This video shows active object recognition using a Kinect RGBD and an RGB eye-in-hand camera mounted on the manipulator. Three objects on a table (different flavours of the same juice) are recognized. The recognition probability of the initial recognition is not reliable enough as the objects are too similar. The robot then uses the camera on the manipulator to systematically inspect the objects by reading barcode and text on the objects in question. All objects are recognized reliably to a very high probability.
Link to this comment:
All Comments (0)