Learning Social Robot Navigation By Sensing Human Legs

2026-07-30

AI Summary

This work addresses the limited social compliance of existing 2D LiDAR–based robot navigation methods, which typically model pedestrians as static geometric obstacles and neglect the dynamic leg motion actually captured by LiDAR. To overcome this, we propose CALF, an end-to-end neural architecture that integrates convolutional layers, attention mechanisms, and multilayer perceptrons to directly interpret human leg dynamics from raw LiDAR scans and generate socially compliant navigation commands. We introduce, for the first time, human legs as the primary perceptual target, developing a dedicated gait model and a lightweight JAX-based ray-traced simulator, LegNav, enabling efficient policy training via deep reinforcement learning. Our approach outperforms both classical and learning-based baselines in navigation performance and social compliance, trains within one hour on a single consumer-grade GPU, and achieves zero-shot real-world deployment on a TurtleBot 4, producing smooth and socially appropriate trajectories.

Abstract

Robots navigating among pedestrians typically sense their surroundings with a 2D LiDAR mounted close to the ground. At that height, the sensor mostly sees moving legs rather than whole people, yet most learning-based navigation methods still treat pedestrians as simple shapes like circles. This paper addresses that gap with CALF (Convolutional Attention for Leg Features), an end-to-end neural architecture that combines convolutional layers, attention, and MLP to interpret leg motion directly from LiDAR scans and produce safe navigation commands. The CALF policy is trained using deep reinforcement learning algorithms within LegNav, a custom lightweight 2D simulator that combines 2D LiDAR ray tracing with a novel pedestrian gait model. The resulting policy is compared against classical and learning-based baselines in terms of navigation performance and social compliance. The approach is validated through real-world experiments via zero-shot deployment on a TurtleBot 4, yielding smooth and socially compliant trajectories. Written in JAX, the LegNav simulator enables the training of a deployment-ready CALF policy in under an hour on a single consumer GPU.

Problem

• social robot navigation
• leg sensing
• 2D LiDAR
• pedestrian modeling
• human-aware navigation

Innovation

• CALF
• leg-based navigation
• social robot navigation
• deep reinforcement learning
• LiDAR perception