Driver Scoring Using Clustered Telematics and Maintenance Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for tracking driver behavior in on-demand transportation services are inadequate, relying on passenger feedback and maintenance records, which can be incomplete or inaccurate, failing to effectively assess driver performance and accountability.
Innovation Solution
A system and method that utilize a server to receive and analyze maintenance, booking, and behavioral data to segregate feature values into clusters, training a classifier to determine a driver score based on performance indicators, ensuring accurate and efficient tracking of driver behavior.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If passenger feedback and maintenance records are used to track driver behavior, then driver performance can be assessed, but the assessment is incomplete and inaccurate due to missing data and potential fraud
Solution Approach 1:
The system segments driver behavior tracking into multiple independent data sources: passenger feedback, maintenance records, and telematics data. Each source captures different aspects of driver behavior, and their combination provides a comprehensive view that overcomes the limitations of any single source.
Solution Approach 2:
The system introduces telematics devices and machine learning models as intermediaries between the driver and the assessment system. These intermediaries automatically collect and analyze data, reducing reliance on manual feedback and records that are prone to incompleteness and fraud.
2Measurement precision
If passenger feedback is collected for each ride, then driver performance can be measured, but the feedback does not capture driver behavior when the vehicle is not booked
Solution Approach 1:
The telematics device continuously monitors driver behavior and vehicle status regardless of booking status. This continuous data collection ensures that driver behavior is tracked both during and outside of passenger rides, providing a complete picture of driver performance over time.
Solution Approach 2:
The system collects and stores telematics data continuously in advance, so that comprehensive driver behavior information is already available when needed for assessment, eliminating gaps in coverage during unbooked periods.
3Measurement precision
If maintenance records are used to assess driver behavior, then vehicle maintenance compliance can be tracked, but the records do not consider driver performance or negligence on the road
Solution Approach 1:
The system merges maintenance records with telematics data that captures driver behavior on the road. This combination allows the assessment system to evaluate both maintenance compliance and driving performance together, providing a holistic view of driver behavior that neither data source could provide alone.
4Measurement precision
If manual feedback collection and maintenance record analysis are used, then driver behavior can be tracked, but significant human effort is required and the process is inefficient
Solution Approach 1:
The telematics system automatically collects, stores, and transmits driver behavior data without manual intervention. The machine learning model automatically analyzes this data and generates assessments, eliminating the need for manual feedback collection and analysis while maintaining high tracking capability.
Solution Approach 2:
The system replaces manual mechanical processes of feedback collection and record analysis with automated electronic telematics devices and machine learning algorithms, dramatically improving efficiency while maintaining or enhancing tracking precision.
Data Source
AI summary
A method for determining a driver score for a first driver of a first vehicle. The method comprises receiving, from a database server, first maintenance data, first booking data, and first vehicle data associated with a plurality of vehicles, and first driver behavioral data for a plurality of drivers. The method includes obtaining a plurality of features and a plurality of feature values based on the first maintenance data, the first booking data, the first vehicle data, and the first driver behavioral data. The method further includes segregating the plurality of feature values into a plurality of clusters. The method includes training a classifier to determine the driver score. The method further includes receiving from the database server, a first dataset associated with the first vehicle and the first driver. The method includes determining the driver score for the first driver based on an output of the trained classifier.