Camera-Based Steering Control Using Lane Area Values
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vehicle control systems based on bird's eye view coordinate systems require a large number of parameters and increase the computational load on onboard computers, and the alignment of output data from multiple sensors is cumbersome.
Innovation Solution
A vehicle control device that calculates target steering and vehicle speed based on area values within input images captured by a camera, using a neutral reference line and dividing lines to determine steering and speed adjustments, reducing the need for complex parameter tuning and computational load.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If bird's eye view coordinate system is used for vehicle control, then control accuracy is improved, but computational load and parameter tuning complexity increase
Solution Approach 1:
The patent replaces the complex bird's eye view coordinate system transformation with a direct image coordinate system approach. Instead of converting images to bird's eye view and then calculating control amounts, the system directly calculates steering and acceleration control amounts from the original image coordinates, eliminating the need for coordinate transformation and reducing parameter tuning complexity
Solution Approach 2:
The patent extracts only the essential control information (steering angle and acceleration) directly from the image data without performing full bird's eye view transformation. By taking out only the necessary control parameters and calculating them directly from image coordinates, the system maintains control accuracy while reducing computational complexity
2Measurement precision
If bird's eye view coordinate system is used for vehicle control, then control accuracy is improved, but load on onboard computer increases
Solution Approach 1:
The patent substitutes the computationally intensive bird's eye view transformation process with a direct image coordinate calculation method. By replacing the multi-step transformation pipeline with a direct calculation from image coordinates to control amounts, the system maintains accuracy while significantly reducing computational load on the onboard computer
3Measurement precision
If bird's eye view coordinate system is used for vehicle control, then control accuracy is improved, but alignment work for multi-sensor data increases
Solution Approach 1:
The patent creates a universal control framework that works directly with image coordinates, making the system applicable to multiple sensor types without requiring separate alignment procedures. The image coordinate system serves as a universal reference that can be directly used with cameras, LIDAR, and other sensors, eliminating the need for complex alignment work when integrating multiple sensor outputs
Data Source
AI summary
A vehicle control device includes: an input image acquirer that acquires an input image; a target steering amount calculator that calculates a target steering amount based on the input image; and a steering controller that manipulates an electric power steering to a positive side or a negative side based on the target steering amount. The target steering amount calculator is configured to execute: processing of recognizing a positive-side dividing line LP and a negative-side dividing line LN of the vehicle; processing of calculating at least either one of a positive-side area value between a neutral reference line L0 determined virtually and the positive-side dividing line LP, and a negative-side area value between the neutral reference line L0 and the negative-side dividing line LN; and processing of calculating the target steering amount based on at least either one of the positive-side area value and the negative-side area value.