|
" Kinematic-based Trajectory Optimization and Tracking Framework for Autonomous Ground Vehicles "
Majd, Keyvan
Homaifar, Abdollah
| Document Type
|
:
|
Latin Dissertation
|
| Language of Document
|
:
|
English
|
| Record Number
|
:
|
1052561
|
| Doc. No
|
:
|
TL51678
|
| Main Entry
|
:
|
Majd, Keyvan
|
| Title & Author
|
:
|
Kinematic-based Trajectory Optimization and Tracking Framework for Autonomous Ground Vehicles\ Majd, KeyvanHomaifar, Abdollah
|
| College
|
:
|
North Carolina Agricultural and Technical State University
|
| Date
|
:
|
2019
|
| Degree
|
:
|
M.S.
|
| student score
|
:
|
2019
|
| Note
|
:
|
104 p.
|
| Abstract
|
:
|
Trajectory planning is seen as a key component for emerging autonomous vehicles as it provides a safe maneuver and dynamically feasible trajectory for the vehicle's control system. Depending on the application, the trajectory planning algorithms can be generally divided into two main categories of kinematic-based trajectory planning and dynamic-based trajectory planning. In this thesis, we revisit the car-like robot kinematic model trajectory tracking and control problem. We introduce a novel trajectory optimization and tracking framework to minimize the error of position, velocity, and acceleration. The consideration of velocity and acceleration errors allows the implementation of optimal control techniques in the trajectory planning problem since the control inputs explicitly appear in these terms. We explore the optimal solution methods for two cases of constrained and unconstrained vehicle motions. In the unconstrained case, we propose two numerical and analytical solutions. The numerical solution is obtained by solving a set of two-point boundary value nonlinear differential equations derived from the variational approach. Then, the input-output linearization technique is employed to transform the nonlinear problem into a linear formulation by defining an auxiliary input. By using the variational approach, a globally exponentially stable analytical solution is obtained, and the boundedness of internal dynamics are guaranteed. In the constrained case, we first propose a numerical solution using the variational approach to minimize the proposed optimization framework subject to the nonlinear kinematic model, states, and control constraints. Finally, a model predictive control (MPC) based optimization framework is developed in a discrete-time domain. The MPC-based framework solves the trajectory planning problem for a linearized time-variant kinematic model subject to the constraints on states, control inputs, and the slew rate of inputs.
|
| Descriptor
|
:
|
Electrical engineering
|
|
|
:
|
Robotics
|
| Added Entry
|
:
|
Homaifar, Abdollah
|
| Added Entry
|
:
|
North Carolina Agricultural and Technical State University
|
| |