Vehicle Object Perception Using Dual Labeling and Shape Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing autonomous driving systems face challenges in accurately and precisely perceiving surrounding objects due to variations in object detection and recognition, necessitating improved perception accuracy for safe and reliable vehicle operation.
Innovation Solution
A dual perception process involving labeling-based and shape-based methods is employed, where actual data from sensors is compared with reference labeling and shape data to enhance object recognition accuracy by ensuring matching conditions are met in both processes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a single perception method (either labeling-based or shape-based) is used, then the system complexity is low, but the object perception accuracy is insufficient
Solution Approach 1:
The patent combines labeling-based perception and shape-based perception into a unified dual perception system. The processor executes both perception methods simultaneously, comparing actual labeling data with reference labeling data while also comparing actual shape data with reference shape data, then integrates the results to achieve more accurate and reliable object perception than either method could achieve alone.
2Measurement precision
If dual perception process is implemented, then object perception accuracy is improved, but the processing time increases
Solution Approach 1:
The system performs preliminary actions by pre-storing reference labeling data and reference shape data in the memory before actual perception tasks. When an object needs to be perceived, the processor can immediately compare sensor data against these pre-prepared references without needing to generate or compute reference data during the perception process, thereby reducing processing time while maintaining dual perception accuracy.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
An apparatus for perceiving an object includes memory that is memory for storing reference labeling data required to perceive an object around an ego vehicle, wherein the reference labeling data defines the type of object that is a target of perception, and a processor configured to extract actual labeling data, corresponding to the reference labeling data, from the results of the detection of a target object by a sensor of the ego vehicle and to perceive the target object by comparing the extracted actual labeling data and the reference labeling data.