Object detection and tracking for autonomous navigation in dynamic environments
We address the problem of vision-based nav- igation in busy inner-city locations, using a stereo rig mounted on a mobile platform. In this scenario seman- tic information becomes important: rather than mod- elling moving objects as arbitrary obstacles, they should be categorised and tracked in order to predict their fu- ture behaviour. To this end, we combine classical ge- ometric world mapping with object category detection and tracking. Object-category specific detectors serve to find instances of the most important object classes (in our case pedestrians and cars). Based on these detec- tions, multi-object tracking recovers the objects' trajec- tories, thereby making it possible to predict their future locations, and to employ dynamic path planning. The approach is evaluated on challenging, realistic video se- quences recorded at busy inner-city locations.