Target Detection Apparatus Correcting Width Errors from Road Gradient
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing collision avoidance systems using radar and cameras face inaccuracies in target detection due to measurement errors caused by road gradient differences and vehicle pitching, leading to potential erroneous determination of target identity.
Innovation Solution
A target detection apparatus that specifies regions on an X-Y plane using radar and camera data, determines target overlap, estimates true target width, and corrects image region positions to improve accuracy, specifically by using a collision reduction ECU connected to millimeter-wave radar and monocular camera systems.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If the criterion for judgment is relaxed depending on the distance to the target, then targets are easily determined to be the same, but accuracy in determination concerning the identity of the targets is lowered due to unreflected road gradient differences and vehicle pitching influences
Solution Approach 1:
The patent dynamically adjusts the judgment criterion based on multiple parameters including distance to target, road gradient, and vehicle pitching angle. Instead of using a fixed threshold, the system modifies the acceptable range for target identity determination by incorporating these varying parameters, thereby maintaining both adaptability to different distances and precision by compensating for environmental influences.
Solution Approach 2:
The system uses feedback from road gradient sensors and vehicle attitude sensors to continuously adjust the judgment criterion. By feeding back information about road conditions and vehicle posture, the system can compensate for shifts in the point at infinity on the image, thereby maintaining accurate target identity determination even when distance alone would be insufficient.
2Measurement precision
If radar and camera are used to detect the target, then measurement accuracy is improved by mutual complementation, but a step is required to determine identity between targets based on detection information from both sensors
Solution Approach 1:
The patent merges radar detection information with camera image information by establishing a correspondence relationship between the two. The system combines the distance measurement strength of radar with the visual identification capability of the camera, creating a unified target detection framework that leverages the complementary strengths of both sensors while streamlining the identity determination process.
Solution Approach 2:
The system introduces an intermediary judgment criterion that bridges radar and camera data. This intermediary mechanism uses the detected target position from radar and image information from the camera together with road gradient and vehicle posture data to determine whether they refer to the same target, thereby simplifying the overall determination process while maintaining high accuracy.
3Ease of operation
If the point at infinity on the image is used for target detection, then target width and direction can be measured, but measurement error is generated when road gradient differs between vehicle and target positions or when vehicle pitching occurs
Solution Approach 1:
The system performs preliminary correction by detecting road gradient and vehicle pitching angle before conducting target width measurement. By anticipating and compensating for the effects of road gradient differences and vehicle posture changes on the point at infinity, the system prevents measurement errors from occurring in the first place, thereby maintaining both ease of operation and measurement precision.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enhances the accuracy of target position detection and width measurement by correcting for errors caused by road gradient and pitching influences, preventing erroneous target identification and improving collision avoidance control.
Implementation Method 1
a target is detected by a millimeter-wave radar... the distances can be measured based on radar waves reflected from the target
Implementation Method 2
the target can be identified based on an actually picked up image... the image information of the camera is suited to measure the target width and the direction
Data Source
AI summary
A first specifying-means specifies a first-region including a first detection-point, which is of a first-target on an X-Y plane on which a width direction of a vehicle is defined as an X-axis, and a longitudinal direction of the vehicle is defined as a Y-axis. A second specifying-means specifies a second-region including a second detection-point, which is of a second-target on the X-Y plane, based on a direction of the second detection-point and a target-width, which is a width along the X-axis of the second-target. A determination means determines that the first- and second-targets are the same, provided that an overlapping portion exists therebetween. An estimation means estimates a true-value target-width, based on the direction of the second detection-point and the first detection-point, provided that the first- and second-targets are the same. A correction-means corrects a position of the second-region specified next time by using the true-value target-width as the target-width.