Human Centred Artificial Intelligence for Agile Manufacturing
The STAR project operates in the IBER (Iber-Oleff SA) value chain with the objective of boosting scientific and technological capacities and competences, and carrying out research and development activities whose results will allow IBER to expand and strengthen its value chain in the fields of products, processes and services, namely:
- Streamline the use of agile and adaptive manufacturing systems by incorporating advanced online process sensing & monitoring systems and different parameters and variables into the manufacturing process and products, including non-destructive testing techniques, to measure dimensions and defects, with contactless and agile technologies;
- Promoting greater flexibility of the production process through a new and innovative integrated management concept based on the flexibility, versatility and synchronization of production cells, using dedicated information systems.
A key aspect of the IBER pilot project is the manufacture of a customized product in an agile production system that allows for greater efficiency, speed, operation and maintenance monitoring. The concept of this pilot project is human-centred and will be supported at the production management decision-making level by a future artificial intelligence platform, connected to the existing digital factory.
It is expected that the future artificial intelligence platform will allow to accelerate the decisions at the production management level, through simulations of the production processes in real time, reaching the desired flexibility of them. An agile production system will allow rapid reconfiguration of production processes to accommodate sudden customer orders, produce a highly configurable product with quality and in the shortest possible time frame.
Advanced online process sensing & monitoring systems and different parameters and variables into the manufacturing process and products, including non-destructive testing techniques, to measure dimensions and defects, with contactless and agile technologies should be incorporated on these adaptive manufacturing systems. The maximization of productivity due to the linearization will have important consequences in the reduction of the work peaks and consequently in the reduction of the number of extra work hours, as well as the reduction of the production down times.
Throughout this process the human factor plays a predominant role, given that our production processes are not fully automated (nor are they intended to be). Nowadays the training of human resources in the various production processes is planned and carried out in advance, in accordance with appropriate operational methods. The flexibility of human resources training, to accommodate the agile production mentioned above, will be achieved with the help of the future artificial intelligence platform, which will contribute, on the one hand, to the improvement of operative methods and consequently the reduction of human errors associated with learning and assimilation, and, on the other hand, due to continuous monitoring of production processes, by reducing human errors associated with overspecialization, which usually occurs when an operator is performing the same functions in the same workplaces over a given period of time. Practically the phenomenon can be described as the loss of operator concentration due to overconfidence, with consequences on productivity and quality of work performed.
The implementation of the future artificial intelligence platform, fully integrated with other existing digital platforms, will succeed if in terms of confidentiality, integrity, evaluability, non-repudiation and authenticity of transmitted, existing or processed elements in the databases, it is achieved. The artificial intelligence system will have to be developed to identify inappropriate data, eliminate it from processing, so as to prevent the contamination of the learning process itself, leading to wrong analyses and alerts. Inadequate data will be those that have been read by improperly functioning sensors, those that can be injected into the system by humans and are poorly conditioned, or those that are brute force altered by elements outside the company.
By: Mihail Fontul, IBER-Oleff