Blind-Region Vehicle Recognition With Adaptive Detection Sensitivity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vehicle control systems prioritize either rapidity or reliability in target recognition, failing to achieve both when detecting targets emerging from blind regions, leading to potential errors or delayed responses.
Innovation Solution
An outside environment recognition device that divides the traveling direction into multiple regions, applying different recognition methods based on contact possibility, using high sensitivity for imminent targets and lower sensitivity for distant ones.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If object recognition prioritizes rapidity when a target is detected in a shielded region, then the response time is reduced, but the reliability of detection decreases leading to potential erroneous detection
Solution Approach 1:
The shielded region is divided into multiple sub-regions based on distance from the own vehicle. The determination unit segments the detection approach into two phases: initial detection in distant sub-regions using reliable but slower recognition, and final detection in near sub-regions using rapid but less reliable recognition. This segmentation allows the system to optimize detection strategy for each spatial zone.
Solution Approach 2:
The system performs preliminary object recognition using reliable methods when targets are first detected in distant sub-regions of the shielded area. This preliminary detection establishes a baseline before switching to rapid recognition methods as the target approaches closer sub-regions, ensuring that reliable detection occurs before time-critical moments.
2Reliability
If object recognition prioritizes reliability when a target is detected in a shielded region, then the accuracy of detection is improved, but the response time increases causing delayed avoidance operations
Solution Approach 1:
The object recognition method dynamically switches between reliable and rapid recognition strategies based on the target's position and proximity to the own vehicle. As the target moves from distant to near sub-regions, the system transitions from reliable recognition to rapid recognition, making the detection approach adaptive rather than static.
Solution Approach 2:
Reliable object recognition is performed in advance when targets are in distant sub-regions, establishing detection confidence before the target enters critical near sub-regions. This preliminary reliable detection ensures accuracy is secured before the time-critical phase of close-proximity detection begins.
3Device complexity
If a single object recognition method is used for all regions in blind spots, then the system complexity is reduced, but the overall detection performance cannot achieve both rapidity and reliability simultaneously
Solution Approach 1:
The blind spot region is segmented into multiple sub-regions with different detection requirements. Rather than using a single uniform recognition method, the system applies different recognition strategies to different spatial segments, optimizing performance for each zone while maintaining manageable system complexity through structured division.
Solution Approach 2:
Different quality levels of object recognition are applied to different local regions of the blind spot. Distant sub-regions receive high-quality reliable recognition, while near sub-regions receive rapid recognition optimized for speed. This local quality differentiation allows the system to achieve optimal overall performance without excessive complexity.
Data Source
AI summary
There is provided an outside environment recognition device that rapidly performs target recognition processing when there is a possibility that a target coming into contact with an own vehicle appears. There is provided an outside environment recognition device that recognizes a target. An arithmetic operation device includes a blind region edge detection unit that detects an edge of a blind region where the target is likely to appear, based on outside environment information, a traveling path estimation unit that estimates a future path of the own vehicle based on a traveling direction and a speed of the own vehicle included in vehicle information, a recognition region division unit that divides a region between the edge and the path into a first region to which first target recognition processing is applied and a second region to which second target recognition processing is applied, based on a contact possibility between the own vehicle and a virtual target appearing on the path from the edge, and a target detection unit that detects a target appearing from the edge by applying the first target recognition processing in the first region, and detects the target appearing from the edge by applying the second target recognition processing in the second region. The first target recognition processing and the second target recognition processing have different recognition sensitivities.