Measuring Vehicle Data Collection Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for collecting and processing sensor data from measuring vehicles are inefficient, as they struggle to categorize and analyze the vast amount of data, leading to incomplete recording of relevant driving situations, increased data transfer times, and redundant data processing, which hampers the development of autonomous driving systems.
Innovation Solution
A method and system that determine measurement categories, adjust measurement settings based on data from measuring vehicles, and use feature vectors to optimize data collection and transmission, employing techniques like auto-encoders and k-means clustering to categorize and reduce data redundancy, allowing for targeted data collection and efficient processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If measurement data is recorded continuously during measurement drives, then comprehensive driving situations are captured, but data volume increases significantly leading to redundant data storage and transfer
Solution Approach 1:
The system performs preliminary categorization of measurement data using feature vectors and clustering algorithms (k-means, GMM) to identify relevant driving situations before full data storage. This preliminary action filters out redundant data, storing only categorically significant measurements while maintaining comprehensive coverage of important driving scenarios.
Solution Approach 2:
The patent extracts essential features from measurement data to create feature vectors that represent driving situations. By taking out only the relevant features and categories rather than storing all raw sensor data, the system reduces data volume while preserving the information needed for autonomous driving system evaluation.
2Loss of information
If all recorded measurement data is transferred to remote computing systems, then complete data is available for analysis, but transfer time and processing time increase significantly
Solution Approach 1:
The system performs preliminary categorization and filtering of measurement data using feature vectors and clustering algorithms before transfer. This preliminary processing identifies and retains only relevant driving situations, reducing the data volume that needs to be transferred and processed while ensuring that complete information about important scenarios is preserved.
Solution Approach 2:
The patent extracts and transfers only the essential categorized data and feature vectors rather than all raw measurement data. This extraction of relevant information maintains data completeness for analysis purposes while dramatically reducing transfer time and processing requirements.
3Reliability
If measurement campaigns run for extended periods to capture rare driving situations, then sufficient relevant data is collected, but measurement duration and resource consumption increase
Solution Approach 1:
The system implements feedback loops where measurement data is continuously categorized, analyzed, and used to adjust measurement strategies. The categorization results provide feedback on which driving situations have been adequately captured and which require continued monitoring, enabling optimized measurement campaign duration while ensuring sufficient coverage of rare but critical scenarios.
Solution Approach 2:
The patent uses preliminary categorization of measurement data to identify rare and relevant driving situations early in the measurement process. This allows the system to focus resources on capturing sufficient data for underrepresented categories, reducing overall measurement duration while ensuring adequate coverage of all important scenarios including rare events.
Data Source
Figure 1~2
Figure 3~4
Figure 5~6
AI summary
A method for storing and transmitting measurement data from measuring vehicles is provided, comprising the steps of determining a number of measurement categories, determining a number of required measurements for each of the determined measurement categories, starting a measurement campaign with a number of measuring vehicles, wherein each measuring vehicle is configured with the determined categories, changing a measurement setting for at least some of the number of measuring vehicles in case an information from at least one measuring vehicle is received indicating that a determined number of required measurements has been reached for at least one determined category. The change in the measurement setting can especially be a change in the route planning of at least one measuring vehicle.