Child Detection Using Height Estimation for Mobile Collision Avoidance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing mobile systems, such as self-driving robots and vehicles, struggle to differentiate between adults and children for effective collision avoidance, necessitating tailored control strategies.
Innovation Solution
A method and device that utilize semantic segmentation and bird's-eye view image generation to detect foot and head pixel coordinates, estimate distance and height, and differentiate between adults and children, enabling customized collision avoidance policies.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If mobile systems use general detection methods to identify people, then they can detect the presence of persons, but they cannot differentiate between adults and children for effective collision avoidance
Solution Approach 1:
The patent segments the person detection task into multiple components: detecting the person's position, estimating their height relative to the vehicle, and classifying them as adult or child based on height thresholds. This segmentation allows the system to obtain detailed information (age group) from basic detection data.
Solution Approach 2:
The patent uses height estimation as an intermediate parameter to differentiate between adults and children. By calculating the person's height based on their position in the image and the vehicle's height, the system transforms basic detection data into actionable classification information.
2Reliability
If mobile systems implement tailored control strategies for children, then collision avoidance effectiveness is improved, but the system complexity increases
Solution Approach 1:
The patent applies different control strategies based on the detected age group of the person. When a child is detected (height below threshold), the system implements more conservative collision avoidance measures compared to when an adult is detected. This local quality approach optimizes safety without requiring completely different systems for different scenarios.
Solution Approach 2:
The patent implements dynamic control adjustments based on real-time detection results. The control strategy changes dynamically according to whether a child or adult is detected, allowing the system to adapt its behavior to the specific situation rather than using fixed complex rules.
3Device complexity
If mobile systems use simple detection methods, then the system remains simple, but they cannot accurately estimate person height and distance for proper classification
Solution Approach 1:
The patent introduces height estimation as an intermediary measurement that connects basic detection data to final classification. By using the person's position in the image and the vehicle's known height as intermediate parameters, the system can derive accurate height and distance information without requiring complex direct measurement devices.
Solution Approach 2:
The patent uses the vertical dimension (height) in the image as an additional parameter for classification. By analyzing the person's position relative to the vehicle height in the vertical dimension, the system can estimate both distance and height accuracy, transforming a 2D image problem into a 3D spatial understanding.
Data Source
AI summary
A method and a device may allow for detecting a child and controlling a mobile system (e.g., a vehicle and/or robot) that drives in a space where people may be. The method may include acquiring an image from a camera of the mobile system; extracting a first foot pixel coordinate corresponding to a foot of a person and a head pixel coordinate corresponding to a head of the person from the acquired image by using semantic segmentation; generating a bird's-eye view image from the acquired image; acquiring a second foot pixel coordinate, corresponding to the first foot pixel coordinate, in the bird's-eye view image; estimating a distance between the mobile system and the detected person based on the first foot pixel coordinate; and estimating a height of the detected person based on the distance and the head pixel coordinate.