OhhO View
Four cameras, one smart view. Fuse base-mounted cameras into a single calibrated bird's-eye-view your robot and your operators can both rely on — CPU-only, ROS 2-ready.
Four cameras. One smart view.
A robot that can only see straight ahead is a robot that bumps into things. OhhO View stitches up to four base-mounted cameras into one top-down, surround bird's-eye-view (BEV) of everything around the robot.
View computes geometric inverse-perspective-mapping from your robot's camera poses, producing a true fused top-down image — not a 2×2 tile. A calibration UI dials it in, and it runs on the CPU, so it doesn't compete with your GPU for VLA inference.
The BEV feeds both humans and models: operators get instant situational awareness, and policies like SmolVLA consume the same surround view as an input — closing the loop between perception and action.
- True fused BEV. Geometric inverse-perspective-mapping from URDF camera poses produces a single top-down image, not a tiled mosaic.
- Calibration UI. A bring-up wizard aligns the cameras; a calibration file takes priority when you want pixel-perfect.
- CPU-only. Runs on the robot's compute without touching the GPU your models need.
- Model-ready output. Publishes a standard ROS 2 image topic consumed by OhhO Serve policies and recorders alike.
- Any camera layout. Configure front / rear / left / right poses, field of view and mounting height for your specific robot.
Related: OhhO Build, OhhO Serve, OhhO Pilot