Learning Human Perception Dynamics for Informative Robot Communication

• DOI: 10.48550/arxiv.2502.01857
• Publication Date: 2025-02-03
• Authors: Shenghui Chen; Ruihan Zhao; Sandeep Chinchali; Ufuk Topcu
• Keywords: Dynamics; Human–robot interaction

Human-robot cooperative navigation is challenging in environments with incomplete information. We introduce CoNav-Maze, a simulated robotics environment where robot navigates using local perception while human operator provides guidance based on an inaccurate map. The can share its camera views to improve the operator's understanding of environment. To enable efficient human-robot cooperation, we propose Information Gain Monte Carlo Tree Search (IG-MCTS), online planning algorithm that balances autonomous movement and informative communication. Central IG-MCTS neural dynamics model estimates how humans distill information from communications. collect dataset through crowdsourced mapping task CoNav-Maze train this fully convolutional architecture data augmentation. User studies show outperforms teleoperation instruction-following baselines, achieving comparable performance significantly less communication lower cognitive load, as evidenced by eye-tracking metrics.