Image-Based Ergonomic Risk Root Cause Detection and Solution Guidance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods struggle to automatically identify ergonomic risks and their corresponding solutions in industrial settings due to the scarcity of ergonomic experts, leading to challenges in preventing work-related musculoskeletal disorders (WMSDs) affecting millions of workers.
Innovation Solution
An image-based system utilizing advanced computer vision and natural language processing techniques, including pre-trained vision-language models and bidirectional transformers, to analyze worker images and generate captions indicating ergonomic risks and solutions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If ergonomic experts are used to identify ergonomic risks and solutions, then identification accuracy is improved, but system cost and complexity increase due to scarcity of experts
Solution Approach 1:
The patent replaces the mechanical system of human expert assessment with an automated computer vision system using deep learning models. The system processes images through multiple neural network components (backbone network, detection head, pose estimation modules) to automatically identify ergonomic risks, substituting human expertise with algorithmic analysis while maintaining identification accuracy.
Solution Approach 2:
The system enables self-service by allowing organizations to independently assess ergonomic risks without requiring external expert intervention. The automated pipeline processes workplace images and generates risk assessments with actionable solutions, empowering users to conduct their own ergonomic evaluations and implement corrections without relying on scarce expert resources.
2Loss of information
If expert visits to workplaces are conducted, then comprehensive ergonomic assessment is improved, but time consumption and cost increase
Solution Approach 1:
The system enables continuous ergonomic monitoring by processing images in real-time as they are captured from workplace environments. Unlike periodic expert visits, the automated system can continuously analyze worker postures, equipment configurations, and environmental factors, providing ongoing assessment without interruption to work operations.
Solution Approach 2:
The system performs preliminary ergonomic assessments automatically before expert intervention is needed. By pre-processing images and identifying obvious ergonomic risks, the system prepares initial findings and recommendations that can be reviewed by experts only when necessary, reducing the frequency and time of expert visits while maintaining assessment completeness.
3Measurement precision
If manual ergonomic assessment methods are used, then solution accuracy is improved, but productivity decreases due to inability to scale to all workplaces
Solution Approach 1:
The patent creates a universal system that can assess multiple types of ergonomic risks across diverse workplace settings using the same automated pipeline. The model handles various tasks including overhead work, lifting, pushing/pulling, and repetitive motions, making it applicable to manufacturing, construction, logistics, and other industries without requiring task-specific expert assessments.
Solution Approach 2:
The system maintains solution accuracy while scaling by changing the operational parameters from manual expert review to automated machine learning inference. The deep learning models, trained on comprehensive datasets of ergonomic scenarios, achieve expert-level accuracy in identifying risks and generating solutions, enabling simultaneous assessment of numerous workplaces without sacrificing precision.
Data Source
AI summary
Disclosed herein is an image-based system configured to identify root causes of industrial ergonomic risks and their corresponding solutions. An example system comprises a computing device configured to encode an image of a worker performing a work task to generate an embedding vector, transmit the embedding vector to an image-grounded text decoder, while generating first tokens to instruct the decoder to generate a first sentence indicating a root cause of an ergonomic risk identified in the image, compute first relative sensitivity scores relating to the first tokens and extracted image features, generate second tokens of the first sentence based on the first relative sensitivity scores, while generating third tokens to instruct a text decoder to generate a second sentence indicating a solution to the ergonomic risk, calculate second relative sensitivity scores relating to the second and third tokens, and generate the first and second sentences accordingly.


