Vehicle Object Detection via Radar Data Association
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing object detection systems face challenges in continuously and correctly identifying a preceding vehicle when irregular detection occurs, leading to potential collisions due to unstable radar reflection points and incorrect object recognition.
Innovation Solution
An object detection device equipped with a radar detection object acquiring part and an object recognition part, featuring a computer system with a preceding vehicle selection part, irregular-detection detecting part, and association processing part, which associates history data of a stable object with an unstable object to maintain accurate detection even during irregular detection events.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If radar reflection points are used for object detection, then object detection capability is provided, but irregular detection occurs causing difficulty in continuously and correctly selecting the same object as a preceding vehicle
Solution Approach 1:
The system performs preliminary actions by detecting multiple candidate objects before final selection, storing their detection results in advance. When irregular detection occurs, the system can switch to previously stored candidate objects to maintain continuous and correct detection of the preceding vehicle.
Solution Approach 2:
The system introduces an intermediary mechanism by detecting multiple candidate objects (first object and second object) as intermediaries between the unstable radar reflection points and the final preceding vehicle selection. This allows smooth transition when irregular detection occurs.
2Area of stationary object
If multiple radar detection objects are detected due to irregular detection, then detection coverage is improved, but correct recognition of the preceding vehicle becomes difficult
Solution Approach 1:
The system segments the detection process by dividing multiple detected objects into different categories (first object used for selection, second object not used). This segmentation allows the system to maintain high detection coverage while accurately identifying which object is the true preceding vehicle through subsequent association processing.
Solution Approach 2:
The system implements feedback by performing association processing that compares detection results across multiple frames. The detection result determination unit uses feedback from the association processing to correctly identify the preceding vehicle even when multiple objects are detected, by analyzing temporal consistency and spatial relationships.
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
Enables continuous and correct selection of the preceding vehicle, reducing the risk of missing the object and ensuring safe vehicle control by stabilizing object recognition through data association, even in conditions of irregular radar reflection point variations.
Implementation Method 1
An object detection device is a known device using a radar device
Implementation Method 2
The object detection device receives radar waves which have been transmitted from the radar device and reflected by an object located in front of the own vehicle
Data Source
AI summary
An ECU as an object detection device and a radar device are mounted on a vehicle. The radar device acquires front-view images containing a detection-target object such as a preceding vehicle present within a predetermined area around the vehicle. The radar device recognizes the detection-target object in time series based on the acquired front-view images. The ECU realizes a preceding vehicle selection part, an irregular-detection detecting part and an association processing part. The selection part selects the object when a relationship between the object and the vehicle satisfies a predetermined condition. The detecting part detects occurrence of an irregular detection when a first object selected as the preceding vehicle and a second object not selected as the preceding vehicle are detected based on the object. The association processing part performs an association process of associating history of the first object to the second object when the irregular detection has occurred.