Artificial Intelligence (AI) can help predict and achieve higher efficiency levels which directly translate to cost savings. With the use of AI, manufacturers can expect to create a huge impact on the bottom line in due course, writes Pawan Chhibba.

 

Artificial intelligence (AI) is the ability of a machine to give answers to various problems based on its learnings from the information and lessons it has been subjected to in the past. There are various ways with which a machine can be taught to provide answers and the most common ones are – supervised and unsupervised learnings. In supervised learning, the algorithm (algo) is provided with both an input (training data) and the output data. Based on this information the algo does a comparison of its answers with the actual answers and improves its accuracy. Equipped with this information the algo is able to predict the output, with minimal prediction errors, for a new set of inputs. A common application is a linear regression.

 

In contrast in an unsupervised learning the input data is known to the machine sans the relationship to the output information/ results. The algo needs to extract the relationship/ understand the patterns on its own in order to be able to reach a conclusive prediction. Clustering is an example of unsupervised learning wherein an imperfect set of data is used by the algo to combine similar objects together. Upon receiving new set of data, the algo based on the variables considered initially to cluster the data, allocates the data points to each of these clusters thus providing labels to these points.

 

AI in manufacturing

Those already part of the manufacturing domain would appreciate the fact that manufacturing has been subject to automation (a subset of AI) much before we all started talking about implications of AI in manufacturing. Seeing a robotic arm executing a welding job in a manufacturing setup is not a new phenomenon for most. Watching programmed CNC machines conducting predefined sets of operations on a sheet of metal is also a common scenario. When one thinks about such examples, the question that presents itself is – What difference does AI creates for manufacturers? What value does it generate for an organisation?

 

The answer to this question can be given in two words – Smart Factory. My definition of smart factories is – A factory with a distributed sensor network, built in intelligence and connected to an outer world, which can learn from the established data on people, material and processes and can make changes to achieve performance far better than that achieved by human intervention. Performance is a broad term and can mean different things to different people. 

 

At the very basic level, performance can relate to the overall efficiency the entire value chain can achieve owing to the improvements injected by the smart factory. It’s the shear ability of the factory to mutate that makes it way different from a mere automation that we all know of. A smart factory has the ability to learn and transform in ways which help organisations analyse, predict and adapt to both technological and ecological changes without the need of human intervention and by the use of the embedded analytical power. Smart factories are agile and help organisations anticipate and fulfill the needs of their customers through autonomous reconfiguration of processes, design of information and material flows, timely and automatic dissemination of information such as placing orders on the vendors, sending alerts to maintenance staff in case of anomalies, charting out predictions for machine failure and mapping out probable scenarios to meet key performance indicators.

 

Accommodating a product mix on an assembly line in the early days of industrial revolution used to be a far-fetched possibility, as the changeover process used to be cumbersome and it most of the times led to loss of productivity. Not anymore. Mass customization is the new normal in manufacturing, thanks to the use of AI. Because of its ability to integrate the entire value chain, AI can help achieve customisation for an order quantity as low as one unit without hindering the efficiency of the manufacturing system. It also helps organisations achieve quick turnaround time, near zero defects, and reduced operational costs through improved utilisation of scarce resources.

 

Use cases

A use case of AI in manufacturing can be related to prediction of efficiency of both machines and human resource. Sensors are being used on the machines to gather data on duration for which the machine was on (uptime), the frequency of breakdowns, the time it took to rectify the issue, and the associated loss of production. This information is then passed on to a cloud equipped with a mathematical model which analyses the data to arrive at a prediction for achieved efficiency vis-à-vis planned. This information is then used by AI to establish changes to the process, the maintenance schedules, the parts availability etc. to minimise the variations.

 

Video analytics is another interesting use case. Productivity for manpower can be predicted through use of video analytics which helps compare the actual presence of a resource vis-à-vis the planned presence and helps arrive at the loss to production. The reasons behind the absence can then be investigated and this information can be fed into the production roster to make necessary changes/ adjustments. Integration with the messaging service can also provide an alert to the production manager in case a resource is missing from a work station and corrective / preventive measures can be taken. Thus, AI can help predict and achieve higher efficiency levels which directly translates to cost savings. With the use of AI manufacturers can expect to create a huge impact on the bottom line in due course.

 

Asset sharing is another opportunity which AI can contribute to and which the manufacturers must consider. Availability of correct information on asset status through a centralised data repository is vital for an AI system to arrive at asset sharing scenarios and associated costs.

If we zoom out to the level of the entire value chain, a seamless integration can be achieved by allowing machines to talk to each other and to share information on current and required production levels, plugging the cost leakages through increased system efficiency, automated adjustment to raw material and model mix, automated allocation of resources, auto-planning and requisitions for power required from the grid and for other natural resources.

 

Augmented reality (AR) is another area which can change the way associates at the shop floor work. Based on the component being used, the AI can project the work instructions and pull up the related drawings on the screen or a wearable of an associate and can even use historic information, attained through experience of other associates, to provide inputs on the best methods to assemble a product. If during the assembly process a part reaches a ROP (reordering point) the AI can search the database for the item and associated vendor code and can automatically place an order based on the MOQ (minimum order quantity).

 

Because EHS is one of the most critical regulatory requirements, AI can keep a track of the working conditions for the associates on the shop floor and based on the attributes such as temperature, humidity, oxygen levels, chemical composition of the air etc, can predict the possibility of associates falling sick, informing the supervisors to plan for manpower accordingly.

 

Conclusion

AI is transforming the entire manufacturing landscape at an unimaginable rate. Organisations involved in manufacturing need to conduct an internal due diligence of the processes, technological maturity, organisational structure & culture, ingrained agility, and risks before infusing AI into the DNA of the organisation. Once the mutation commences, AI is expected to transform the organisation creating greater value for customers in the form of high quality customised products delivered at never before costs and quality and a far greater value for organisations by reducing costs associated with waste of material & manpower, driving greater efficiencies & productivity, providing predictive analytics for better inventory & manpower planning and creating competitive advantages. 

 

AI is the new normal for organisations that are part of manufacturing sector. Those organisations which turn blind eye towards AI would soon find themselves competing for resources and customers and would render themselves irrelevant.

 

 

About the author:

Pawan Chhibba is Business Strategy & Program Management Professional at Indian Institute of Management (IIM), Calcutta