Generative AI Work Support for On-Site Know-How Guidance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing work support technologies for plants and infrastructure are complex and time-consuming, requiring detailed keyword inputs and lacking immediate on-site support for inquiries, leading to potential delays in solving issues.
Innovation Solution
A work support system utilizing a generative AI model to generate replies based on acquired know-how information, allowing for simplified and practical support through a wearable AR terminal that records and processes inquiries from unskilled workers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If conventional keyword-based know-how extraction is used, then know-how information can be extracted from skilled technical persons, but the process becomes complicated and time-consuming requiring detailed 5W1H keyword inputs
Solution Approach 1:
The patent replaces the manual mechanical process of keyword input with automatic image recognition and AI-based question generation. The system automatically captures images at work sites, recognizes objects and situations, generates appropriate questions without requiring 5W1H keyword input, and extracts know-how information automatically, thereby eliminating the time-consuming manual keyword entry process while preserving complete know-how information.
Solution Approach 2:
The system enables self-service by allowing the work support apparatus to automatically perform know-how extraction without requiring skilled technical persons to manually input keywords. The apparatus autonomously captures images, generates questions based on recognized content, and extracts answers, making the process self-sufficient and eliminating the need for complex user interaction.
2Reliability
If conventional keyword-based inquiry system is used, then know-how information can be retrieved, but immediate on-site support is not available causing delays in solving work issues
Solution Approach 1:
The system performs preliminary actions by pre-capturing images and pre-generating questions at the work site before actual inquiries arise. The apparatus continuously captures work site images and prepares potential questions in advance, so when an inquiry occurs, the system can immediately retrieve relevant know-how information without delay, ensuring both reliability and speed of response.
Solution Approach 2:
The patent replaces the manual process of forming and inputting inquiry keywords with automatic image-based question generation. The system captures images of the actual work situation, automatically generates appropriate inquiry questions, and retrieves answers immediately, eliminating the time delay associated with manual keyword formulation and enabling instant on-site support.
3Measurement precision
If detailed keyword input is required for work support, then precise know-how information can be extracted, but the operation becomes complex and difficult to use at work sites
Solution Approach 1:
The patent replaces the complex manual process of inputting detailed 5W1H keywords with automatic image recognition and AI-based question generation. The system captures images of the work situation, automatically recognizes objects and contexts, generates precise questions without requiring user knowledge of keyword structures, and retrieves accurate know-how information, thereby maintaining precision while dramatically improving ease of operation.
Solution Approach 2:
The system uses image copying to capture and analyze the actual work site situation. Instead of requiring users to describe complex situations through keywords, the apparatus copies the visual state of the work environment, automatically interprets it, and generates appropriate questions, making the system as easy to use as taking a picture while maintaining high precision in know-how retrieval.
Data Source
Figure 1
Figure 2~3
Figure 4~5
AI summary
Provided is a control unit (103) configured to acquire know-how information on work related to at least a plant (P1) or infrastructure from a worker terminal (10) that is used for the work performed by a person engaged in the work in a work site, accumulate the know-how information, generate, when inquiry information related to the work is accepted, based on the inquiry information, a question that requests generative AI that uses a large language model to generate a reply produced using the know-how information, generate presentation information that is presented to the worker terminal (10) corresponding to an inquiry source based on the reply to the question received from the generative AI, and transmit the presentation information to the worker terminal (10).