Odometers Error Estimation of the Two-Wheeled Mobile Robot

This article considers the approach of using of a neural network to estimate the odometry error in the mobile robot localization based on the odometers. As the odometers were used inexpensive optical odometers with a resolution of 48 counts per wheel revolution. Data obtained from the odometer using for calculating the position of the robot by means of a mathematical model of its differential kinematic schemes. Odometers data are also used for training the neural network. For training the neural network was used the trajectory of the robot as the reference data, obtained through chart pattern recognition system is mounted on the robot