Autonomous vehicles and research pertaining to them have been an important topicin academia and industry in recent years. Developing controllers that enable vehiclesto performpath and trajectory following is a diverse topic where many differentcontrol strategies are available. In this thesis, we focus on lateral and longitudinalcontrol of autonomous vehicles and two different control strategies are considered:a standard decoupled control and a new suggested coupled control.In the decoupled control, the lateral controller consists of a linear time-varying modelpredictive controller (LTV-MPC) together with a PI-controller for the longitudinalcontrol. The coupled controller is a more complex LTV-MPC which handles bothlateral and longitudinal control. The objective is to develop both control strategiesand evaluate their design and performance through path following simulations in aMATLAB environment.When designing the LTV-MPC, two vehicle models are considered: a kinematic modelwithout tyre dynamics and a dynamic bicycle model with tyre forces derived froma linear Pacejka model. A research on how model complexity affects tracking performanceand solver times is also performed. In the end, the thesis presents thefindings of the different control strategies and evaluate them in terms of trackingperformance, solver time, and ease of implementation.
Model Complexity and Coupling of Longitudinal and Lateral Control in Autonomous Vehicles Using Model Predictive Control
2015-01-01
Theses
Electronic Resource
English
DDC: | 629 |
Model Predictive Control of Autonomous Vehicles
Tema Archive | 2014
|LONGITUDINAL AND LATERAL CONTROL FOR AUTONOMOUS GROUND VEHICLES
British Library Conference Proceedings | 2014
|