Correcting mismatch in a data fusion system for ultrasonic mapping with a mobile robot

Abstract

The goal of an ultrasonic mapping system is to produce an outline of the room from range data sensed by the robot as it moves through the room. An ultrasonic sensor measures the range to the nearest object from the time-of-flight to the first echo. By fusing sensing with robot motion, low-level features can be extracted from range data. Mismatch between measurements of motion with range sensing and with odometry result in errors in the map. Before these features can be reliably fused into outlines and object descriptions the errors caused by mismatch have to be corrected.