Print Email Facebook Twitter An optimization based approach to autonomous drifting Title An optimization based approach to autonomous drifting: A scaled implementation feasibility study Author Verlaan, Bram (TU Delft Mechanical, Maritime and Materials Engineering; TU Delft Delft Center for Systems and Control) Contributor Keviczky, Tamas (mentor) Degree granting institution Delft University of Technology Programme Mechanical Engineering | Systems and Control Date 2019-04-30 Abstract Development of the autonomous vehicle has been a trending topic over the last few years. The automotive industry is continuously developing Advanced Driver-Assistance Systems (ADAS) that partially take over the driver’s workload. This has resulted in an increase in vehicle safety and a decrease in fatal crashes [1]. Full vehicle autonomy has not yet been reached, as the control systems involved are not yet capable of handling every situation. One of these critical situations is when a vehicle enters the unstable motion of drifting. A vehicle is prone to drifting on low-friction surfaces, and also during these generally unstable maneuvers, the autonomous system should be able to remain in control. The performance of an autonomous drifting controller should be exemplified by the experience of rally drivers in how to handle a vehicle and keep control of a vehicle while drifting. The objective of this thesis is to design a control system which is capable of handling a vehicle during a drifting motion and to follow a certain desired path. Vehicle dynamics are modeled as a three-state bicycle model to simplify the complex dynamics of the vehicle and the interaction between tyre and road. The definition of longitudinal wheel slip is reformulated to a smooth alternative to accommodate gradient based solving. With the system dynamics defined, the drifting motion is analyzed and equilibrium points are identified, showing differences between low- and high friction surfaces. Initially, a Model Predictive Control (MPC) strategy is applied with the purpose of steering the vehicle to desired drifting equilibria. Hereafter, the control system is extended to provide path following properties and addition of a dynamic velocity controller allows for a larger range of equilibria to be reached. The simulation setup intends to capture the experimental environment in the Network Embedded Robotics DCSC lab (NERDlab) at the Delft Center for Systems and Control (DCSC) department. Simulating a 1:10 scaled model allows to investigate the challenges that arise when implementing the control strategy on a scaled vehicle. These simulations show that autonomous drift control using the designed MPC strategy is possible, even when accounting for possible uncertainties such as delay, noise, and model mismatch. Subject optimizationcontrolautonomousdriftingvehicle To reference this document use: http://resolver.tudelft.nl/uuid:6d0e608e-b4d6-4d7f-8f6e-1ffed2802347 Part of collection Student theses Document type master thesis Rights © 2019 Bram Verlaan Files PDF Thesis_Final.pdf 24.88 MB Close viewer /islandora/object/uuid:6d0e608e-b4d6-4d7f-8f6e-1ffed2802347/datastream/OBJ/view