INDUSTRY SCENARIOS – TEXTILE PRODUCTION

Scenario 1: AI-Assisted On-The-Job Training for Machine Operators

This scenario will address the shortage of qualified labor force with an innovative solution that will reduce the time and cost of machine operator trainings. COALA will identify suitable textile manufacturing activities and machinery to demonstrate the solution considering, amongst others, the following factors: factory noise, needed worker capabilities, variety of production machines, expected product and process quality, time-criticality, and worker safety. A key goal of this scenario is to maintain the worker’s autonomy instead of promoting the unquestioned execution of instructions. Piacenza will provide the testbed from raw materials to fabric delivery and, to clothing sale to consumers to prove the skills and qualification of the trained employees. 

Baseline. Piacenza realizes all its production in the Italian textile district of Biella. The company performs its production organization fully in Italy and it manages all aspects from raw materials to finished fabric, except spinning phase, which its sub suppliers perform. 

Piacenza’s production strategy bases on the strict and integrated control of production. Piacenza keeps internal those production phases which give an added-value perceived by the customer (e.g. raw material acquisition, finishing, and inspection) or a production flexibility and cost advantage (e.g. weaving, and yarn dyeing). A limited number of specialized sub-suppliers take the remaining ones based on a long-term business relationship that helps to preserve quality and service.

The textile workers focused in this scenario are the machine operators that use their skills and knowledge to prepare and supervise machine operations under strict consideration of the product quality. Using COALA solution. Piacenza expects that the solution will allow machine operators to request advices, explanations, and other information via the digital assistant running on a mobile device. A custom dialog modeland the cognitive advisor service will identify the worker’s current skill level, and advice how to prepare and operate a machine. It uses the PA service to access shop floor context information, especially related to machine parameters, machine performance, and production planning information. The assistant will adapt the advising behavior with progressing skill increases and under consideration of human feedback about the learning experience, for instance it will be more instructional in the early phase and more supervising in later phases. The explanation engine extension will demonstrate prototypically how workers can ask “why” questions about the advices. The anonymization service will obfuscate personal worker data to minimize ethical risks.
 
Using the COALA solution will allow textile companies to employ less skilled workers whose training would otherwise be too costly – effectively, this will increase the supply of potential labor force. The company also expects that the training support will reduce defects created during the manufacturing process due to human errors.

Scenario 2: Development Of AI Competencies of Manufacturing Workers

One of the key problems in luxury fashion is the shortage of qualified labour force that can produce high quality and high value textiles. To face this problem a consortium of companies of the Biella district (including Zegna and Loro Piana) and the local textile school Città Studi created an academia to train the new generation of workers to face the requirements of the market. This scenario focuses on the development and demonstration of a didactic concept for professional textile workers that develops their competencies in working with AI-based systems, such as the digital assistant developed in scenario 1. 
 
Regional innovation infrastructures, such as DIHs, can play an important role in strengthening the AI competencies of workers because they support local companies in evaluating and using new technologies. STAM, BIBA, ICCS, and TUD, for instance, operate DIHs and will use their infrastructures to elaborate opportunities to transfer COALA results to other domains.

Baseline. Today the CIT academy courses are structured with a balance between the theoretical part and the practical part. The practical part is made in collaboration with some companies that host our students and make some production machines available for education; here students can learn how to work on a specific machine, how machine works and what are the possible problems that can occur during work, with the supervision of one or more teacher; to do this we can also simulate issues or bad working conditions to help students to better understand these issues and how to fix them. The practical period lasts at least 120 hours but can be extended if necessary. In order to do this practical part we have signed agreements with some companies that rent us their plants and/or machines (e.g. looms, and ring spinning frames) for the practical periods; with these plants and buying different fibers and/or yarns suitable for didactical purposes we are able to offer to our students a wide range of user-cases that covers the most parts of the needs that our companies requires from our courses.
STAM, a manufacturing-focused DIH in Genova (Italy), has expertise in application experiments. They are connected to a variety of potential future COALA solution users. Human-AI collaboration is one of their topics and strengthening regional companies in this field an important goal.

Using COALA solution. The main innovation related to this scenario is the didactic concept developed by ITB and CIT to educate textile machine operators. CIT will use this new concept and its learning materials to educate machine operators regarding the opportunities, challenges, and risks when collaborating with AI. The concept will use the COALA assistant as an example for an AI-system in this context.

STAM’s role as a DIH will be to further exploit and spread the concept and the COALA solution in the region. They will reach for potential users outside the textile sector as well and coordinate their efforts with the other DIHs in the COALA consortium (T5.4).
Because of the different fabric manufacturing needs, some of which supported by old technology looms and some others requiring new ones (such as the brand new jacquard looms bought recently by Piacenza) in the same production facility it is very common to find looms of different ages, provided by different OEMs with various PLC. The integration with all of them would be very time expensive and could not be justified by the provided ROI. In order to provide an easy to adopt and affordable solution, especially for SMEs which are >90% of textile companies, COALA will adopt a progressive approach: instead of a direct integration with the machine each loom, we will provide the machines, for first, with a barcode based on which the mobile device supporting the operator will be able to access the machine specifications to support the training and the operations of the worker on that specific loom. The COALA solution will carry out the training of the mobile based on the map of the looms provided by each textile company in advance. The COALA solution will also collects the feedback of the operators as regards the use of the specific machine, in order to support OEMs in the ergonomic design enhancement. 
For the most advanced looms, available in the Piacenza testbed, we will explore the direct integration with the machine PLC to support the exploitation of the COALA solution, such as the direct collection of machinery tuning by the operator in relation with support provided by COALA to collect an additional dataset to train future training and support models. We will provide a cost/benefit analysis during the exploitation to verify the affordability for SMEs (T7.2).