Linear Estimation of the Odometric Parameters for Differential-Drive Mobile Robots

In this paper a calibration technique aimed at identifying the odometric parameters of differential-drive mobile robots is proposed. The proposed algorithm is based on two successive least-squares estimations based on the continuous-time kinematic equations of motion; the time-discretization error, thus, is avoided. The use of the least-squares technique is made possible by observing a linear mapping between the unknowns and the measurements and it is not the results of a linearization. Another advantage of the proposed technique is that no predefined path is required, i.e., the robot can move under a preferred navigation algorithm. The basic technique makes use of video-camera measurements and absolute position readings of the wheels´ encoders, the use of different sensors and measurements of the wheels velocities is also discussed. Experimental results with the mobile robot Khepera II confirm the effectiveness of the proposed technique