The AI-PRISM project is oriented towards a human-centred AI-based solutions ecosystem targeted to manufacturing scenarios with tasks difficult to automate and where speed and versatility are essential. To facilitate the assessment of the performance, transferability, scalability, and large-scale deployment of AI-PRISM solutions, we will conduct research under real operational environments in five pilots involving key manufacturing sectors: furniture, food/beverage, built-in appliances, Electronics and one generic discrete manufacturing.
Regarding the electronics manufacturing demonstrator, researchers from Cranfield University and PIAP visited VIGO Photonics facilities in Vigo, Poland. In this first approach, all the attendees made a tour around the company, visiting each department and, more concretely, learning individual processes performed in the production department. The use case of this pilot will take place in a clean room, and the shop floor has controlled environmental parameters.
This use case focuses on a solution that will feature automatic positioning of the electronic component against the wire, to be glued, with the support of electromechanical or/and pneumatic effectors. Furthermore, they will have input from an AI-enhanced vision system that will recognize the appropriate place to attach the wire to the electronic component based on the shape, colour and additional markers if needed. Moreover, AI algorithms will adapt to the differences in components in case of their exchange. Finally, the operator will teach the robot to position the elements to be glued together correctly. This will be the initial role of the operator, who, at a later stage, will still be in the loop for the task, mainly for its most critical phase – to confirm the correct positioning of the elements but also to perform supporting actions, such as glue application.
Collecting data and verbalizing steps for the task analysis
During their visit, Cranfield University’s partner collected eye-tracker data while two operators performed the assembly (one with four years of experience and a novice) at the assembly station. While collecting the eye-tracking data, operators verbalized their steps for the task analysis. Later, via a short interview, they discussed the most challenging aspects of the task, common errors, solutions/mitigations, and what potential changes would happen to be needed to make the process easier. In parallel, the AI-PRISM team documented all the experience with photos and had technical discussions.
Before AI-PRISM, positioning and glueing used XY adjustable tables with position control under a microscope and support of measuring software. Due to the differences between holders for wires and the wires’ curvature, the positioning procedure is repeated for every pair of components. The manual positioning lasts a few minutes. It is a bottleneck of the production process, as the processing machines that work on glued components sometimes need to be used for the next delivery.
Researchers studied three areas of interest: monitor used for calibrating the microscope, microscope, and the component to see how operators approached and completed the process. The novice operator had a more spread fixation on the environment and broader dwell range than the experienced operator, who focused only on areas of the AOIs. The numerical comparison for the dwell times of these two operators indicated that although the novice operator had a more considerable dwell time in all three AIOs, the difference was the most evident in the microscope AOI. Interestingly, the discussion with the operators allowed us to determine that awareness and mental workload might impact the assembly situation depending on the automation steps introduced.
A new robotic solution to overcome technical barriers
The studied processes are based on the worker’s experience, and automating this work using conventional methods is difficult as the components may change in size. VIGO Photonics has yet to find a robotic solution to overcome the barriers of the process. With AI-PRISM, they expect improved production reduction in the time spent on training workers and performance while increasing assembly precision. Ultimately, they want to reduce environmental pollution with waste from producing semiconductors.
Some of the reasoning problems to solve at the agent (robot) level to improve manufacturing agility is the components handling based on AI vision output to achieve components perpendicularity, surface parallelism and precision alignment. The challenge will be to provide an easy reconfiguration to identify different chip mask patterns and wire/rod/shaft diameters and remote monitoring of the process for the VIGO’s engineering team. The AI-PRISM team is excited to see where this project will take us. We will apply the same methods and technologies in the other industries that conform to our five pilot use cases, requiring high manual dexterity and expertise.
Do you want to learn more about AI-PRISM and the team behind the project? Subscribe to our newsletter and follow us on LinkedIn and Twitter, so you don’t miss a thing!