ABSTRACT: In this paper the methodology of convex optimization was applied to the problem of odometry calibration of differential drive mobile robots. In order to estimate the pose (i.e. position and orientation) of a mobile robot two calibration parameters are introduced to the kinematic model of a mobile robot. Optimal calibration parameters are the minimizers of the cost function penalizing the discrepancy between the exact and the estimated final position of the robot’ s trajectory. The conventional optimization methods are not applicable since the cost function – formulated as a nonconvex polynomial function – has several local minima. By employing the method of successive relaxations the problem of minimizing a polynomial function can be recasted as a sequence of semidefinite programs, for which efficient computational algorithms exist. Moreover, these computational algorithms also provide a proof of global optimality of the solution. The methodology can be is easily applied to different calibration models and expanded to the problem of online odometry calibration.

BibTeX entry:
@inproceedings \{Spudic2008_466,
author = \{Spudi\'{c}, V. AND Baoti\'{c}, M. AND Peri\'{c}, N.},
title = \{Odometry calibration of mobile robots via semidefinite programming}, booktitle = {Proceedings of the 17th International Electrotechnical and Computer Science Conference ERK 2008}, volume = \{B}, pages = \{187  190}, year = \{2008} }
