AI and robotics are reshaping industries by automating processes, enhancing efficiency, and driving innovation. In this article, we delve into insights shared by Chirag Agrawal, Global Head of Data Science at Novelis, on how these transformative technologies are being implemented across sectors. From overcoming adoption challenges to ensuring safety and measuring ROI, this discussion provides actionable strategies for decision-makers navigating the complexities of industrial automation.

Read on in this deep dive interview to discover how industries are leveraging AI and robotics to improve operations, the hurdles they face in adoption, and the critical steps needed to ensure successful implementation and measurable outcomes.

Interview:

How are companies introducing this tech?

Chirag explores how AI and robotics are automating processes, with examples like recycling and safety improvements:

AI and robotics, these are the two technologies which are absolutely transforming the way we are driving a lot of an automation. There are a lot of old manual processes in the plan and if we can automate that using AI robotics, then we get better results every time, we get a better quarter product quality. 

How AI and Robotics are helping to boost efficiency

The way companies are incorporating, like for example, in the case of metal, right? When I'm talking about recycling or sustainability, in recycling, let's say if we have the sorting system set up with the help of computer vision and a robotic arm, anytime there is bad material, it just separates it out. Anytime there a good can, it separates it out. So it can help us increase recycling. That's just one example. For example, we can use computer vision to have better safety within the plant because there are a lot industries, especially chemical industries where safety is a big issue. There are areas where it is not safe for humans to go and perform a task. That's where robots can go there, perform the task, and make the environment safer. 

Key lessons of implementing new technologies

There are many key lessons as we're trying to adopt AI. The reason why adoption is one of the biggest challenges is that people are not ready for new technology. For example, if I'm drinking coffee every day in the morning for the last 10-15 years, and someone says start drinking tea because it's better for health, I agree it is better for my health, but would I change? Same thing with new technology, AI as well. It's going be a good for industry, but it'll take some time to change people and make the workforce ready. And so that's one of the biggest roadblocks at this point. 

Testing artificial intelligent solutions

So for testing any AI solution, and that is super important as we are talking about the industries which can impact the people around them. For example, if an autonomous car has an accident in real time it can be very serious. The same applies to the heavy industry. If you have applied autonomous AI robots there and if they get into a serious accident with someone who's working together with them, it's a big hazard for the company. 

How do we make these safer? 

We test more and how do we test more? Well we have always have a baseline. When a task has been performed by human, we check, can the robot or AI do a similar task at equal or maybe 10% less, 20% less. We can always create those baselines that give you a success criteria. If my robot or AI breaks this threshold, then I'm gonna deploy in a maybe a contained environment. So do not deploy in a free environment. Even autonomous cars at this point are deployed in a very small city, small area, not everywhere. They're still slowly scaling that to bigger areas. So two things:

1. Test it, always have the baseline.
2. Start small. Start with a contained environment. 

Understandiing return of investment

If I talk about return on investment for any technology, if you're investing $10 in something, we want return for that $10. And same goes with AI as well. So what could be those return on investment measures or what could be those KPIs for industrial companies be, and it really depends on what kind of use cases we are doing. For example in safety, if we have deployed computer vision or cameras, and we are making sure that people's lives are safer, we reduced the number of incidents happening within the plant, so that could be one fact area or one return on investment measurement. 

Another way to measure your return on investment could be about predicting maintenance. Have we reduced the unplanned downtime of the plants or the machines that have been working or have been operated within the industry center? So that would be another KPI that can be measured for calculating the return on investment. Now the question is how do we get return on investment? So that could be different criteria based on your baseline. So I just mentioned about safety, the unplanned downtime. It could be how many defects we had before AI and after AI. Are we able to reduce the number of defects with the time, but after implementing AI that could be another key. So this could be some of the criteria, how we can measure ROI in an industrial setting.

Watch the full interview on Streamly here.