Trip Type Identification Using Hybrid Segmentation and Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing telematics systems face challenges in accurately identifying the type of trip, such as transportation mode and role of the person, which is crucial for applications like vehicle insurance and driving safety analysis, often resulting in errors and irrelevant data inclusion.
Innovation Solution
The system improves trip type identification by using historical information and sensor data from mobile devices to derive features such as motion patterns, location, and device usage, employing hybrid segmentation and classification models like Hidden Markov Models to accurately classify transportation modes and roles.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If sensor data is collected continuously during a period, then data completeness is improved, but data processing complexity and energy consumption increase
Solution Approach 1:
The patent segments continuous sensor data into discrete trips based on motion thresholds and temporal patterns. By dividing the continuous data stream into meaningful trip segments, the system maintains data completeness while reducing processing complexity through focused analysis of segmented data rather than continuous processing.
Solution Approach 2:
The system employs periodic sampling and threshold-based detection to identify trip events. Instead of processing every sensor reading continuously, the system periodically evaluates motion thresholds and trip conditions, reducing energy consumption and processing complexity while maintaining reliable trip detection through structured periodic analysis.
2Measurement precision
If trip type identification uses only recorded motion data, then system simplicity is maintained, but identification accuracy deteriorates
Solution Approach 1:
The patent merges multiple data sources including motion data, location data, device usage patterns, and historical trip information to identify trip types. By combining these diverse data sources, the system achieves high identification accuracy while managing complexity through integrated multi-source analysis rather than relying on a single complex system.
Solution Approach 2:
The system performs preliminary processing of sensor data to extract features and patterns before final trip type classification. By pre-processing motion data, location data, and device usage information into meaningful features, the system improves identification accuracy while reducing the complexity of the final classification stage through structured feature extraction.
3Measurement precision
If manual labeling is used for trip data, then data accuracy is improved, but time consumption and labor costs increase
Solution Approach 1:
The patent implements automated trip type identification using machine learning models that analyze sensor data, location information, and device usage patterns to classify trips independently. This self-service approach eliminates manual labeling requirements, achieving high data accuracy through automated classification while completely eliminating the time consumption and labor costs associated with manual annotation.
Solution Approach 2:
The system incorporates feedback mechanisms where identified trip types and patterns from historical data improve future classification accuracy. By using feedback from accumulated trip data to refine classification models, the system achieves high accuracy automatically without requiring manual labeling, reducing both time consumption and dependency on human annotators.
Data Source
AI summary
Among other things, an operation of a computational device in identifying a type of a given trip of a person is improved. Historical information is stored about prior trips of the person or of other people or both. The historical information is based on other than recorded motion data of the trips. Features are derived about the prior trips from the historical information. Features indicative of the type of the given trip are identified by the computational device. The type of the given trip is identified based on the features derived from the historical information.