In-Cabin Child Presence Detection Under Visual Occlusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing child presence detection methods in vehicles face challenges due to visual occlusions and the inability to capture granular details of an individual's size and pose, leading to errors in detecting age and presence, which can compromise safety and regulatory compliance.
Innovation Solution
The use of sensor data from images and RADAR to detect occupants, classify them as children or adults, and estimate age through machine learning models, combining predictions from face-based, child seat-based, and limb length-based assessments to enhance detection accuracy and reliability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If visual-based age estimation methods are used, then the system can detect occupant age from images, but detection accuracy deteriorates due to visual occlusions such as blankets over child seats or extreme head poses
Solution Approach 1:
The patent combines multiple detection methods (visual age estimation, child seat detection, and RADAR-based presence detection) into a unified system. The machine learning model integrates predictions from different sensor types and detection approaches to compensate for weaknesses in individual methods, particularly when visual occlusions are present.
Solution Approach 2:
The system introduces an intermediate detection layer that detects child seats as proxies for child presence. When visual occlusion prevents direct age estimation, the system uses child seat detection as an intermediary indicator to infer child presence, and combines this with RADAR data to maintain detection reliability.
2Reliability
If RADAR sensors are used to detect child presence, then the system can measure distance and movement parameters, but it struggles to capture granular details of individual size and pose needed for age detection
Solution Approach 1:
The patent merges RADAR sensor data with visual sensor data in a multi-modal sensor fusion framework. The machine learning model processes both RADAR measurements (distance, movement, speed) and visual features (face images, body pose, limb length) simultaneously, allowing the system to leverage the reliability of RADAR for presence detection while using visual data for precise age estimation when available.
Solution Approach 2:
The system applies different detection strategies to different parts of the detection process. RADAR is used for reliable occupancy and presence detection, while visual sensors are used for detailed age estimation when conditions permit. The system dynamically selects which sensor modality to trust based on local conditions such as occlusion levels and sensor quality.
3Reliability
If multiple detection methods are combined to improve accuracy, then detection reliability improves, but system complexity increases due to multiple sensors and machine learning models
Solution Approach 1:
The machine learning model serves multiple functions: it performs age estimation, detects child presence, identifies child seats, and fuses data from multiple sensor types. This multi-functional approach consolidates what could be separate complex systems into a single unified model, reducing overall system complexity while maintaining high reliability through multi-modal input.
Data Source
AI summary
In various examples, sensor data (e.g., image and/or RADAR data) may be used to detect occupants and classify them (e.g., as children or adults) using one or more predictions that represent estimated age (e.g., based on detected limb length, a detected face) and/or detected child presence (e.g., based on detecting an occupied child seat). In some embodiments, multiple predictions generated using multiple machine learning models (and optionally one or more corresponding confidence values) may be combined using a state machine and/or one or more machine learning models to generate a combined assessment of occupant presence and/or age for each occupant and/or supported occupant slot. As such, the techniques described herein may be utilized to detect child presence, detect unattended child presence, determine age or size of a particular occupant, and/or take some responsive action (e.g., trigger an alarm, control temperature, unlock door(s), permit or disable airbag deployment, etc.).