Learning Robot Visual Navigation in Crowds via Intention-Aware Scene Representations

2026-06-24

AI Summary

This work addresses the limitations of existing vision-based navigation methods in crowded environments, which often rely on oversimplified scene representations and struggle to effectively integrate human intent with environmental structure. To overcome this, the authors propose iCrowdNav, a novel approach that uniquely combines human pose cues with visual scene information to construct an intent-aware representation. Leveraging egocentric visual input, iCrowdNav employs a spatiotemporal encoder to extract scene occupancy features and introduces an Intent-Interact Former module to infer pedestrian motion intentions, producing compact state embeddings for training deep reinforcement learning policies. Experimental results demonstrate that iCrowdNav significantly outperforms current baselines in both simulated and real-world environments, achieving efficient crowd navigation using only visual observations.

Abstract

Robot crowd navigation requires the ability to infer human intentions while accounting for the structural constraints of the environment. Currently, deep reinforcement learning (DRL) provides a promising method for learning navigation policies that understand human intentions. However, most of them rely on limited scene representations, treating pedestrians as simple 2D points and ignoring rich visual cues from both humans and the environment. To address this issue, we introduce iCrowdNav, a novel visual crowd navigation method with intention-aware scene representations, to encode behavioral and structural context from egocentric visual observations. Our method employs two key components: a spatio-temporal encoder for extracting occupancy features of the scene, and Intent-Interact Former (I$^2$ Former), an attention-based module that encodes human poses to infer pedestrians' motion intentions. These features are integrated into a compact state embedding that supports effective DRL policy training. Extensive experiments show that our method achieves superior performance over baselines, and real-world deployment demonstrates vision-based crowd navigation.

Problem

• robot navigation
• crowd navigation
• human intention
• scene representation
• visual cues

Innovation

• intention-aware scene representation
• visual crowd navigation
• spatio-temporal encoder
• Intent-Interact Former
• egocentric vision