Autonomous Vehicle Trajectory Control Using Iterative Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for controlling autonomous vehicles to repeatedly follow a predetermined trajectory, such as in parking or racing scenarios, face challenges in optimizing tracking performance and reducing tracking errors.
Innovation Solution
The method employs Iterative Learning Control (ILC) to generate an altered control signal based on the actual and target trajectory signals, using an iterative learning control device that adjusts the control model to minimize tracking errors and optimize performance over subsequent iterations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional control methods (PID, state-space control, feedback control, MPC) are used to control an autonomous vehicle to follow a desired trajectory, then the vehicle can be steered along a trajectory, but the tracking performance cannot be optimized across repeated iterations and tracking errors persist
Solution Approach 1:
The system performs preliminary action by using the actual trajectory data from previous iterations to pre-calculate adjustments to the control signal before the next iteration begins. The iterative learning controller processes the recorded actual trajectory signal and generates an altered control signal in advance, so that when the next iteration starts, the vehicle already has an optimized control strategy that anticipates and corrects for systematic tracking errors, rather than reacting to them in real-time.
Solution Approach 2:
The system implements feedback by continuously comparing the actual trajectory followed by the vehicle with the target trajectory, recording the deviation, and using this information through the iterative learning controller to generate corrected control signals. The feedback loop operates across iterations: the actual trajectory signal from iteration n feeds into the iterative learning controller, which produces an altered control signal for iteration n+1, systematically reducing tracking errors over time.
2Measurement precision
If the control signal is adjusted based on actual trajectory feedback during each iteration, then tracking performance improves over iterations, but the complexity of the control system increases
Solution Approach 1:
The system applies copying by recording the actual trajectory signal as a data copy of the vehicle's actual path, and using this copied information to generate the altered control signal without modifying the physical vehicle or control hardware. The iterative learning controller works with signal copies (trajectory data) rather than requiring complex real-time sensor arrays or multiple redundant control systems, reducing hardware complexity while maintaining improved tracking accuracy through iterative refinement.
Data Source
Figure 1~2
Figure 3
AI summary
A method for controlling an autonomous vehicle to repeatedly follow a same predetermined trajectory comprises: a) receiving a target trajectory signal indicative of the predetermined trajectory; b) generating a control signal adapted to steer the vehicle along the predetermined trajectory; and, for at least one of a number of iterations: c) steering the vehicle along the predetermined trajectory by feeding the control signal to the vehicle; d) measuring an actual trajectory followed by the vehicle in response to being steered according to the control signal; e) recording an actual trajectory signal indicative of the measured actual trajectory; f) using an iterative learning controller to determine an altered control signal using the control signal, the actual trajectory signal and the target trajectory signal. The method allows to improve a tracking performance during a subsequent iteration.