Indoor Robot Odometry Modelling and Calibration Incorporating Floor Surfaces and Acceleration

Mobile robot odometry estimates a robot’s position and orientation from measuring wheel angle changes. Calibration of odometry involves estimating wheel radii and wheel separation. This thesis devises novel odometry calibration methods that detect different floor types in real time, and also models the effects of linear acceleration and path curvature to significantly improve robot localisation and map accuracy. This involves novel floor classification with colour and motor current sensors using a Support Vector Machine algorithm. More accurate odometry models are proposed using data driven methods and theoretic support from tyre mechanics.