Predictive Industrial Safety: An AI framework for industrial wood saw safety applications
| Reference number | |
| Coordinator | ManoMotion AB |
| Funding from Vinnova | SEK 2 999 036 |
| Project duration | October 2021 - December 2022 |
| Status | Completed |
| Venture | Joint R&D projects for small and medium-sized enterprises in Sweden-Germany |
| Call | German-Swedish Call for joint R&D projects by Small and Medium-sized Enterprises spring 2021 |
Important results from the project
In this project we have designed and developed an AI framework to extend the level of industrial safety by understanding and predicting human behavior and thereby discovering dangerous situations before happening. The AI system extracts different information from humans and the working environment to predict incidents and provide safety signals, warn the operators or stop the machines on time. Thereby human-machine interaction will be done in a significantly safer, smarter, and more efficient manner leading to a significant reduction in injuries and less down time for machines.
Expected long term effects
The AI framework implemented in this project for predictive safety combines real time information from hand tracking, gesture analysis, human motion and environment to generate smart safety signals while operators interact with machines, and avoid dangerous situations. The system is designed with the real data from industrial use cases, and evaluated in various environments with different human subjects to ensure a robust performance. Based on the outcomes, demonstrations, and feedback from the industrial partners, a business model and IPR strategy have been developed.
Approach and implementation
The developed AI framework for predictive industrial safety combines the visual input from a set of cameras looking at human operators within the interaction range of the industrial machines. The system extracts relevant information regarding the human motion, hand tracking & gestures, face & attention and environment to predict the human intent while interacting with machines. The framework generates smart warning and emergency signals to avoid dangerous situations, human injuries and damages to the machine. The system performance has been evaluated by industrial partners.