Sensor fusion method of distance data and image data with data vagueness for objects detection

摘要

A sensor fusion algorithm considering possibility of obstacle existence for obstacle detection is proposed. In the sensor fusion of both image data and distance data observed by a laser range finder, vagueness of data, which come from the unknown disturbance, is paid a 16t of attention. Using co-relation factor between the distance data without any object and the observed distance data, possibility index of obstacle existence is calculated. Similarity of intensity distribution of image data in each horizontal direction compensates the possibility index of existence. Objects with possibility index of existence are detected. For the pedestrians and vehicles, experimental detections are carried out in real ground using laser range finder. The results conclude the algorithm gives the breakthrough to solve the dilemma problem of under-detection and over-detection.