Identification-based modelling of a wheeled robot

The aim of this thesis is to investigate the state of the art of dynamic modelling of vehicles including existing dynamic models, friction forces and Pacejka’s “Magic Formula” and reproduce a more realistic model to be used in a new simulator, collect a bunch of sensor data from the real system and design a learning-based algorithm able to cope with the collected data and able to fit a set of differential equations describing the real vehicle dynamics. Exploiting a grey box approach, the thesis starts from a kinematic model, and it finishes producing a dynamical model which involves forces, acceleration, slip angles and steering angles. All these steps are performed using many tools like Python, MATLAB, and ROS.