Combing multiple sensory inputs is a hard problem in itself. Say Tesla uses LIDAR with vision...they can have a great vision experience, a great LIDAR experience, but fuses those two experiences together can create lots of problems that wind up being worse than either by itself.