youBot Vision Navigation Integration

The Webots world is already in place—KUKA youBot, colour-coded cubes, matching bins, camera stream, wheels, arm, and a starter Python controller that can see colours and drive in straight lines. What I still lack is dependable object localisation and navigation that can guide the robot from cube to bin while skirting obstacles, then hand the target pose to the manipulator for a clean pick-and-place cycle.

I would like the navigation layer to blend a small set of pre-planned waypoints with real-time obstacle avoidance for unforeseen clutter. Once the base is at the drop-off point, the arm and gripper should finish the job, ideally with smooth motion planning and minimal jitter; feel free to add task-planning optimisation or trajectory smoothing if you have a proven approach.

OpenCV, Webots Python API, and any lightweight motion-planning libraries you favour are all fair game as long as everything stays inside a single Python controller.

Deliverables
• A Python controller that:
– localises cubes and bins in camera space, converts them to world coordinates, and navigates with the hybrid strategy described
– triggers arm IK to pick a cube, lift, travel, and place it in the correctly coloured bin without collisions
• Clear setup notes and inline comments so I can iterate on your work.

Please centre your proposal on relevant Webots, robotics, or computer-vision experience; sample videos or repos that show similar navigation and manipulation tasks will help me judge fit quickly.