aiqbench analysis
NavVerse introduces a critical physics-enabled benchmark for testing embodied agents across continuous indoor-to-outdoor environments, addressing the significant adaptation gap currently hindering real-world robot deployment. The study reveals that while end-to-end Vision-Language Agents achieve higher zero-shot success rates in cross-context navigation, modular approaches offer superior safety profiles and highlight kinodynamic failures as a major bottleneck. This work advances robotics research by providing executable interfaces to evaluate task-success, path-efficiency, and safety metrics under realistic transition conditions.
NavVerse: Benchmarking Indoor-to-Outdoor Embodied Navigation in Continuous Robot Simulation
arXiv:2607.19695v1 Announce Type: new Abstract: Robots deployed in delivery, campus, and emergency-response settings often need to navigate from buildings to streets within a single continuous episode. Existing benchmarks usually evaluate indoor and outdoor navigation separately, and many abstract away robot execution, leaving exit finding, boundary traversal, adaptation, and kinodynamic failures underexplored. We introduce NavVerse, a physics-enabled benchmark for indoor-to-outdoor embodied navigation. NavVerse contains 100 indoor scenes, 50 urban outdoor scenes, and 50 indoor-to-outdoor scenes, and 10,000 episodes spanning Object Navigation, Vision-and-Language Navigation, and Place Navigation tasks, where agents search for semantic points of interest such as restaurants or banks. Agents are evaluated through executable robot interfaces using task-success, path-efficiency, and safety metrics. Zero-shot experiments with RL, VLA, and modular baselines show that current agents remain far from solving cross-context navigation: end-to-end VLAs obtain the highest zero-shot success, while the modular method provides the strongest safety profile. PlaceNav further reveals a clear drop from outdoor to indoor-to-outdoor scenes, indicating that adaptation remains major bottleneck.
Abstract:Robots deployed in delivery, campus, and emergency-response settings often need to navigate from buildings to streets within a single continuous episode. Existing benchmarks usually evaluate indoor and outdoor navigation separately, and many abstract away robot execution, leaving exit finding, boundary traversal, adaptation, and kinodynamic failures underexplored. We introduce NavVerse, a physics-enabled benchmark for indoor-to-outdoor embodied navigation. NavVerse contains 100 indoor scenes, 50 urban outdoor scenes, and 50 indoor-to-outdoor scenes, and 10,000 episodes spanning Object Navigation, Vision-and-Language Navigation, and Place Navigation tasks, where agents search for semantic points of interest such as restaurants or banks. Agents are evaluated through executable robot interfaces using task-success, path-efficiency, and safety metrics. Zero-shot experiments with RL, VLA, and modular baselines show that current agents remain far from solving cross-context navigation: end-to-end VLAs obtain the highest zero-shot success, while the modular method provides the strongest safety profile. PlaceNav further reveals a clear drop from outdoor to indoor-to-outdoor scenes, indicating that adaptation remains major bottleneck.
Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
arXiv:2607.19695 [cs.RO]
(or arXiv:2607.19695v1 [cs.RO] for this version)
https://doi.org/10.48550/arXiv.2607.19695 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 257 words
Related coverage
I2O2I-Nav: A Benchmark with a Modular Agent for Long-Horizon Indoor-to-Outdoor-to-Indoor Navigation
arXiv · Oct 14, 2026
Embodied-BenchClaw: An Autonomous Multi-Agent System for Embodied Spatial Intelligence Benchmark Construction
arXiv · Jun 11, 2026
ABot-AgentOS: A General Robotic Agent OS with Lifelong Multi-modal Memory
arXiv · Jul 20, 2026
Benchmarking Interaction-Oriented Aerial Mobility across Heterogeneous Environments
arXiv · Oct 14, 2026