Physical AI: The Next Big Leap in Manufacturing

Physical AI equips machines with the ability to perceive, learn, and act in the real world by integrating AI algorithms with sensors and actuators in physical systems like robots and vehicles.

Kindly note that this illustration generated using ChatGPT has only been posted for representational purposes.

The real value of artificial intelligence (AI) in manufacturing will come only when physical AI scales to a more mature level, as efficiency gains have already been achieved due to automation, Sreenivasa Chakravarti, head of Internet of Things (IoT) and digital engineering at TCS, said.

“Most of these large companies are extremely mature with automation. Automation has done wonders over the last two decades.

“So what is the delta that any kind of AI will provide on top of that?

“It will come with physical AI,” he said during an interaction with Business Standard.

Physical AI is expected to be the next frontier which will change manufacturing enterprises.

It equips machines with the ability to perceive, learn, and act in the real world by integrating AI algorithms with sensors and actuators in physical systems like robots and vehicles.

The shift comes at a time when manufacturers are batt­ling macro uncertainties, supply chain challenges, rising costs, and labour shortages.

Nvidia believes that phy­sical AI will revolutionise the $50 tri­llion manufacturing and logistics ind­u­stries with everything from trucks to cars being robotic and embedded by AI.

It is, however, difficult to quantify the gains immediately.

Chakravarti explains, “For example, you can avoid some downtime.

“So if you are at an OEE (overall equipment effectiveness) of 80 per cent, you will go to 85 or 90 per cent certainly.

“But if you want to get to real gains, that is when physical AI starts manifesting itself at scaled levels.

“Today, it is already there in warehouses and other places.”

Similarly, on the product engineering side, test validation and scenario gene­ration are being automated to a large extent but the core product needs to get transformed into a software-defined product.

“And once that happens, then how quickly can you add features with an AI-based approach.

“That is when the real value will come.

“So I think it is still going to take some time before you can come to accurate numbers,” he explained.

Chakravarti, who became the head of this unit earlier this year, said his focus will be on application lifecycle management and service lifecycle management to drive growth as more traditional products become software-defined ones.

“When you build that product and build in the intelligence into it, you need to, or you have the opportunity then to create an ecosystem of services around those intelligent products.”

TCS’s IoT and digital engineering service is seeing increased demand as its manufacturing customers invest in plant transformations and connected IoT platforms.

This is largely led by utilities, energy and life science segments.

Smart manufacturing and digital thread services saw a positive movement in the first quarter with demand primarily from South America and Europe.

India’s engineering research and development (ER&D) sector can be worth about $100 billion by 2030, from $56 billion currently, Nasscom said last month.

It is one of the fastest-growing in India’s technology industry, which has otherwise witnessed muted growth.

“Utilities is a very sweet spot. You have assets, which have transformers and grid networks.

“Today, you can orchestrate a lot of that with AI-based software,” Chakravarti said.

“For example, since non-conventional energy is on an upswing across the world, you need to make sure that you have conventional power and non-conventional power synchronising.

“And there is a desire to balance availability, cost and carbon.

“So an AI-based orchestration layer becomes the most logical choice.

“So will we see a lot of movement in this space?

“The answer is yes,” Chakravarti said.

Feature Presentation: Ashish Narsale/Rediff