BIA-Brukerstyrt innovasjonsarena

FREM, what is it and what has been done? The goal has been to develop and renew equipment and processes for design so that the limit for profitable automation is lower and the quality is higher. One challenge is that robots do not have advanced human senses and will often have problems with assembly of electromechanical equipment. Therefore, we have focused on product design and robotic assembly as well as the use of artificial senses for robots. Barco and Norautron have mapped and analyzed existing and new products. New set of rules related to Design For Automated Assembly (DFAA) version 1.0 have been introduced and work is continuing on an improved version 2.0. For this, Barco has developed a flexible robotic test setup that can be adapted to different processes and adjusted quickly. This provides important feedback to the development process of new products. The use of the design rules has resulted in many good changes and new concepts at Barco and Norautron use this as a tool to start collaboration with customers in an earlier phase for a better overall solution. Barco and Norautron have flexible robot solutions where setups, apart from small adaptations, are largely reused on new products. Barco use it to assemble light sources that require precision and cleanliness in addition to being a repetitive process. Norautron has ABB YuMi robots and now performs automated placement and feeding of hole-mounted components, automated soldering and works with several new concepts. Industry 4.0 opens for the development of new technology that can be used in robotics to increase flexibility and reduce conversion time. Robots must understand what they "see and feel" and perform safe actions based on this. Sensors for what the robot "feels" are well known, but what it "sees" and what kind of technology is suitable in different conditions is not as well known. The project has worked with both 2D and 3D machine vision. A demonstrator for this has been set up at SINTEF Manufacturing. Initial use was the validation of precision for feeding parts in assembly lines. Later, it has been used to explore opportunities for quality inspection. A step towards more flexible lines / cells is indeed 3D vision. Work has been done here with both 3D metrology and FLIR. Combined with CAD tools, this will be able to provide faster adjustment. 2D vision in combination with image recognition and deep learning can also be used for this in many contexts. These are areas where the development of both hardware and software is now progressing rapidly. There are also processes where machine vision cannot be used or is necessary. Here, solutions for force control are just as important. A concept for such an automatic assembly has been developed and tested with good results. The project has also focused on the use of automation in quality assurance. The aforementioned demonstrator has been used to look at various methods of inspecting assemblies unsuitable for automation. The use of 3D views on pure mechanical assemblies works relatively well. Here, choosing the right equipment in terms of resolution of details and type of assembly you are looking at is important. Programming of the process is done almost exclusively based on an annotated 3D model. Barco has seen challenges with such inspection in assemblies with optical components. These are difficult to see with 3D / 2D vision as the components are transparent. Here, tests have been done, with good results, where you illuminate the assembly with a laser and read the position of the reflectors with normal 2D vision. A concept study has also been done for the use of 2D views on an optical mechanical assembly where one looks at how light is formed through the assembly and by means of image recognition reveals errors. This is presented for production in Barco. Common to these solutions is a desire to detect human or component defects during assembly and raise the quality. Norautron has looked at the use of 2D vision in combination with deep learning and the use of the right type and shape of lighting as part of a cell that will detect faults or missing components on circuit boards. The intention of the project has been to achieve cost-effective production of electromechanical products in Norway. Design solutions adapted to the industry have been and are being developed. Same for solutions that utilize artificial sensory apparatus and give robots an understanding of what action to perform when a given object is present. The degree of complexity in electromechanical production will lead to humans having to be present and collaborate with robots and the results of the project show that this is possible. We see that the project has led ?FREM? to results that have increased profitability and quality of the products, which in turn raises competitiveness. It will also be a significant innovation in the industry and an important contribution to the research front in Industry 4.0

Deltagerne i prosjektet har bygget kompetanse og funnet gode løsninger innenfor følgende områder: - produktdesign for automasjon (DFAA) - menneske robot samarbeid (der helautomatstasjon ikke er aktuelt) - fleksible robot celler med enkel omstilling mellom oppgaver Prosjektet har med dette bidratt til å senke kostnader knyttet til produksjon, samt øke kvaliteten i flere av produktene til bedriftene som har deltatt. Deltagerne har delt sine funn og erfaringer videre igjennom presentasjoner for og i samarbeid med andre bedrifter.

Automatisert montasje av produkter som består både av elektronikk og mekanikk er uvanlig i dagens industri. Dette er hovedsakelig på grunn av produktenes komplekse oppbygging, variasjon og relativt lavt antall som vil selges i løpet av produktets levetid. Slike produkter består som regel av en kombinasjon av små og relativt store deler. En del av de kan for eksempel bestå av ett til flere kretskort, kabler og kontakter som ofte skal loddes, skrues eller limes. Noen produkter består av flere små delprodukter som monteres sammen til større sluttprodukter. Disse oppgavene krever normalt nøye inspeksjon og evnen til å føle seg frem. Gode metoder for å lære en robot til å utføre disse typisk menneskelige montasjeoppgavene er ikke etablert i dagens industri. Det snakkes i dag mye om den fjerde industrielle revolusjonen, noe som betyr at maskiner vil i enda større grad kommunisere seg i mellom, og gjerne med selve produktet som blir produsert. Dette åpner muligheten for svært mange nye og mer fleksible robotløsninger. Disse kan lettere tilpasses og skaleres etter produkttype og etterspørsel i markedet. Roboter vil være avhengig av avansert kamera- og sensorteknologi. Dette vil hjelpe roboter til å forstå omgivelsene, utføre trygge valg og de kan dermed samarbeide med mennesker uten fare for liv og helse. Et samarbeid mellom menneske og robot vil gjøre det mulig å produsere elektromekaniske produkter i Norge og være lønnsomt og konkurransedyktig i forhold til lavkostland. Dette er et veldig viktig steg for bedrifter både for å bli bedre på produktdesign og produksjon, og for samfunnet generelt vil det øke kompetansenivået, industriutvikling og sysselsetting. Automatisering av Norsk produksjon vil også bidra positivt for miljøet ved å redusere CO2 utslipp gjennom transport, noe som potensielt vil gjøre produktene mer attraktive for kundene. Det kan også øke antallet kortreiste produkter som faktisk blir lønnsomme å produsere i Norge.