University West uses cookies to ensure you get the best experience on our website. By continued use of the site, you agree to the use of cookies. Read more about cookies
Antingen stödjer din webbläsare inte javascript, eller är javascript inaktiverat. Denna webbplats fungerar bäst om du aktiverar javascript.
The project aims at performing an in-depth and systematic needs analysis of the error recovery process and knowledge integration demands between involved human stakeholders at the partner companies and based on that design a prototype for an AI&AR-based error recovery support system for operators.
Husqvarna is one of the companies participating in Restart. Some of the participants at the company's production facility in Brastad are (from left) Sandra Mattsson, Jörgen Jensson and Alexandra Hallqvist Lehtinen.
In automated production processes, unplanned stops occur due to tool failure, material defects, unmanaged program situations or human interventions, etc. Restart of production defined as an error recovery process is a complex problem, involving the three phases: error detection, error diagnoses, and resetting the system to a restartable state. Short, long and unplanned production stops are expensive, unavoidable and cannot be predicted and therefore cannot be handled in advance in system and process models. There is no one-size-fits-all solution, rather each situation must be handled by the operator or maintenance personnel.
The envisioned system prototype is expected to facilitate in particular the error diagnosis aspect of the recovery process through hands-on AR guidance tuned to the specific situation and skill level of the operator. The guidance will be based on error cause diagnosis strategies prerecorded by expert maintenance personnel, strategies which the system automatically links to the current unexpected shop floor situation through AI. For recording and representing these strategies, various technologies will be investigated including state-of-the art AI activity sequencing approaches, eye tracking, and AR-based subtle guidance approaches.
2022 - 2024