376 0

Real-Coded Micro-Genetic Algorithm based Model Predictive Control in Collision Avoidance

Title
Real-Coded Micro-Genetic Algorithm based Model Predictive Control in Collision Avoidance
Author
손로만
Advisor(s)
허건수
Issue Date
2017-08
Publisher
한양대학교
Degree
Master
Abstract
Nowadays, with the growth of computational power it became possible to apply advanced control and optimization algorithms, such as Model Predictive Control (MPC) and Genetic Algorithms, in real-time problems. In this thesis, an MPC augmented with Artificial Potential Field (APF) approach is applied to design a Collision Avoidance (CA) system for an autonomous vehicle. It is assumed that the vehicle environment is unknown with different types of obstacles. An MPC controller is designed to keep the vehicle as close as possible to the reference trajectory. Whereas an APF behaves as an external force, which pushes the vehicle away from the reference trajectory, which is occupied by obstacles. A single-track vehicle dynamics model is defined and used to predict future states of the vehicle. Different types of Potential Functions (PFs) for various obstacles are defined to generate the vehicle environment. The CA problem is formulated as a constrained objective function minimization. To minimize the objective function a fast and effective solver is required. To match these requirements a Real-Coded Micro-Genetic Algorithm (RμGA) is adopted in this thesis. To verify the performance of the proposed CA controller a number of simulations have been carried out using commercial software, MATLAB/Simulink and CarSim. Simulation results show that the proposed CA controller effectively performs the assigned CA task. As compared to other solvers, using RμGA allows to formulate an MPC optimization problem in a more flexible manner, satisfying the requirements.
URI
http://hdl.handle.net/20.500.11754/33347http://hanyang.dcollection.net/common/orgView/200000431044
Appears in Collections:
GRADUATE SCHOOL[S](대학원) > AUTOMOTIVE ENGINEERING(미래자동차공학과) > Theses (Ph.D.)
Files in This Item:
There are no files associated with this item.
Export
RIS (EndNote)
XLS (Excel)
XML


qrcode

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.

BROWSE