A multi-paradigm object tracker for robot navigation assisted by external computer vision

Tracking multiple persons/robots/pets and moving objects is an essential task for situation awareness in robot navigation and operation. It is also a relatively complicated problem of computer vision and multiple solutions have been proposed in literature. In this paper we are exploring a novel method of object tracking using computer vision by fusing multiple techniques into a single tracker implementation. The main goal of this method is to perform high confidence data associations as soon as possible in order to be able to provide tracking information to a moving robot in real time with attempts to minimize the CPU utilization for tracking whenever possible since the Base Station computer is being shared with multiple software modules.