For long, Bengaluru has been a back-office function for international tech firms. However, that is changing now with Drishti, a California-based AI company that  is driving disruption for manufacturers of all stripes. In fact, it is among the top 56 early growth stage companies recognised by the World Economic Forum for developing a cutting-edge technology that keeps humans in the loop in the manufacturing sector. Drishti is one of the first AI companies producing deep learning solutions which will not automate factory workers out of their jobs, but augment their productivity. 

It is one of the first companies to commercialise action recognition technology specifically for manufacturing. Now, their Bengaluru-based engineering team led by Dr Krishnendu Chaudhury is solving some of the toughest problems in AI, up and down the entire stack, as Dave Prager, Head of Marketing and Business Development at Drishti puts it. 

Some of the critical solutions Drishti is solving are in the area of Industry 4.0. “Computer vision is playing an increasingly key role in manufacturing, as it is in other industries. At Drishti, we use vision technology to augment human workers,” said Prager. Citing an example, he shared that similar to technology like spell check, where the computer sees a mistake and suggests a correction to the typist, one of the many things that Drishti’s computer vision solution sees is if an assembly line worker misses a step, it triggers a correction. 

In that way, we are able to keep the benefits that humans provide – cognition and adaptability – and marry them to the benefits of machines, augmenting and extending the efforts of the human worker, he shared. 

Bengaluru Engineering Team Produces Cutting-edge Solutions

Drishti has drawn some of its most talented, hard-working individuals from India. Bengaluru office boasts of double-digit numbers of employees who are working on developing Drishti’s signature action recognition technology, a new deep learning architecture that generates real-time analytics on highly variable human actions using a video stream. “It’s orders of magnitude more difficult than object recognition, in the same way that parsing a spoken sentence is much more difficult than analysing the written word,” said Prager.  

The solution’s architecture goes beyond Deep Learning. The team built a vertically integrated solution from the sensing layer all the way to change management in collaboration with customers. “Our engineers love working at Drishti because, unlike with other Bengaluru offices, they’re not QA’ing at the periphery of other people’s innovations – here, with Drishti, they are the real innovators,” he said. 

How Computer Vision Can Revolutionise Assembly Line Work

Of late, the scope of AI in factory automation has increased significantly. Manufacturers are increasingly recognising that AI has the potential to transform their business. “One piece of advice we give companies who are implementing AI is to bypass the lab; it’s important to test AI applications, as with any new technology, in their destination environment, with all of the variability and challenges that come with it,” said Prager. Laboratory deployment doesn’t provide a real world equivalent, and therefore many AI solutions that succeed in the lab ultimately fail on the floor. 

Factories Are Not Ruled By Robots 

We usually believe that factories are full of robots with a few humans in lab coats wandering here and there. The truth is the exact opposite. The company conducted primary research with AT Kearney, the global strategic management consulting firm wherein the findings indicated that 72% of tasks in the factory are still performed by humans. 

Drishti addresses a longstanding manufacturing problem that is pervasive across the industry: manufacturers are using outdated manual time and motion studies to gather data on tasks performed by humans. Drishti changes that by automatically collecting data on every task that’s performed on the floor, and providing that data to manufacturers. 

The market for factory automation is huge globally. Manufacturing represents $12 trillion of an $80 trillion global economy; in India, manufacturing makes up nearly $3 trillion of the national economy. With the country’s Make In India initiative, more investments are being made to push those numbers up, which opens up even more opportunity. 

AI Is The Best Fit For Factory Automation 

AI is well-suited to factory automation. Manufacturing is defined by predictable and repetitive systems. Because AI can be applied more easily in the factory than in highly variable environments, such as self-driving cars, the possibilities to solve real-world problems in real-world environments are greater. Also, because of the massive scale of manufacturing, and its impact on global economies, the companies who find an edge in AI can be extremely competitive, providing a huge incentive to invest in new technology.

Of late, the manufacturers have also started thinking about data. In India specifically, AI has been less of a focus for the government than other technologies. “Despite this and given the large need, you will see a very large “annotation industry” growing in India, just as we saw the BPO and call center businesses in the past,” said Prager.