HA-VLN 2.0: An Open Benchmark and Leaderboard for Human-Aware Navigation in Discrete and Continuous Environments with Dynamic Multi-Human Interactions
Researchers have developed a new benchmark for navigation in environments with multiple humans. The HA-VLN 2.0 benchmark includes a standardized task and metrics that capture both goal accuracy and personal-space adherence. It also provides a dataset and simulators modeling multi-human interactions, outdoor contexts, and finer language-motion alignment. The results show that explicit social modeling improves navigation robustness and reduces collisions.
Researchers have developed a new benchmark for navigation in environments with multiple humans. The HA-VLN 2.0 benchmark includes a standardized task and metrics that capture both goal accuracy and personal-space adherence. It also provides a dataset and simulators modeling multi-human interactions, outdoor contexts, and finer language-motion alignment. The results show that explicit social modeling improves navigation robustness and reduces collisions.
---
Why it matters: This matters to engineers working on AI navigation because it highlights the importance of considering human dynamics and social awareness in navigation systems. By using this benchmark, researchers can develop more robust and safe navigation algorithms that can handle complex environments with multiple humans.
Source: https://arxiv.org/abs/2503.14229
This article was originally published at: https://arxiv.org/abs/2503.14229