Cue and sensor fusion for independent moving objects detection and description in driving scenes

Abstract

In this study we present an approach to detecting, describing and tracking independently moving objects (IMOs) in stereo video sequences acquired by on-board cameras on a moving vehicle. In the proposed model only three sensors are used: stereovision, speedometer and light detection and ranging (LIDAR). The IMOs detected by vision are matched with obstacles provided by LIDAR. In the case of a successful matching, the descriptions of the IMOs (distance, relative speed and acceleration) are provided by adaptive cruise control (ACC) LIDAR sensor, or otherwise these descriptions are estimated based on vision. Absolute speed of the IMO is evaluated using its relative velocity and ego-speed provided by the speedometer. Preliminary results indicate the generalization ability of the proposed system. © 2008 Springer US.