Autonomous Vehicle Event Detection Using Machine Learning Classifiers
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for detecting and classifying events in autonomous vehicles rely on human passengers, which are costly, subjective, and not scalable, leading to inconsistent and delayed event detection and classification, especially in simulated driving sessions.
Innovation Solution
The use of machine-learned classifier models that extract features from vehicle data, such as those obtained through continuous wavelet transform, to automatically detect and classify events like high acceleration, deceleration, and juke motions, trained using annotated vehicle data logs to provide consistent and accurate event classification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If human passengers are used to detect and classify events, then event detection can be performed, but the method becomes costly, subjective, and not scalable
Solution Approach 1:
The patent replaces the mechanical system of human passengers performing manual event detection and classification with an automated computer-based system that processes vehicle data through defined algorithms and machine learning models, eliminating human subjectivity and enabling scalable deployment across multiple vehicles
Solution Approach 2:
The system enables vehicles to automatically detect and classify their own events using onboard sensors and processing capabilities, with the vehicle's own data serving as the input for self-assessment of ride quality and event classification, removing the need for external human evaluation
2Ease of operation
If human passengers are used for event detection, then event classification is possible, but the process becomes delayed and inconsistent
Solution Approach 1:
The patent replaces human-based event classification with automated computer processing that continuously analyzes vehicle data in real-time, eliminating the temporal delays and inconsistencies associated with human reaction times and subjective judgment variations
Solution Approach 2:
The system implements continuous event detection and classification through ongoing processing of vehicle sensor data, rather than discrete human evaluations, ensuring uninterrupted monitoring and immediate classification of events as they occur during vehicle operation
3Measurement precision
If human passengers are used, then subjective event assessment is achieved, but cost increases significantly
Solution Approach 1:
The patent substitutes human passengers with automated computing systems that process vehicle data through algorithms and machine learning models, eliminating the operational costs associated with hiring, training, and compensating human evaluators while maintaining or improving assessment accuracy through consistent automated judgment
Solution Approach 2:
The system uses inexpensive computational resources and algorithms to perform event detection and classification, replacing the expensive human resource with cost-effective automated processing that can be deployed across multiple vehicles without proportional cost increases
Data Source
AI summary
The present disclosure provides systems and methods for automatic event detection and classification for autonomous vehicles. One example method includes obtaining, by one or more computing devices, vehicle data descriptive of vehicle conditions associated with an autonomous vehicle during an autonomous driving session. The method includes extracting, by the one or more computing devices, a plurality of features from the vehicle data. The method includes determining, by the one or more computing devices using a machine-learned classifier, a classification for each of one or more candidate events based at least in part on one or more of the plurality of features that are respectively associated with the one or more candidate events. The method includes associating, by the one or more computing devices, the classification determined for each of the one or more candidate events with the vehicle data.