Approaches to Pole Detection using Ranged Laser Data

In building up maps for mobile robot naviga- tion the detection of clearly identifiable land- marks is important. Also important is the con- fidence level that recognised landmarks are re- ally what we identify them as. This paper looks at using laser range sensing to both iden- tify poles and to attach a probability to poles observed. Adapting the standard method of circle detection using Hough Tables a new ap- proach to assigning probabilities to features is proposed. These probabilities are assigned ac- cording to the level of support in the Hough Table and according to aspects germane to the objects itself. Results are given demonstrating the real time application of this approach and its usefulness in mapping landmarks.