urban Navigation from Web-Scale Videos

TL;DR

CityWalker leverages web-scale urban videos to train embodied agents for complex navigation tasks, significantly improving performance in dynamic city environments without relying on costly annotations.

Contribution

The paper introduces a scalable data-driven approach using web videos for training urban navigation agents, enabling large-scale imitation learning without manual annotations.

Findings

Training on large-scale diverse datasets improves navigation performance.

The approach surpasses existing methods in urban navigation tasks.

Web-sourced videos effectively teach complex navigation behaviors.

Abstract

Navigating dynamic urban environments presents significant challenges for embodied agents, requiring advanced spatial reasoning and adherence to common-sense norms. Despite progress, existing visual navigation methods struggle in map-free or off-street settings, limiting the deployment of autonomous agents like last-mile delivery robots. To overcome these obstacles, we propose a scalable, data-driven approach for human-like urban navigation by training agents on thousands of hours of in-the-wild city walking and driving videos sourced from the web. We introduce a simple and scalable data processing pipeline that extracts action supervision from these videos, enabling large-scale imitation learning without costly annotations. Our model learns sophisticated navigation policies to handle diverse challenges and critical scenarios. Experimental results show that training on large-scale, diverse datasets significantly enhances navigation performance, surpassing current methods. This work shows the potential of using abundant online video data to develop robust navigation policies for embodied agents in dynamic urban settings. Project homepage is at https://ai4ce.github.io/CityWalker/.