WildOS: Open-Vocabulary Object Search in the Wild
WildOS: Open-Vocabulary Object Search in the Wild
This research presents WildOS, a unified system for autonomous navigation in complex outdoor environments that combines safe geometric exploration with semantic visual reasoning. The system utilizes a sparse navigation graph maintained by a foundation-model-based vision module (ExploRFM) to score frontier nodes for traversability and object similarity, enabling real-time onboard semantic navigation. Additionally, a particle-filter-based method is introduced for coarse localization of open-vocabulary target queries beyond the robot's immediate depth horizon. Extensive closed-loop field experiments across diverse off-road and urban terrains demonstrate that WildOS enables robust navigation, significantly outperforming purely geometric and purely vision-based baselines in both efficiency and autonomy.
The paper introduces WildOS, a system combining geometric exploration with semantic visual reasoning for long-range object search in unstructured environments, validated through closed-loop field experiments on off-road and urban terrains.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
- Control
- Reported autonomousOpen source S5
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
ate depth horizon, enabling effective planning toward distant goals. Extensive closed-loop field experiments across diverse off-road and urban terrains demonstrate that WildOS enables robust navigation, significantly outperforming purely geometric and purely vision-based baselines in both efficiency and autonomy. Our results highlight the potential of vision foundation models to drive open-world robotic behaviors tha
Open source S5
t are both semantically informed and geometrically grounded. Project Page: https://leggedrobotics.github.io/wildos/
Open source S6
Source：arXiv Robotics — research abstracts · arxiv.org