FAST MOBILE ROBOT OBSTACLE DETECTION USING SIMULTANEOUS FIRING OF SONAR RING SENSORS

This paper describes a new method of scheduling multiple Ultrasonic Ra nge Sensors (URSs) mounted on a sonar ring designed for fast mobile ro bot obstacle detection. This method employs simultaneous firing of all on-board URSs operating on the same frequency where false range data caused by interference among multiple URSs are consequently rejected f rom scan data on the basis of crosstalk modeling. This task is perform ed by simulating a simple backpropagation neural network. The new meth od allows full panorama scanning six to ten times faster than those cu stomary in conventional applications so that mobile robot can detect o bstacles which appear in its path suddenly, and which may not be detec ted by the low-frequency scans in time for the robot to safely avoid t hem. The presented experimental results show the fundamental functiona lity of the new method.