Abstract
Proximity sensing is effective to increase safety in human-robot collaboration. Capacitive proximity sensing can detect objects in proximity over a wide range. However, its detection accuracy suffers from output drift. To compensate for the drift, this study combines a capacitive sensor with an optical sensor that has a narrow viewing angle and estimates the target distance by sensor fusion. A switching unscented Kalman filter was implemented, whose filter setting was switched based on the availability of the optical sensor’s output. A human hand motion was measured by using the two sensors, and an artificial output drift was added on the capacitive sensor’s output. Then, the filter estimated the hand position. It was confirmed that the implemented filter could remove the output drift based on the optical sensor’s output and that it could realize uninterrupted estimation for the viewing angle wider than that of the optical sensor.
