Development in the field of autonomous vehicles has been rising over the past decade. Most autonomous driving systems must be able to perform a wide variety of tasks, such as driving across an intersection, following another car, or negotiating complicated city streets.
Even though these seem to be complex problems, they can be further decomposed into three major fields of development. Detection, where the car first has to observe the different objects in its environment. Decision making, as the car has to have a high-level strategy based on the detected objects, deciding where it wants to go, and identifying obstacles to avoid. Finally, the actual control values, the needed vehicle speed and steering angle value have to be defined, so the car can actually perform according to the high-level decisions.
Current work presents possible control methods for trajectory tracking of self-driving cars. At first, a proper mechanical model, the well-known bicycle model is demonstrated. While this model represents the behavior of a vehicle within a linear range, the added effects of different tire models are also shown. The model is transformed into a state-space with distance and angular deviation from the trajectory as state variables. This form of the bicycle model can be used for controller design.
Four different control methods are presented, the Stanley controller, that won the DARPA Challenge in 2005, the first competition where driverless cars were competing against each other. A linear quadratic regulator (LQR), a model predictive controller (MPC) and a model predictive controller with input and state constraints. By testing the implemented controllers with the described model it is shown, that although the LQR method seems to be the one with the best dynamics, the constrained MPC can handle the proposed task more robust, as we can be sure that the calculated values by this method can be realized by an actuator.
Trajectory Tracking Control of an Autonomous Car
Lect.Notes Mechanical Engineering
Vehicle and Automotive Engineering ; 2020 ; Miskolc, Hungary November 25, 2020 - November 26, 2020
2020-10-20
12 pages
Article/Chapter (Book)
Electronic Resource
English
Trajectory Tracking Control of an Autonomous Car
TIBKAT | 2021
|LSTM-Based Trajectory Tracking Control for Autonomous Vehicles
British Library Conference Proceedings | 2022
|LSTM-Based Trajectory Tracking Control for Autonomous Vehicles
SAE Technical Papers | 2022
|