Workspace Optical Monitoring With Access-Area AI Verification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing person detection systems in industrial work areas face challenges in accurately recognizing individuals with varying external characteristics, such as hair, skin color, height, and posture, especially when influenced by clothing or protective equipment, due to incomplete test data and changing assumptions, leading to potential false positives or false negatives.
Innovation Solution
An online testing system is implemented where every person entering the work area must pass through an access area with an AI-based person detector, which continuously evaluates and verifies the position information using multiple sensors, ensuring accurate detection by comparing position data from the access area detector with additional sensors like light grids or foot mats, and only allowing access if the deviation is within a predetermined level, thus ensuring safe and precise recognition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If test data is used to determine detection threshold, then detection accuracy is optimized for test cases, but results cannot be transferred to practical use if test data does not match real data
Solution Approach 1:
The system performs preliminary detection in the access area before the person enters the workspace. This preliminary action verifies the detector's ability to recognize the person under controlled conditions, establishing baseline performance. Only after successful preliminary detection does the system allow the person to enter the monitored workspace, ensuring detection reliability is validated in advance.
Solution Approach 2:
The system continuously monitors detection results in both access area and workspace, comparing positions and consistency of detections. This feedback loop allows the system to adapt to varying conditions and detect patterns that may indicate detector degradation or environmental changes affecting generalizability.
2Measurement precision
If multiple sensors are used to verify position information, then detection accuracy is improved, but system complexity increases
Solution Approach 1:
The system divides the detection process into two independent stages: access area detection and workspace detection. Each stage has its own detector and confidence threshold, allowing optimization for different requirements. The access area detector uses a higher threshold to ensure only clearly detected persons trigger access control, while the workspace detector uses a lower threshold to ensure all persons are captured for safety monitoring.
Solution Approach 2:
The access area serves as an intermediary zone between the outside environment and the monitored workspace. This intermediary space allows for preliminary verification of detection systems without exposing the main workspace to potential failures. The access area detector acts as a mediator, filtering and verifying persons before they enter the critical workspace area.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
To implement a system (1) for the optical monitoring of a workroom (AR), it is proposed to place an access area (ZB) in front of the workroom (AR), which is designed so that it must first be passed through in order to reach the access (T), wherein the access area (ZB) has an access area detector (1a) which is functionally similar to the workroom detector (2a) and is also designed - with an imaging sensor for recording digital image files or point clouds and - with a test evaluation unit (1b), which is designed to evaluate the image files or the point clouds in the same way as provided in the evaluation unit (2b) of the workroom detector (2a), and to output a position information (PZ) of a person (M) detected in the access area (ZB).