Communicating Vehicles in Convoy and Monocular Vision-based Localization

This article deals with mobile robotics in convoy. An autonomous vehicle has to follow the path of the vehicle ahead of it, but without having the vehicle ahead in its field of view. So, the task for the second vehicle is to follow the leader's path with a variable time delay. Moreover, relatively strong constraints are imposed in terms of speed of processing, precision of the localization and for communication. The vehicles of the convoy have to be localized in their environment without being detectable. So, the use of active sensors such as rangefinders, radars or laser sensors, is prohibited in the same way as GPS devices. The communication data rate between consecutive vehicles is low (10 kbits/s). The presented method allows in this context the replay of a path from input data based on 3D positions of some points supplied by the leader vehicle. This method combines data acquired by low cost sensors (odometer, gyroscope) and by a monocular camera through an Extended Kalman Filter (EKF). Simulation results as well as results of experiments conducted on real data demonstrate the effectiveness of the proposed approach.