Don't Freeze, Don't Crash: Extending the Safe Operating Range of Neural Navigation in Dense Crowds

TL;DR

This paper introduces a reinforcement learning method for dense crowd navigation that generalizes to unseen crowd densities, maintaining safety and reducing freezing compared to existing approaches.

Contribution

It proposes a density-invariant observation encoding and density-randomized training for zero-shot density generalization in crowd navigation.

Findings

Achieves over 99% goal-reaching success in dense crowds.

Attains 86% collision-free success rate in unseen crowd densities.

Reduces freezing behavior compared to analytical methods.

Abstract

Navigating safely through dense crowds requires collision avoidance that generalizes beyond the densities seen during training. Learning-based crowd navigation can break under out-of-distribution crowd sizes due to density-sensitive observation normalization and social-cost scaling, while analytical solvers often remain safe but freeze in tight interactions. We propose a reinforcement learning approach for dense, variable-density navigation that attains zero-shot density generalization using a density-invariant observation encoding with density-randomized training and physics-informed proxemic reward shaping with density-adaptive scaling. The encoding represents the distance-sorted $K$ nearest pedestrians plus bounded crowd summaries, keeping input statistics stable as crowd size grows. Trained with $N\!\in\![11,16]$ pedestrians in a $3\mathrm{m}\times3\mathrm{m}$ arena and evaluated up to $N\!=\!21$ pedestrians ($1.3\times$ denser), our policy reaches the goal in $>99\%$ of episodes and achieves $86\%$ collision-free success in random crowds, with markedly less freezing than analytical methods and a $>\!60$-point collision-free margin over learning-based benchmark methods. Codes are available at \href{https://github.com/jznmsl/PSS-Social}{https://github.com/jznmsl/PSS-Social}.