Skip to Main Content
In this paper, a system capable of obtaining the 3-D pose of a mobile robot using a ring of calibrated cameras attached to the environment is proposed. The system robustly tracks point fiducials in the image plane of the set of cameras generated by the robot's rigid shape in motion. Each fiducial is identified with a point belonging to a sparse 3-D geometrical model of the robot's structure. Such a model allows direct pose estimation from image measurements, and it can easily be enriched at each iteration with new points as the robot motion evolves. The process is divided into an initialization step, where the structure of the robot is obtained, and an online step, which is solved using sequential Bayesian inference. The approach allows the proper modeling of uncertainty in measurements and estimations, and at the same time, it serves as a regularization step in pose estimation. The proposed system is verified using simulated and real data.