Companies normally aim to beat their competitors through their prices or the products they offer. But one other way in which companies can gain a significant competitive edge is by satisfying the individual needs of clients, through on-demand or order-based production.
Static optimisation: no longer up to the task
Such a solution requires accurate forecasts and short turnaround times, which has an impact not only on the manufacturer, but also on the rest of the supply chain, given the difference in supplier needs from one product to another and the individual nature of logistical planning.
Compared to mass production, this has implications for the supply chain given the need to continuously adapt to the individual needs of all stakeholders along the chain. It’s a situation which can no longer be handled using static optimisation methods.
Luce Brotcorne, director of research at Inria Lille Nord Europe (and member of INOCS)
What’s more, the majority of companies who have already invested in systems and processes aimed at making forecasts more accurate, reducing delays linked to systems and limiting stock have not got the results they were looking for, and have had to deal with issues linked to surpluses or stock shortages.
Adapting manufacturing processes through real-time data and artificial intelligence
The “Supply Chain Optimisation” and “Game theory, decision theory” topics that were addressed during the joint workshop held by Inria and the DFKI in Nancy on 20 and 21 January presented an opportunity to launch a research project between the two institutes through two research teams : INOCS, which is based at the Inria Lille Nord-Europe centre, and the IWi team (Institute for Information Systems) for the DFKI.
Their aim is to use real-time data in order to develop an intelligent system capable of the proactive and partially-automated optimisation and adaptation of manufacturing processes in an integrated context, taking account of current and predicted commercial situations and events, both externally and internally.
These events can either be anticipated (e.g. a supplier fails to meet an agreed delivery deadline), or unforeseen (e.g. a machine breaks down or a key employee is unexpectedly absent).
The team will draw upon data and detailed descriptions from multiple case studies (data relating to production and unforeseen events) the DFKI has access to, in addition to INOCS’ expertise in modelling and integrated problem solving under uncertainty. They will be submitting a project between now and November. In the meantime, phase 1 of the project, the goal of which was to arrive at a precise definition of the problem being studied and to determine an initial model and results, has been validated.