Extracting lines, circular segments and clusters from radar pictures in real time for an autonomous mobile robot

Autonomous mobile systems have to convert their sensor data in real time into meaningful data structures to navigate with their help through their environment. The authors show how to extract a description of the environment from a radar picture in real time with help of a curvature function. The mobile system MOBOT III has as its environment an office like indoor environment and obstacles detected by a laser radar are described as lines, circle segments and clusters as building blocks of maps. The line extracting algorithm is compared to other line fitting algorithms and the time needed measured from a number of different scenes. The algorithm allows to process laser radar data with 250 ms/picture.<<ETX>>