All applications that were implemented as part of "Converged MES" follow the modelling approach introduced by ERP5 (IEEE article XXX) and already used in XXX Box system for XXX and XXX production management. This application modeling is based on the idea that one should first represent tangible reality into business applications and initially ignore any data flows.

This idea is consistent with the approach of Industry 4.0 to "create a virtual representation of physical reality" for any cyber physical systems (CPS).

In the case of a "Converged MES", physical reality consists of:

• a shop floor layout;
• made of multiple areas;
• where certain operators do certain operations on vehicles;
• and where what is produced in one area if pushed to the next area considered for a given model of vehicle.

Configuring a "Converged MES" therefore consists of defining vehicle models, bill of operations for each model, areas, possible operations for each area, assignments of staff to each area and XXX graph ??  XXX .

The defect/reworks application adds to this basic model the ability to:

• declare a defect at before leaving an area;
• declare that the defect was reworked before entering the next area. 

The performance management application adds to this basic model the ability to:

• declare a quantity produced in an area (XXX) for a given time period;
• explain if necessary why the quantity target in that area (XXX) was not met.  

Initial configuration specific to defect/reworks application consists of defining different types of defects.

Initial configuration specific to performance management application consists of defining different types of explanations.

Let us now have a look at a demo, starting with management of defects and reworks. This case is implemented with a dedicated min-application that runs on a smartphone. It is connected in background to ERP5 framework which contains all base data and logic.

The process has four steps:

1. One operator declares a defect
2. Another operator declares an rework that closes the previous defect declaration
3. Shop Floor manager displays a pareto diagram
4. Shop Floor manager discusses priorities with operators to reduce defects 

Let us now have a look at a demo of the second mini application which implements performance management.

The process has four steps:

1. Operator declares a quantity produced during a certain time period (XXX and target quantity ?)
2. Operator explains if necessary why quantity did not meet target
3. Shop Floor manager displays a pareto diagram
4. Shop Floor manager discusses priorities with operators to match target production in the next time periods

During our visit of a "body in-white" workshop (ferrage ), we found out that operators use specific gloves to touch the car's body and detect defects. Those gloves prevent operating a smartphone. This is why we have considered another approach to enter data, directly inspired from current practices in  "body in-white" workshops: a laser wand.

Recent models of laser wand are actually built using Android based smartphone electronics. They are thus just a kind of smartphone with a laser, a sensor and a big button. By selecting bar codes on a large board placed on the wall, operators no longer need to select a value in a pop-up menu on a smartphone, which would require to remove gloves. Bar code selection, from a software point of view, works in the same way as manually entering a value with a keyboard. It is thus very easy to use with those laser wands the same application as the one which was designed to be operated from a smartphone. Only minor customisation is required, using Javascript.

The current Converged MES is for now only operated by humans. It could be extended:

• to be operated also by digitally controlled machines present in a Shop Floor;
• to operate digitally controlled machines present in a Shop Floor.

It would then become a true cyber physical system. By collecting logs from machines (if it is possible and cost efficient), just like we collect defects from human, Converged MES could provide technical assistance thanks to supervised machine learning between defect declaration and log or picture analysis. This type of application of machine learning is very simple. ERP5 backend already includes complete support for scikit-learn machine learning library and storage of logs of any size. Pictures are already being included as part of MES integration in XXX Box ERP.

The same can be said about performance management. Correlation of defects and performance, or of logs and performance, could lead to predict performance issues and their causes in real time, rather than at the end of a time period. This could be useful to take actions immediately or - in other words - to shorten the feedback look of performance management.

Those type of applications will become easier thanks to the current trend of industrial automation towards open hardware, Free Software and Edge Computing. As one of the inventors of Edge computing 10 years ago, Nexedi has started evaluating the possibility to provide an Edge gateway for industrial automation based on open hardware and SlapOS Edge orchestration system. A complete presentation of our ideas is available here: "Edge Computing for Industrial Automation and Control". Nexedi will present in Munich on November XXX its strategy for Industry 4.0, in particular for smart manufacturing execution.