Vehicle Wheel Impact Detection Using Deep Neural Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Vehicle wheel impacts, especially in autonomous or semi-autonomous vehicles, can go unnoticed, leading to potential damage and maintenance needs, as occupants may not be aware of the risks posed by such impacts.
Innovation Solution
A system utilizing a deep neural network (DNN) and threshold-based algorithms to detect and assess vehicle wheel impacts by analyzing data from sensors, including velocity, acceleration, and wheel speed, and actuating vehicle components accordingly.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If a vehicle operates autonomously or semi-autonomously, then vehicle operation is improved, but occupant awareness of wheel impact risks deteriorates
Solution Approach 1:
The system continuously monitors wheel impact events using sensors and provides feedback to the occupant through alerts or notifications. This resolves the contradiction by maintaining automation while compensating for the loss of occupant awareness through an artificial feedback loop that informs the occupant of impact events that would otherwise go unnoticed.
2Measurement precision
If wheel impact detection sensitivity is increased, then impact detection accuracy is improved, but false positive rate increases
Solution Approach 1:
The detection system is segmented into multiple independent components: vibration sensors detect mechanical impacts, acoustic sensors detect impact sounds, and image recognition analyzes visual data. Each component operates with its own thresholds and algorithms. By segmenting the detection function across multiple modalities, the system achieves high sensitivity for true impacts while reducing false positives, as an actual wheel impact will likely trigger multiple sensors simultaneously whereas false triggers are isolated to single sensors.
3Reliability
If multiple detection methods are combined, then impact detection reliability is improved, but system complexity increases
Solution Approach 1:
The patent merges vibration sensing, acoustic sensing, and image recognition into a unified wheel impact detection system. The sensors and processing units are integrated to work together, with a central system that receives data from all modalities and synthesizes a comprehensive detection result. This merging approach improves reliability through multi-modal verification while managing complexity through integrated architecture rather than separate independent systems.
Data Source
AI summary
Data describing operation of a vehicle is provided to a deep neural network. A vehicle wheel impact event is determined based on output of the deep neural network. Alternatively or additionally, it is possible to determine the wheel impact event based on output of a threshold based algorithm that compares vehicle acceleration and the velocity to one or more thresholds.