Manufacturing is a dynamic industry, and a shop floor is a frenetic and unpredictable world.
Have you ever visited a shop floor in a factory? If yes, you must have definitely heard the following:
– “Side channel is not fitting properly between the frames.”
“The air pressure is low; so, can’t rivet the skin.”
“Special fasteners are in shortage.”
“Calibration of tool has just expired.”
An assembly shop supervisor gets to hear this on a daily basis and has to do fire-fighting to ensure throughput despite such issues. It requires considerable human intellect, experience, intuition, and problem solving skills to manage such situations.
Applying Artificial Intelligence to Shop Floor
The decision making process is quite time consuming, as the concerned professionals first decide on the data to be searched, then the report to be fetched for a particular time period, and then analysis of the report and data for insights. Most of the time they find themselves doing it as a corrective or reactive measure, once a problem has already halted the production. However, the rise of artificial intelligence (AI) has given all of us a hope for proactive decision making and problem solving, much before the problem has occurred. We have seen how application of AI in the Personal Assistant (PA) apps such as Apple’s Siri and Google Assistant has brought about a revolution in the way people interact with their mobile phones and smart devices . I believe a similar PA app can be designed for the shop floor, or even a business enterprise. We can call it “Digital Apprentice” (DA), and it can go a long way in assisting the supervisors to take an optimal decision.
Digital Apprentice – Leveraging ecosystem of DSS and physical processes
Manufacturing professionals leverage various decision support systems (DSS) such as simulation, 3D modelling, manufacturing execution, inventory management, and plant maintenance system. These systems provide information, analysis, and visualization, which can be utilized by the user to take a decision. DA can be integrated with these systems to and can also be programmed to take inputs from the physical processes such as welding, painting, and assembly operations via supervisory control and data acquisition (SCADA) layer (ISA 95). Leveraging inputs from these cyber-physical worlds and using cognitive capability developed with the user shadowing technique, DA will be able to suggest the best solution approach to the user.
Think of a time in future when in case of a probable problem scenario, DA will ‘sense’ a problem by taking advantage of its integration with physical process, natural language, context awareness capabilities and will define clear problem statement. Then, it will create ‘models’ by analysing previous problem solving approaches taken up by the user. Further, DA will choose an appropriate analytical model to solve the problem and arrive at all plausible options. It will not stop here and would ‘collaborate’ with various team members and working groups to socialize the feasible options, take feedback, and execute workflows. In the end, DA will ‘suggest’ the best option to user with pros and cons, that too prior to actual occurrence of the problem.
To understand how a DA would work, consider a scenario. A production supervisor wearing DA arrives at a sub-assembly station where some special fasteners are in shortage. DA, using its language and context awareness capability, picks up the issue, creates logs, and communicates (over both voice and text) problem statement to user. It then uses an existing part shortage model to come up with possible corrective actions such as – fulfil part shortage or – defer special fastening operations to end of line. Then, leveraging AI-enabled DSS, it will look for the fastener in inventory of market place and lineside bins of similar stations. It will even search for an alternate engineering part that could serve the purpose. Based on available options, it will suggest the next course of action to the supervisor. It’s quite similar to Siri or Google Assistant suggesting you to go to a specific restaurant based on your location, time, and preference. I have chosen to write about DA’s applicability in the production department, but it can be customized and deployed across business functions of an enterprise.
Superior Decision Making with DA
In no way am I suggesting that DA will replace its user. But, its use will definitely aid the decision-making capabilities of an enterprise. This will help minimize flaws such as intuition and biases in human decision making, which sometimes overshadow our cognitive thinking. Moreover, loss of knowledge and expertise, which most enterprises have to face when people leave the organization, can be controlled.
Don’t you think DA can prove to be your best office buddy, especially in the decision making situations?