Modeling odometry and uncertainty propagation for a Bi-Steerable car

This paper addresses the problem of modeling the odometry (dead-reckoning) for a Bi-Steerable car (BS-car) and its uncertainty propagation model. The attractive feature of BS-robots is their enhanced maneuverability compared to car-like robots. At first, we take into account the steering ability of a BS-car in order to obtain the odometry model for vehicles showing this type of kinematics. Subsequently, we obtain a generic model for the uncertainty propagation of the odometry model by using a Bayesian probability approach. We show how it is possible to derive from this generic model a specific error propagation model for a particular system, under clearly stated hypothesis. We evaluate qualitatively and quantitatively the performance of the odometry model by experimenting with our BS-platform, the Cycab robot.