Teaching robots where things are without cameras

Robots need to know exactly where objects are to manipulate them precisely, but optical tracking systems require tedious calibration and fail when objects block the camera. This work replaces optics with physics: the robot's probe itself becomes the sensor. By matching the object's shape against the swept volume the probe traces through space—using both actual contact and near-miss information—the system pinpoints object pose. A global search via 3D Fourier transform finds rough position, then continuous refinement tightens it to surgical accuracy. On a tooth-prep robot, it achieved 0.42 mm positioning without any external sensors.