Rotational odometry calibration for differential robot platforms

Abstract

In order to navigate a robotic base in an unfamiliar environment, some mechanisms to detect positioning and location must be provided. When the base is in the process of navigation and makes use of these mechanisms, errors originating from the environment and base are inserted into the system, resulting in an erroneous position estimation. One relevant approach to reduce the error amplitude is to attack the error at its grounds, through an efficient calibration model, capable of identifying and estimating values for the main sources of uncertainty in odometry calculations. This work presents a novel calibration method, which differs from the classical ones by the way the calibration is performed, based on rotational movement instead of the usual linear displacement, reducing the area covered by the robotic base during the calibration process. The calibration results obtained are equivalent to those from the main classical methods.