Artificial Intelligence (AI) is building new capabilities in manufacturing’
Dr Pradeep Chatterjee
conversation with Industrial Automation
Artificial Intelligence (AI)
How does a typical manufacturing plant, already automated in the conventional sense, make a transition to the current trend of Industry 4.0 or IIoT?
For conventional automated plants to make transition to Industry 4.0 or IIoT, there are several devices available to connect additional sensors to machines and capture that data to IIoT platforms. The basic scalable infrastructure for IIoT as secured network, edge devices, cloud environment, IIoT platform, etc., which are required for conventional or modern automation machines should be put in place. However, once the basic infrastructure is in place, organisations should not plan to connect all conventional/modern automated machines to IIoT platform. It should be need based else it will be difficult to justify return on investments.
Automation is no longer seen in isolation, but as a means to achieve a ‘connected plant’ in the real sense across several plants and locations. Easier said than done?
Yes, achieving connected plants involve lot of activities and sub-systems to be working together. Each machine should be equipped with necessary sensors and actuators connected to gateway devices and network, which finally connects to a cloud environment talking to each other. It involves lot of investments as well as lot of planning to form connected plants across locations. But that’s not the end of it. Many people invest and start collecting data and then try to decide what to do with the data, which is not the right approach. First you need to be very clear what you want to do and then plan what all data you need and how you plan to capture those data. These data might be from single machine or from different machines within a plant or from different plants connected together. Deciding what I want to do and why I want to do it, is the most difficult part.
For large companies like Cummins with dozens of plants and heavy investments in capital equipment, with a mix of new and not-so-new machinery, how difficult is this?
Several new machines today are coming with IIoT capabilities, which make it easier. Definitely some challenges and investments are involved for connecting not-so-new machines. But as I mentioned, if there is business value, same should be pursued and investments should be done. If a machine has IIoT capability it necessarily need not be hooked into IIoT environment. However, readiness from skill, talent, network, cyber security and necessary infrastructure are pre-requisites which any plant/organisation should be ready with to leverage the capabilities of IIoT as and where required. Setting up the complete eco-system takes more time and effort than connecting any modern machine or not so new machine with additional sensors.
Artificial Intelligence is making rapid inroads in manufacturing. What are the implications?
Artificial Intelligence (AI) is building new capabilities in manufacturing. Conventional intelligence is based on mathematical relationships of data to draw meaningful conclusions. AI goes beyond that and able to see unknown pattern and co-relation between data through perception based training, as human perceptions, which goes beyond mathematical formulae. It led to development of AI based solutions for quality improvement, predicting failures of machines/products or forecast markets, improvement in supply chain, services, bringing new business models, etc. With AI these predictions become more realistic and accurate since it is able to capture unknown relationships of data.
There is that talk of data deluge with all these sensors integrated in machines and equipment generating humongous amount of data. Is there a method in this madness?
Collecting all sorts of data without visibility what to do with it will lead to this madness. So one should take conscious decision what he wants to do and then decide which data to be collected. It is also necessary to decide which data requires real-time analytics which may be performed at edge devices and need not store and which requires storage for future or historical references. It is necessary to decide which data to be collected at what frequency. People need to analyse if I can do same level of analysis with less data using AI capabilities as AI is capable to meet requirements with limited or partial data. There should be plan for data purging .On an overall basis, proper plan of data capture and utilisation of data needs to be worked out at different levels for each data point to avoid humongous amount of unwanted data generated.
Cyber security is now a critical area. Is there a truly secure facility amidst this cat and mouse chase?
Cyber security is a never ending race and it will continue like this. You make something secured and then new threats will come up and you address same and it goes on. Our focus should be to setup environments which are difficult to access by intruders. Through various layers of security as demilitarised zone and then creating another industrial demilitarised zone on top of it for shop-floor, creates different layers of security so that it becomes difficult for hacker to hack into your environment. It is just like locking your door with one lock when you go out or put two or more locks. It can only make things difficult and time consuming for a thief to break open your door which can give you a chance of early detection before he really breaks open. Once you come to know about it, you change the lock which opens with different set of keys or add an additional lock.
Is there a direct connection between making process efficient and rendering human jobs redundant? Is it as simple as that?
It all depends which processes are becoming efficient. There are set of processes which are not possible for human to do but can be done by automation. For example observing frequently top of a high chimney with high temperature can now be done with drones and image processing, which is a challenge for human to do. Here it does not make any human job redundant but adding new capability to do something which was earlier not possible. There can be set of processes which will assist human to take better decisions for example an operator picking up a part for fitment can be alerted with automation if he picks up a wrong part. This is not making human job redundant but assisting the operator to do fitment of right part. There can be some human intensive jobs which if automated can result in human job redundancy. So a generic statement that making processes efficient lead to human jobs redundancy might not be a right statement.
India with its abundant manpower now aspires to raise its manufacturing activity to the level of advanced economies which is impossible without high level automation. Is there a contradiction here?
I believe high level of automation leads to more job creation which will be good for India with abundant manpower. People had similar fears when computers started coming. Emails and work process flows in IT took away the jobs of office peons but gave birth to the IT industry where scope of jobs increased several times. The IT business automation created far more number of jobs than the number of jobs it took away. The peon might not be doing the job of a peon in an office but he may be working in a data centre with a different job profile. Skills will change, people have to learn new technologies and adopt the change so that India leads the world in the field of automation with abundant manpower just like it is leading in field of IT.
© 2017 IED All right reserved.