Edge-Based Vehicle Mode Determination With Sensor Feature Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vehicle mode determination methods rely on central processing that fail to account for data collection time, type, and features, leading to accuracy loss and increased computing resource utilization.
Innovation Solution
Implementing vehicle mode determination systems on-board using edge-computing, which preprocesses data from accelerometers, gyroscopes, and GPS to generate feature vectors, apply machine learning models like SVM classifiers, and determine vehicle modes locally.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If vehicle data is collected and transmitted to a central server for processing, then comprehensive analysis can be performed, but data transmission time and network resource usage increase
Solution Approach 1:
The patent segments the data processing task by dividing it into two parts: feature extraction is performed locally on the mobile computing device, while only the extracted features (not the raw sensor data) are transmitted to the central server for final classification. This segmentation reduces the amount of data that needs to be transmitted and processed centrally, thereby reducing transmission time and network resource usage while maintaining determination accuracy.
Solution Approach 2:
The patent applies preliminary action by performing feature extraction locally before data transmission. The mobile computing device pre-processes the raw accelerometer, gyroscope, and GPS data to extract relevant features, and only these processed features are then transmitted to the server. This preliminary processing step reduces the data volume for transmission and allows the server to focus on the final classification task, improving overall efficiency.
2Measurement precision
If large amounts of vehicle data are transmitted over networks, then more data is available for analysis, but network bandwidth and computing resources are consumed
Solution Approach 1:
The patent extracts only the essential features from the raw vehicle data locally using unsupervised learning algorithms. Instead of transmitting all raw sensor data to the central server, only the extracted features (such as statistical moments, frequency domain characteristics, and temporal patterns) are transmitted. This extraction process significantly reduces the data volume for transmission and processing, thereby conserving network bandwidth and computing resources while maintaining the accuracy needed for vehicle mode determination.
3Power
If central server processing is used, then computational power is available, but data transmission requirements and response time increase
Solution Approach 1:
The patent implements local quality by enabling the mobile computing device to perform feature extraction and preliminary processing locally using unsupervised learning algorithms. This local processing capability reduces the dependency on continuous central server processing, allowing the system to operate more autonomously. The mobile device handles data collection, preprocessing, and feature extraction locally, while only transmitting processed features to the server for final classification, thereby reducing response time and system complexity.
4Loss of energy
If feature extraction is performed locally on mobile devices, then data transmission is reduced, but local processing capability must be sufficient
Solution Approach 1:
The patent replaces the mechanical approach of transmitting all raw data with a computational approach using unsupervised learning algorithms for automatic feature extraction. The mobile computing device uses machine learning models to automatically identify and extract relevant features from the sensor data without requiring manual feature engineering or complex preprocessing pipelines. This substitution reduces the computational burden on the mobile device while still achieving effective feature extraction, thereby reducing data transmission requirements without significantly increasing on-board processing complexity.
Data Source
AI summary
Methods, computer-readable media, software, and apparatuses may determine, based upon edge-computing operations, that a vehicular trip has been initiated and cause one or more sensors to collect vehicle data. One or more trip segments for at least a portion of the vehicular trip may be determined. In some aspects, for each trip segment, a first plurality of time features and a second plurality of frequency features may be determined, and may be concatenated with a third plurality of GPS features to form a feature vector. An accuracy measure may be determined based on the feature vector, and a mode for the vehicle may be predicted.