Periodic Reporting for period 2 - COALA (COgnitive Assisted agile manufacturing for a LAbor force supported by trustworthy Artificial Intelligence)
Berichtszeitraum: 2022-04-01 bis 2023-09-30
Knowledge-intensive manufacturing processes require qualified employees. Their training is very time-consuming and cost-intensive for companies. The current shortage of skilled workers, for example, in sectors like textile manufacturing, makes this even more critical. Another competitive challenge for the manufacturing industry is the constantly shortening production cycles and the increasing variety of products.
COALA addresses these challenges through the innovative design and development of a human-centred and trustworthy voice-enabled Digital Intelligent Assistant (DIA) for the manufacturing industry that provides a more proactive and pragmatic approach to support operative situations characterised by cognitive load, time pressure, and little or zero tolerance for quality issues. COALA helps shape the complementarity in the collaboration between the AI-based assistant and the human so that 1) the AI can take over time-consuming and stressful tasks reliably and credibly, while 2) the human can focus on understanding and problem-solving in complex, knowledge-intensive situations.
COALA AI-based digital intelligence solutions will make manufacturing companies more sustainable through the best use of digital technologies and the development of employees' skills, allowing companies to increase their production process and product quality.
COALA aims to improve workers' social skills by developing an AI-focused education and training concept to help workers build competencies in human-AI collaboration. We developed a didactic concept to help the workers improve their professional, personal, learning and methodological competencies and strengthen their control and responsibility while using the COALA solution.
COALA Digital Intelligent Assistant has a conversational user interface that allows worker interaction by voice or text and visualises shop floor data. COALA's feature for on-the-job training aims to provide advice and support to novice operators in their learning and working activities while reconfiguring and operating production lines. The training approach focuses on introducing changes through the advice of and dialogue with the assistant, resulting in measurable changes in the operator's behaviour.
COALA's feature Augmented Manufacturing Analytics aims to support workers in tasks that formerly required expertise in data science. Digital voice assistants and chatbots interact intelligently with humans via natural language – these AI-driven programs guide their users through complex analytics processes and support the non-data scientist workers in performing knowledge-intensive activities that significantly impact product and process quality, allowing the workers to utilise and customise data analytics during product quality tests.
The combination of all these efforts is expected to have a positive economic impact in the manufacturing sector, among others, in improving the productivity of workers and new workers and reducing poor quality costs.
COALA addresses these challenges through the innovative design and development of a human-centred and trustworthy voice-enabled Digital Intelligent Assistant (DIA) for the manufacturing industry that provides a more proactive and pragmatic approach to support operative situations characterised by cognitive load, time pressure, and little or zero tolerance for quality issues. COALA helps shape the complementarity in the collaboration between the AI-based assistant and the human so that 1) the AI can take over time-consuming and stressful tasks reliably and credibly, while 2) the human can focus on understanding and problem-solving in complex, knowledge-intensive situations.
COALA AI-based digital intelligence solutions will make manufacturing companies more sustainable through the best use of digital technologies and the development of employees' skills, allowing companies to increase their production process and product quality.
COALA aims to improve workers' social skills by developing an AI-focused education and training concept to help workers build competencies in human-AI collaboration. We developed a didactic concept to help the workers improve their professional, personal, learning and methodological competencies and strengthen their control and responsibility while using the COALA solution.
COALA Digital Intelligent Assistant has a conversational user interface that allows worker interaction by voice or text and visualises shop floor data. COALA's feature for on-the-job training aims to provide advice and support to novice operators in their learning and working activities while reconfiguring and operating production lines. The training approach focuses on introducing changes through the advice of and dialogue with the assistant, resulting in measurable changes in the operator's behaviour.
COALA's feature Augmented Manufacturing Analytics aims to support workers in tasks that formerly required expertise in data science. Digital voice assistants and chatbots interact intelligently with humans via natural language – these AI-driven programs guide their users through complex analytics processes and support the non-data scientist workers in performing knowledge-intensive activities that significantly impact product and process quality, allowing the workers to utilise and customise data analytics during product quality tests.
The combination of all these efforts is expected to have a positive economic impact in the manufacturing sector, among others, in improving the productivity of workers and new workers and reducing poor quality costs.
The main technical development activities started with eliciting the initial technical, social, and ethical requirements and defining the use case scenarios of the three business cases (textile, white goods, and detergent productions). By M9, we specified the COALA system architecture and its interfaces. Furthermore, we developed the initial prototype of the core demonstrator, COALA Digital Intelligent Assistant (DIA), and some key components/services, such as Prescriptive Quality Analytics and Cognitive Advisor services.
Until M15, we developed the second prototype of DIA core demonstrator components, along with the prototype of some other services such as Product Avatar and Data Collection service, Data Anonymisation service, Dialogue and Interface Localization service. We also identified the first principles for the Why Engine as a new, experimental solution component that allows the assistant to answer "why" questions. Accordingly, we developed an integration and deployment plan for the COALA components/services into the COALA Solution. Furthermore, we defined a testing and evaluation plan to verify the proper functioning and performance of the integrated COALA Solution. We use the ALTAI (Assessment List for Trustworthy Artificial Intelligence) tool to perform the first self-assessment of the COALA components to ensure compliance with the Trustworthy Artificial Intelligence aspect's requirements and to allow early identification of any risks that may arise during the development and integration activities. In the non-technical aspects, we developed the first version of the didactic concept for the factory workers and the change management approach.
By M18, all technical components are in their final configuration. We started preparing the first validation tests with the end users. We deployed two COALA main demonstrators: (1) "DIA for Augmented Manufacturing Analytics" in the White Goods Production (Whirlpool) use case, and (2) "DIA for on-the-job factory worker training" in the Detergent Production (Diversey) use case and Textile Production (CITTA/PIACENZA) use case.
Between M18 and M24, the project's main focus is on further development, deployment and integration of the COALA components/services and setup of the first tests of the COALA demonstrators with the end-users. We analysed the system usability and user experience feedback and then adjusted and refined the demonstrators accordingly. From M25, we intensified the integration of live data from the end-user facilities and finalised the translation of the dialogue into the local languages, Dutch and Italian. The final evaluation took place between M28 and M36.
In the evaluation framework, we engaged the regional innovation infrastructures, such as Digital Innovation Hubs (DIHs), to support local companies in evaluating and using new technologies since they play an essential role in strengthening the AI competencies of workers. Specific contributions to Europe's AI communities in manufacturing have been defined and carried out. We established COALA's presence in the AI4Europe community. We actively disseminated and communicated COALA in more than 30 conferences and 20 industrial events. The project has developed 11 Innovative Exploitable Assets.
Until M15, we developed the second prototype of DIA core demonstrator components, along with the prototype of some other services such as Product Avatar and Data Collection service, Data Anonymisation service, Dialogue and Interface Localization service. We also identified the first principles for the Why Engine as a new, experimental solution component that allows the assistant to answer "why" questions. Accordingly, we developed an integration and deployment plan for the COALA components/services into the COALA Solution. Furthermore, we defined a testing and evaluation plan to verify the proper functioning and performance of the integrated COALA Solution. We use the ALTAI (Assessment List for Trustworthy Artificial Intelligence) tool to perform the first self-assessment of the COALA components to ensure compliance with the Trustworthy Artificial Intelligence aspect's requirements and to allow early identification of any risks that may arise during the development and integration activities. In the non-technical aspects, we developed the first version of the didactic concept for the factory workers and the change management approach.
By M18, all technical components are in their final configuration. We started preparing the first validation tests with the end users. We deployed two COALA main demonstrators: (1) "DIA for Augmented Manufacturing Analytics" in the White Goods Production (Whirlpool) use case, and (2) "DIA for on-the-job factory worker training" in the Detergent Production (Diversey) use case and Textile Production (CITTA/PIACENZA) use case.
Between M18 and M24, the project's main focus is on further development, deployment and integration of the COALA components/services and setup of the first tests of the COALA demonstrators with the end-users. We analysed the system usability and user experience feedback and then adjusted and refined the demonstrators accordingly. From M25, we intensified the integration of live data from the end-user facilities and finalised the translation of the dialogue into the local languages, Dutch and Italian. The final evaluation took place between M28 and M36.
In the evaluation framework, we engaged the regional innovation infrastructures, such as Digital Innovation Hubs (DIHs), to support local companies in evaluating and using new technologies since they play an essential role in strengthening the AI competencies of workers. Specific contributions to Europe's AI communities in manufacturing have been defined and carried out. We established COALA's presence in the AI4Europe community. We actively disseminated and communicated COALA in more than 30 conferences and 20 industrial events. The project has developed 11 Innovative Exploitable Assets.
The expected impact of COALA on the industry will cover two major levers:
(1) Improvement of the production performance: agile production processes and improved quality of products and processes.
(2) Improvement of human integration in the production system of manufacturing companies through:
- Education and on-job training will help workers to improve their skills and AI-related competencies.
- Cognitive advisor, Prescriptive Quality Analytics, and Product Avatar functionalities will: guide workers to manage problems they face, reduce intrinsic and extraneous cognitive workloads and the resulting stress, increase individual and shared situation awareness,increase in speed of gaining situation awareness about issues at hand, and promote germane cognitive load by redirecting their attention to cognitive processes that are directly relevant to the construction of schemas (e.g. understanding of the functioning of the process, and factors impacting quality).
- The WHY engine will provide some explanations about the DIA predictions and advices to support decision and sense making, and to reassure workers on the way AI elaborates conclusions (trustworthy AI).
(1) Improvement of the production performance: agile production processes and improved quality of products and processes.
(2) Improvement of human integration in the production system of manufacturing companies through:
- Education and on-job training will help workers to improve their skills and AI-related competencies.
- Cognitive advisor, Prescriptive Quality Analytics, and Product Avatar functionalities will: guide workers to manage problems they face, reduce intrinsic and extraneous cognitive workloads and the resulting stress, increase individual and shared situation awareness,increase in speed of gaining situation awareness about issues at hand, and promote germane cognitive load by redirecting their attention to cognitive processes that are directly relevant to the construction of schemas (e.g. understanding of the functioning of the process, and factors impacting quality).
- The WHY engine will provide some explanations about the DIA predictions and advices to support decision and sense making, and to reassure workers on the way AI elaborates conclusions (trustworthy AI).