Event
OWLS: Zero-Shot Rover Navigation with VLM Feature Maps
1 reports · 1 sources
Chahyon Ku and co-authors submitted a preprint proposing OWLS (Open-World Language-Conditioned Speed Field): given a description of traversable terrain, a vision-language model scores each pixel of a single camera image; the scores are projected onto the ground and adapted to the platform via two morphological operations, then used as the speed field of a minimum-time planner, so that terrain appearance and obstacle spacing jointly shape a single trajectory. The authors claim the method requires no task-specific training, terrain annotations, or persistent map, and no external perception sensors beyond a camera. The authors evaluated it on physical rovers indoors and off-road, on Mars and lunar analog rovers over 200 procedurally generated terrains, and on real planetary imagery.
Report timeline
- OWLS: Open-world language-conditioned speed fields for zero-shot rover navigation
The authors propose OWLS (Open-world Language-conditioned Speed Fields), which plans directly over the dense feature maps of a vision-language model (VLM): given a description of traversable terrain, the VLM scores every pixel of a single camera image, and OWLS projects these scores onto the ground, adapts them to the platform with two morphological operations, and uses them as the speed field of a minimum-time planner, so that terrain appearance and obstacle spacing jointly shape a single trajectory.
arXiv cs.RO ·