Vehicle Child Presence Detection Using Time-Series Vibration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for detecting child presence in vehicles face challenges in accuracy and computational efficiency, particularly in distinguishing between a child and other objects using image and radar data, and lack effective signal processing improvements.
Innovation Solution
A method utilizing time series data and features, especially the evolution of time series features, to improve detection accuracy by analyzing sensor data from a monitoring device, which includes steps such as trigger activation, composing time series data, extracting features, and using a spiking neural network (SNN) for classification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If image sensors and gesture recognition are used for child detection, then the system can detect child presence, but the system may be deactivated by a child's gestures and lacks reliability
Solution Approach 1:
The patent changes the detection parameter from static image-based gesture recognition to dynamic time-series vibration analysis. By monitoring vibration patterns over time and analyzing their evolution, the system can distinguish between intentional deactivation gestures and natural child movements, thereby maintaining reliable system activation while achieving accurate child detection.
Solution Approach 2:
The patent replaces the image-based detection system with a vibration-based detection system. Instead of using cameras and gesture recognition algorithms, the system uses vibration sensors to detect mechanical vibrations caused by the child's presence and movements, providing more reliable and difficult-to-falsify detection.
2Measurement precision
If multiple sensors are used to improve detection sensitivity, then detection accuracy improves, but computational effort and complexity increase
Solution Approach 1:
The patent segments the vibration signal analysis into distinct time periods: an initial period immediately after door closing to capture vibration characteristics, and a subsequent period to monitor evolution. This temporal segmentation allows the system to process sensor data more efficiently by focusing on critical time windows rather than continuous analysis, reducing computational complexity while maintaining high detection accuracy.
Solution Approach 2:
The patent performs preliminary filtering and feature extraction on vibration signals immediately upon detecting door closing. By pre-processing the signals to extract relevant vibration characteristics before full analysis, the system reduces the computational burden on subsequent processing stages while preserving essential detection information.
3Productivity
If traditional discrimination methods are used for radar data processing, then the system can identify objects, but detection speed and accuracy are insufficient
Solution Approach 1:
The patent replaces traditional radar discrimination methods with a vibration analysis approach using time-series signal processing. By analyzing the temporal evolution of vibration patterns rather than relying on static radar cross-section discrimination, the system achieves both faster detection speeds and higher identification accuracy for distinguishing children from other objects.
Solution Approach 2:
The patent introduces dynamic time-series analysis of vibration signals, monitoring how vibration characteristics evolve over time. This dynamic approach captures the unique vibration signatures of different objects and movements, enabling faster and more accurate identification compared to static traditional methods.
Data Source
Figure 1~2
Figure 3~4
Figure 5
AI summary
A computer-implemented method for detecting at least one object or at least one person in a monitored area of a monitoring device, the method comprising the steps of : - detecting a trigger activation (S1), such as a closing of a door of a vehicle, - receiving signal data (S2) from the monitoring device, - composing time series data (S3) based on the signal data received, - extracting times series features (S4) from the time series data, - identifying an evolution of times series features over a predetermined period of time starting with the detected trigger activation, - detecting (S5) the at least one object or the at least one person based on the evolution of times series features.