Learning from Human Gaze: Human-like Robot Social Navigation in Dense Crowds
Proceedings of the AAAI Conference on Artificial Intelligence · 14.03.2026 · 40 (22) · 18728-18736
DOI: 10.1609/aaai.v40i22.38941 ↗Kurz gefasst
Fachartikel in der Zeitschrift Proceedings of the AAAI Conference on Artificial Intelligence, online erschienen am 14.03.2026. Verfasst von Zhecheng Yu und 9 weiteren Autor:innen.
Abstract (Originalsprache)
Robot navigation in dense crowds requires understanding social cues that humans naturally use, yet existing methods struggle with real-world complexity. We investigate two questions: (1) Where do pedestrians look when navigating crowds? and (2) Can eye tracking improve robot navigation? To answer, we introduce GazeNav, an egocentric dataset collected via wearable eye trackers, featuring synchronized video, gaze, and trajectories in crowded environments. Analysis reveals that the gaze of pedestrians is closely related to the semantic presence and movement of other individuals, exhibiting distinct attention patterns across navigation behaviors. Building on this, we propose Gaze2Nav, a modular framework that first predicts human gaze to infer socially salient pedestrians, then incorporates the semantic attention into motion planning alongside visual inputs. Our method achieves 87.6% salient pedestrian prediction accuracy and reduces trajectory error by 15.4% over state-of-the-art baselines. By aligning with human gaze, our framework improves both performance and interpretability, advancing toward human-like, socially intelligent robot navigation.
Zitieren
APA
Yu, Z., Lyu, Y., Yang, C., Chen, T., Zhang, Y., Ling, B., Wang, P., Gao, G., Wu, W., & Lim, B. Y. (2026). Learning from Human Gaze: Human-like Robot Social Navigation in Dense Crowds. Proceedings of the AAAI Conference on Artificial Intelligence, 40(22), 18728-18736. https://doi.org/10.1609/aaai.v40i22.38941
Vancouver
Yu Z, Lyu Y, Yang C, Chen T, Zhang Y, Ling B, et al. Learning from Human Gaze: Human-like Robot Social Navigation in Dense Crowds. Proceedings of the AAAI Conference on Artificial Intelligence. 2026;40(22):18728-18736. doi:10.1609/aaai.v40i22.38941
BibTeX
@article{Yu2026,
author = {Yu, Zhecheng and Lyu, Yan and Yang, Chen and Chen, Tao and Zhang, Yishuang and Ling, Bo and Wang, Peng and Gao, Guanyu and Wu, Weiwei and Lim, Brian Y.},
title = {Learning from Human Gaze: Human-like Robot Social Navigation in Dense Crowds},
journal = {Proceedings of the AAAI Conference on Artificial Intelligence},
year = {2026},
volume = {40},
number = {22},
pages = {18728--18736},
doi = {10.1609/aaai.v40i22.38941}
}
EUVIMED