Fetch.ai x Festo x University of Cambridge

We are incredibly excited to announce that we are collaborating with Festo and the Manufacturing Analytics Group at the University of Cambridge, Institute of Manufacturing (IfM), to provide research and recommendations to successfully develop a multi-agent system architecture for distributed manufacturing. With the use of our Fetch.ai technology stack, including the Autonomous Economic Agents framework and blockchain in synchronized harmony, our goal is to transform the existing manufacturing control systems, delivering a scalable solution for the 21st century and beyond.

Despite advancements in technology, the manufacturing industry remains rife with challenges and inefficiencies, lowering productivity, utilization, production variety. Distributed Manufacturing is a relatively new paradigm proposed to overcome some of these challenges. In Distributed Manufacturing, producers lease excess capacity for customized, low volume high variety orders. Whilst a promising approach to improve productivity and reduce wasted capacity, the take up of Distributed Manufacturing itself has been difficult. One of the issues is a lack of automated mechanisms to match suppliers and buyers. Firms need to spend manual effort to orchestrate matches, which are unlikely to outweigh the cost benefits obtained from a Distributed Manufacturing approach. Another issue has been the monopolization of economic transactions by platform providers, which results in suppliers having to succumb to pressure for reducing prices.

For years, multi-agent systems (MAS) architecture has been considered a possible solution to reducing the above issues associated with the conventional, centralized manufacturing orchestration. MAS offers a way to automatically allocate suppliers of services to buyers, without the associated manual transaction costs. It also allows for decentralized matchmaking, reducing the power of platform providers in suppliers. MAS take up has been slow due to a lack of suitable infrastructure. Ultimately, the missing link has been the application of cutting-edge research in AI and the connection with the blockchain technology that helps us understand the benefits that multi-agent systems can provide within the distributed manufacturing sector.

This collaboration will bridge these gaps, shedding light on the lack of current industry applications available to act as benchmarks to capitalize on the solutions multi-agent systems can provide to the distributing manufacturing sector.

Multi-agent systems to achieve distributed manufacturing

Utilizing multi-agent systems as a solution to distributed manufacturing will address the following:

1. Improve flexibility and computational efficiency, thus enhancing overall system performance.
2. Create distributed platform economies by providing a more natural way of representing task allocation, team planning, user preferences, open environments, and so on, through autonomous agent interaction.
3. Alleviate concerns over the handling of confidential information and creation of bottlenecks.

Our findings so far include

• Our test results show that using multi-criteria decision-making allows order agents, with varying preferences, to select the best manufacturer for their needs.
• The simulation testbed can be used by Festo to run additional tests and find problems with new features before they are deployed in the real marketplace.
• The research has identified the need to reduce the number of bottlenecks and advance the optimization functionality of agents. This will be a trade-off between added complexity to allow more flexibility for manufacturers to use the method that is best suited to their needs.

The Fetch.ai-Festo-University of Cambridge collaboration is set to continue to develop these approaches for further cases and suggest guidelines for future development directions to develop multi-agent systems for distributed manufacturing. Stay tuned for more information and updates on our blog.