The intelligent obstacle sensing and recognizing method based on D–S evidence theory for UGV

Abstract

Unmanned Ground Vehicles (UGVs) have been widely used in Internet of Things, military applications and outer space exploration. Reliable intelligent obstacle recognition is the key requirement and significant challenge for UGVs. To perform robust obstacle recognition under all conditions, data processing and fusing will be necessary based on multiple sensors. This paper carries out obstacle sensing and recognizing research by extracting obstacle feature evidence from camera and laser scanner. The feature evidence extracted from camera includes S/V in HSV color space. The laser feature evidences are related to penetrability value feature and λ3∕λ2 feature based on covariance matrix. Then, the key step is to calculate the basic probability assignment based on these feature evidences. By applying Dempster’s combination rules the current obstacle is recognized based on D–S evidence theory. The test results show the validity of this new method and its application in cross-country environment perception, also give references to Internet of Vehicles including vehicle safety control and assistance driving technology, also the related algorithm can be potentially used in the Internet of Things or big data processing.