Vehicle Data Collection Using Invariant Feature Maps
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vehicle data collection systems rely on pre-defined, rule-based trigger conditions, which limit their ability to adapt to new criteria and can result in false negatives or false positives, reducing the effectiveness of data collection.
Innovation Solution
The use of neural networks to develop invariant feature maps that can be compared with data collected by vehicle sensors, allowing for more precise data collection and reduction of false positives and negatives, while also enabling the collection of data related to new trigger conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If pre-defined rule-based trigger conditions are used for data collection, then the system is simple to implement, but the system cannot adapt to new criteria and produces false positives and negatives
Solution Approach 1:
The patent replaces traditional rule-based trigger conditions with neural network-based semantic understanding. The neural network processes natural language queries and automatically generates appropriate sensor data collection criteria, eliminating the need for manual rule definition and enabling adaptation to new conditions without system reconfiguration.
Solution Approach 2:
The system changes the parameter representation from fixed rule-based conditions to dynamic semantic parameters processed by neural networks. This allows the trigger conditions to be modified through natural language input rather than requiring changes to the underlying system architecture or code.
2Reliability
If all sensor data is transmitted to the server for processing, then complete data is available for analysis, but processing load and power consumption increase
Solution Approach 1:
The server performs preliminary processing by generating invariant feature maps from historical data and transmitting these compressed representations to vehicles. This preliminary action reduces the amount of raw sensor data that needs to be transmitted and processed in real-time, thereby reducing power consumption while maintaining analysis accuracy.
Solution Approach 2:
The system extracts only the essential features needed for analysis by creating invariant feature maps that capture the most relevant information from sensor data. This extraction process removes redundant data, reducing transmission requirements and processing load while preserving the critical information needed for accurate analysis.
3Measurement precision
If invariant feature maps are used for data comparison, then data collection precision is improved, but the amount of data to be processed increases
Solution Approach 1:
The system transforms raw sensor data into invariant feature maps that represent the same information in a more compact and comparable format. This parameter transformation maintains the essential characteristics needed for precise comparison while reducing the computational complexity and effective data volume that needs to be processed.
Solution Approach 2:
Instead of transmitting and processing complete raw sensor datasets, the system creates and transmits invariant feature map copies that contain the essential information needed for comparison. These compressed representations enable precise data matching without requiring the full original data sets, thereby reducing processing requirements.
Data Source
AI summary
A vehicle data collection system includes a vehicle-mounted sensor; a non-transitory computer readable medium configured to store instructions; and a processor connected to the non-transitory computer readable medium. The processor is configured to execute the instructions for generating an invariant feature map, using a first neural network; and comparing the invariant feature map to template data to determine a similarity between the invariant feature map and the template data, wherein the template data is received from a server. The processor is further configured to execute the instructions for determining whether the determined similarity is exceeds a predetermined threshold; and instructing a transmitter to send the sensor data to the server in response to a determination that the determined similarity exceeds the predetermined threshold.