Different Phases of Technology Evolution
Published on : Sunday 13-02-2022
Application of AI/ML models is not limited to the shop floor but in multiple areas of operations in manufacturing, says Dr Pradeep Chatterjee.
Industrial Automation, which started from 1920, underwent different phases of technology evolution. At present it is Artificial Intelligence, Machine Learning, Virtual Reality, Augmented Reality, 3D printing and Intelligent Vision Systems, which are disrupting the manufacturing and other industry sectors. Going forward it will be BCI (Brain Computer Interface) along-with AI (Artificial Intelligence) to create Collaborative Intelligence (CI) of man and machine to bring the future technology evolution for industrial automation.
To leverage the power of AI/ML, it has dependency on voluminous data, which is already available on the shop floor. The challenges are two-fold:
1. Data is there but there are hybrid controllers and all controllers are not connected to central repository, and
2. Old plants where production lines have a mix of old and new machines, old machines may not have data captured for further analysis.
To overcome this, part of the analysis can be done over edge computing devices connected to a machine, which works on the machine data available. Some part of the data can be pushed to the cloud through edge devices, which can be used for predictive analysis. Such data can be collected from multiple plants of the manufacturer and help to leverage in development of predictive models based on data collected from different scenarios. It can also take into consideration some locational factors that impact the predictive models. In case of insufficient data, digital twins can be created to generate synthetic data for training models. Going forward I expect manufacturers will form consortia to share non-competitive data, which can help each other to develop AI/ML based predictive models.
Application of AI/ML models is not limited to the shop floor but in multiple areas of operations in manufacturing. It starts from supply chain predictions to human resource predictions to market forecasts of products. Depending on past data as well as data points of external dominating factors, robust predictive models can be built for all the relevant business areas. In fact to test it out we built a stock price predictive model using all these factors which led to 95-97% accuracies. Similar concepts can be extended to predict operational parameters in manufacturing for supply chain and predict demand of products in the market.
While this is relatively easy to do for large manufacturers, SMEs find it a challenge. To handle this we extend our solutions to the respective vendor communities, which has two-fold advantages:
1. SMEs can get similar solutions, and
2. Some defects are detected early even before it reaches the OEM.
With digital solutions there are complexities involved in developing the solutions. However the user interface can be kept simple and easy to handle. It can be through usual channels of communications if we are using digital channels; or in machines it can be through easy to use HMI solutions. If HMI interfaces are simple to use, technology deployment will not be a challenge even if a complex AI solution is running at the backend. Also intelligent systems should be able to automate corrections or next steps to be taken without any human intervention. However, if it requires human intervention and the existing technologies are not that mature, even simple audio-visual arrangements for defect detection coupled with AI solutions working at backend can be good enough to signal an operator that a particular job needs to be rejected. So a simple user interface can minimise the requirement of reskilling manpower with deployment of advanced technologies at backend.
Dr Pradeep Chatterjee is Head – Digital Transformation & Experience Management Business, at TML Business Services Limited. He has worked in technology innovation, strategy development and deployment, business relationship management and translating concepts to innovation and business opportunity through forecasting technology trends.