Vehicle Collision Characterization Using Multi-Source Telematics
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional collision detection systems rely on accelerometer data, which is unreliable due to high false positive and false negative rates, as they cannot differentiate between actual collisions and non-collision events such as bumps, potholes, or loose telematics monitor installations, leading to inaccurate responses.
Innovation Solution
A computerized method that analyzes data from various telematics monitors, including accelerometers, GPS, and contextual information, to determine the likelihood of a collision by evaluating data before and after a potential collision event, using a trained classifier to classify events and reduce false detection rates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If accelerometer data is used for collision detection, then collision detection capability is provided, but false positive and false negative rates are high
Solution Approach 1:
The patent combines multiple data sources including accelerometer data, GPS data, and contextual information from various telematics monitors to create a comprehensive collision detection system. This merging of multiple independent data streams allows the system to cross-validate signals and distinguish true collision events from false positives such as bumps or potholes, thereby improving both reliability and measurement precision simultaneously
Solution Approach 2:
The patent introduces a trained classifier as an intermediary component that processes and interprets raw telematics data. This classifier acts as a mediator between the sensor data and collision detection decisions, applying learned patterns to accurately differentiate between collision events and non-collision events, thus resolving the precision-reliability contradiction
2Measurement precision
If multiple telematics monitors are analyzed, then detection accuracy improves, but system complexity increases
Solution Approach 1:
The patent pre-trains classification models using historical telematics data before deployment. This preliminary action of training the classifier offline allows the system to handle complex multi-sensor data analysis during runtime with minimal computational overhead, maintaining high detection accuracy while managing operational complexity
Solution Approach 2:
The system uses historical collision and non-collision data to automatically train and improve its own classifier without external intervention. This self-service approach allows the system to adapt to specific fleet patterns and environmental conditions, improving precision while the trained model handles the complexity of multiple data sources
Data Source
AI summary
Described herein are various techniques, including systems and non-transitory instructions, that, in response to obtaining information regarding a potential collision between a vehicle and an object, obtain data describing the vehicle for a time period extending before and after a time of the potential collision. The system may determine a likelihood that the potential collision is a non-collision event based on the data describing the vehicle by performing one or more assessments. The assessments may include telematics monitor assessment, driver behavior assessment, road surface feature assessment, trip correlation assessment, and/or context assessment. In response to determining that the likelihood indicates that the potential collision is not a non-collision event, the system may trigger one or more actions responding to the potential collision.