Object Recognition Apparatus Low-Resolution State Determination
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional object recognition apparatuses face challenges in accurately distinguishing between multiple vehicles traveling side by side due to low angular resolution, leading to deteriorated transverse and longitudinal recognition accuracy, especially when vehicles are far away, causing them to be recognized as a single object.
Innovation Solution
An object recognition apparatus that includes a data reception unit, an estimation processing unit, a low-resolution-state determination unit, and a recognition processing unit, which estimates object state changes, determines if data is in a low-resolution state, and adjusts the reflection degree of object data in recognition data to improve accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If sensor data is used directly for object recognition, then the processing speed is improved, but the recognition accuracy deteriorates when data is in a low-resolution state
Solution Approach 1:
The patent performs preliminary detection of the resolution state of sensor data before proceeding with object recognition. By detecting whether data is in a low-resolution state in advance, the system can adjust the reflection degree of the data accordingly, ensuring accurate recognition results without sacrificing processing speed. This preliminary action prevents accuracy deterioration while maintaining efficient processing.
2Area of stationary object
If multiple vehicles traveling side by side are detected, then the detection coverage is improved, but the ability to separate and identify individual vehicles deteriorates when the angle between them is small
Solution Approach 1:
The patent applies different processing strategies to different spatial regions and data characteristics. When vehicles are detected in a low-resolution state (small angle between them), the system adjusts the reflection degree of the data differently than when vehicles are well-separated. This local quality approach ensures that each data point is processed according to its specific resolution characteristics, maintaining vehicle separation accuracy across the entire detection coverage area.
Data Source
AI summary
There are provided an object recognition apparatus that raises the recognition accuracy for a surrounding object and a vehicle control apparatus, and an object recognition method and a vehicle control method. An object recognition apparatus receives object data, which is a state value of the object, from a first sensor for detecting a surrounding object; compares estimation data obtained through estimation of a state value of the object, based on recognition data calculated in a past period, with the object data, and determines whether or not the object data is data in a low-resolution state; then, in accordance with the determination result, calculates the state value of the object by use of object data and estimation data and then generates the state value as recognition data, so that the recognition accuracy for an object is raised.