Data driven estimation of cabin dynamics in heavy-duty vehicles

Abstract: With increasing demand for autonomous systems and self-driving heavy-dutyvehicles there is an even more increasing demand for safety. In order to achievedesired safety level on the public roads, engineers have to tackle many technicalissues, like decision making, object detection and perception. In order to detect anobject or to have an understanding of its surroundings, autonomous heavy-dutyvehicles are equipped with different types of sensors. These sensors are placed ondifferent parts of the autonomous truck. The fact that some parts of the truckare highly dynamical introduces additional disturbances to the signals comingfrom onboard sensors. One of the most dynamic parts of every truck is its cabin.Moving cabin may induce additional disturbances into data coming from sensorsattached to it. This corrupted data may lead the autonomous trucks to make wrongdecisions. In the worst case, such decisions may be fatal.This thesis uses a data driven modeling approach for creating a mathematicaldescription of cabin movements based on data from onboard sensors. For thatpurpose, tools from system identification field are used. The resulting modelsare aimed to be used for implementation of real-time estimation algorithm forthe cabin dynamics, which in turn can be used for real-time compensation of thedisturbances.