Autonomous Driving Route Deviation Prediction and Compensation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face errors in control due to differences between planned and actual driving routes during maneuvers like lane changes, U-turns, and turns, necessitating a technology to predict and mitigate route deviations for stable and safe autonomous driving.
Innovation Solution
An autonomous driving control apparatus that uses sensors and high-definition maps to calculate expected driving routes, determine route deviations, and adjust driving strategies based on dynamics models and machine learning algorithms, considering risks of collision and vehicle interactions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If autonomous driving control follows a predetermined control-following route, then the driving operation is simplified and automated, but route deviation errors occur due to vehicle dynamics characteristics
Solution Approach 1:
The system performs preliminary calculation of the expected driving route based on vehicle dynamics models before actual driving occurs. By predicting the actual driving route considering vehicle dynamics characteristics (yaw rate, departure distance, departure angle, speed, acceleration) in advance, the system prepares compensation strategies beforehand to correct deviations from the control-following route, thereby maintaining both automation and reliability.
2Reliability
If the control system adjusts driving strategy in real-time to compensate for route deviations, then route following accuracy is improved, but system complexity increases
Solution Approach 1:
The system adjusts driving strategy by changing key vehicle parameters (yaw rate, speed, acceleration) based on the calculated expected driving route. Instead of complex overall route recalculation, the system modifies specific controllable parameters to steer the vehicle back toward the control-following route, thereby improving route following accuracy while maintaining relatively simple control logic.
Solution Approach 2:
The system continuously compares the actual driving route (calculated from vehicle dynamics) with the control-following route, detects deviations, and feeds this information back to adjust the driving strategy. This closed-loop feedback mechanism enables automatic compensation for route deviations without requiring complex manual intervention or overly complicated control architecture.
3Measurement precision
If the system calculates expected driving route using dynamics models and machine learning, then route deviation prediction accuracy is improved, but computational load and processing time increase
Solution Approach 1:
The system applies machine learning models selectively rather than continuously. The pre-trained machine learning-based learning model is used to predict route deviations in situations where high accuracy is critical, while relying more on physics-based dynamics models in other scenarios. This partial application of computationally intensive methods reduces overall processing time while maintaining prediction accuracy when most needed.
Data Source
AI summary
An autonomous driving control apparatus and method are for determining following-route deviation of an autonomous vehicle. The apparatus and method: may obtain surrounding information of an autonomous vehicle; may calculate a control-following route according to a predetermined driving strategy based on the surrounding information and the high definition map information around an autonomous vehicle; may calculate an expected driving route on which the autonomous vehicle is expected to be driven, when autonomous driving according to the control-following route is performed; may determine whether following-route deviation of the autonomous vehicle is expected, by comparing the control-following route with the expected driving route; and may change the driving strategy based on whether the following-route deviation of the autonomous vehicle is expected.