Steering Neutral Point Learning for Curved-Road Lane Keeping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing driver assistance systems face challenges in accurately learning the neutral point of the steering angle, particularly when transitioning from straight roads to curved roads, leading to deviations in steering control and reduced control performance during lane departure prevention and active lane keeping.
Innovation Solution
The system employs a neutral point learning calculator that estimates lateral accelerations based on both steering angle and lane curvature, using sensors and cameras to continuously learn and correct the neutral point, ensuring accurate steering angle control on various road conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If the neutral point is learned only on straight roads using existing methods, then the learning process is simple, but the steering control accuracy deteriorates when transitioning to curved roads
Solution Approach 1:
The patent makes the neutral point learning process dynamic by enabling it to be performed not only on straight roads but also on curved roads. The system adapts the learning conditions based on the actual road geometry detected by the recognition device, allowing the neutral point to be continuously updated under varying driving conditions, thereby maintaining steering control accuracy across different road types.
Solution Approach 2:
The patent changes the learning parameters by incorporating lane curvature information into the neutral point learning process. When the recognition device detects a curved road, the system adjusts the learning algorithm to account for the curvature, modifying how the neutral point is calculated and updated. This parameter adaptation enables accurate neutral point learning on both straight and curved roads.
2Productivity
If feedback control is performed without accurate neutral point calibration, then the control system operates continuously, but the steering angle converges excessively or insufficiently to the target angle
Solution Approach 1:
The patent enhances the feedback mechanism by using the recognized lane curvature as additional feedback information. The recognition device continuously detects road geometry, and this information feeds back into the neutral point learning process. This dual feedback loop (steering angle feedback plus road geometry feedback) enables the system to maintain continuous operation while ensuring accurate steering angle convergence by constantly refining the neutral point based on actual road conditions.
3Adaptability or versatility
If the neutral point is constantly learned using traditional methods, then adaptation to road conditions is attempted, but the learning accuracy is insufficient on curved roads leading to control performance degradation
Solution Approach 1:
The patent makes the neutral point learning system universal by enabling it to function accurately on both straight roads and curved roads. The recognition device detects various road geometries, and the learning algorithm adapts its behavior based on the detected geometry type. This multi-functional capability allows the same learning system to maintain high accuracy across different road conditions, eliminating the limitation of traditional methods that work only on straight roads.
Data Source
AI summary
A driver assistance apparatus includes a neutral point learning calculator. The neutral point learning calculator sets a neutral point learning correction value by which a neutral point of a steering angle instruction value is to be corrected. The neutral point learning calculator includes: a first estimated lateral acceleration calculator that calculates a first estimated lateral acceleration based on a steering angle and a vehicle speed; a second estimated lateral acceleration calculator that calculates a second estimated lateral acceleration, based on the vehicle speed and a lane curvature; a lateral acceleration difference calculator that calculates a lateral acceleration difference based on a difference between the first and the second estimated lateral accelerations; a steering angle difference calculator that calculates a steering angle difference, based on the lateral acceleration difference; and a neutral point learning correction value setter that sets the steering angle difference to the neutral point learning correction value.