Automatic Yield Determination for Cooperative Driving Data Labeling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current systems lack an efficient method for automatically collecting and labeling data on vehicle yielding behaviors during merging driving scenarios, relying on manual processes or outsourcing for data labeling, which is time-consuming and limited in scale.
Innovation Solution
An in-vehicle yield determination unit automatically detects and labels vehicles as yielding or not yielding by analyzing sensed data features and positioning data during lane changes, with labeled data uploaded to a central server to refine driver behavior models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual data labeling processes are used, then data accuracy can be maintained, but time consumption and labor costs increase significantly
Solution Approach 1:
The system performs automatic data labeling using the piloted vehicle's own sensors and processing capabilities. The yield determination unit automatically detects merging scenarios, tracks surrounding vehicles, determines yielding behavior, and labels data without external human intervention, enabling the system to serve itself in the data collection and labeling process
Solution Approach 2:
The patent replaces manual mechanical labeling processes with automated electronic sensing and processing systems. Sensors detect vehicle positions and behaviors, processors analyze the data to determine yielding intentions, and software automatically labels the collected data, substituting human manual work with electronic automation
2Measurement precision
If outsourcing data labeling is used, then specialized expertise can be obtained, but scalability and control are limited
Solution Approach 1:
The piloted vehicle system performs data labeling independently using its own embedded yield determination unit and processing capabilities, eliminating the need to outsource to external parties. This self-service approach enables unlimited scalability since the system can process data at its own pace without external bottlenecks
Solution Approach 2:
The yield determination unit integrates multiple functions including scenario detection, vehicle tracking, yielding behavior analysis, and data labeling within a single system. This multi-functional design allows the same system to handle various data labeling tasks across different merging scenarios, increasing overall productivity and scalability
3Productivity
If automatic yield determination is implemented, then data collection efficiency increases, but system complexity increases
Solution Approach 1:
The automatic yield determination system is divided into distinct functional modules: scenario detection module that identifies merging scenarios, vehicle tracking module that monitors surrounding vehicles, yield determination module that analyzes yielding behavior, and data labeling module that tags collected data. This segmentation allows each module to handle specific tasks independently, improving efficiency while managing complexity through modular design
Solution Approach 2:
The system performs preliminary detection of merging scenarios and pre-processing of sensor data before full analysis. By identifying relevant scenarios in advance and pre-filtering data, the system prepares information for subsequent yield determination and labeling steps, improving overall efficiency by avoiding processing of irrelevant data
Data Source
Figure 1A~1D
Figure 2
Figure 3
AI summary
A yield determination system (200) for automatically collecting, determining, and labeling yield behaviors of vehicles (10, 30, 35) during cooperative driving scenarios. The system (200) includes sensors (205) for detecting the start and stop of the scenario, a data recorder (220) for automatically collecting the data, and an annotation unit (225) for automatically labeling features of interest about a determination vehicle (10) and surrounding vehicles (30 ,35) during the scenario. The labeled data may be automatically uploaded and processed to insert the labeled features into a learning model to predict vehicle behavior in cooperative driving scenarios.