Firm boosts human effort in factories, using computer vision, deep learning

Drishti combines video and AI for data capturing in manufacturing to increase process efficiency and reduce defects

Despite all of the innovation and change in the industry, manufacturers are running blind to the greatest sources of value and variability inside the factory due to manual activities.There’s no easy or accurate way to capture data that represents these actions. 

Drishti Technologies, a company backed by marquee investors such as Andreessen Horowitz and Emergence Capital, is addressing this problem. Its technology is focused on humans so as to make a true digital transformation in the manufacturing world. It does this by making manual processes visible to factory analytics.The idea is to extend human potential in an increasingly automated world by combining the creativity and adaptability of people with the cognitive power of AI (artificial intelligence). 

Drishti is looking back more than 100 years to Ford Motor Company founder Henry Ford’s assembly line as the target for its innovation. It is essentially an eye on the factory floor to monitor thousands of workers at once. Its software uses computer vision to digitize human actions. It takes real-time video footage from cameras installed throughout a factory and then applies deep learning to assign data points to label different types of manual labor. The data is turned into information to help factories increase efficiency or quality.

"For example, if our camera is looking at an automobile assembly, and someone is assembling a brake pad. Let’s say there is a step of tightening a nut. If that tightening is forgotten, and a car with a faulty brake goes out into the world. You can calculate the bad things that can happen,” says Krishnendu Chaudhury, co-founder and chief technology officer (CTO) of Drishti. “On the other hand, if the manufacturing operator is warned right there, he just reaches out and tightens the nut. Problem solved. Thus, the real-timeness of the warning makes all the difference."

Drishti was founded in 2016 by Chaudhury along with Prasad Akella (founder), and serial entrepreneur Ashish Gupta (co-founder and Board Member). This trio is considered to be a top combination that brings to the table expertise and knowledge in manufacturing and AI combined with the business acumen.

Chaudhury  previously worked with tech firms such as Google, Flipkart and Adobe. He is a renowned expert in video analytics and has dozens of patents to his credit in the areas of computer vision and deep learning (AI). He has also authored a book titled “Math and Architectures of Deep Learning.”

Dristi’s founder and chief strategy officer Prasad Akella, transformed manufacturing in the 1990s as leader of the General Motors team behind the world’s first collaborative robots (cobots). With cobots, Akella advanced robotics to safely amplify workers’ physical capabilities. With Drishti, Akella once again returned to the factory to enhance workers’ capabilities. This time, he is doing it by driving advances in computer vision and AI.

Another co-founder, Ashish Gupta, is a well-known investor. Gupta holds a Ph.D. in Computer Science from Stanford University and a Bachelor’s degree from Indian Institute of Technology (IIT) Kanpur.He has co-founded two successful companies - Tavant Technologies and Junglee (acquired by Amazon).

Drishti is also playing a key role in the area of medical manufacturing. For instance, there is a medical manufacturer that builds equipment which sends a catheter inside the human body. There is a step involving adding a drop,or maybe three drops, of oil to a joint.If you forget to squeeze the oil, the machine can get jammed. Worse, it may not get jammed immediately, but can get jammed when there is a catheter in the human body. Then it would refuse to turn anymore and it can’t be taken out. 

“This would be a disaster, involving human life. No testing would reveal the mistake as it will not fail immediately,” explains Chaudhury. “Right now, reasonably senior employees watch over the process and make sure it has happened. Our cameras can look and ensure that that step was done - it’s foolproof."

Drishti said its technology can be applied in many areas such as self-driving cars and for driver habit analytics. The other areas include customer heat maps and behavior analysis in the area of retail and also seeing eye glasses for visually impaired and sports analysis. 

How does Drishti’s technology work? Drishti provides video analytics and video traceability for manual assembly lines. The discrete manufacturing customers stream video from each station into the cloud. Here the software creates incredible volumes of lean production data. Engineers, supervisors, trainers, and operators use this data to make better decisions, faster — with each data point backed by video. Forget incremental, step-by-step improvement. Drishti’s purpose is to improve the entire system all at once. The firm said the customers make giant leaps in productivity, quality, and training.

This all happens on the backbone of the firm’s proprietary neural networks. Its “action recognition” technology uses AI to parse video streams and produce data on manual actions from every station on the factory floor.

That data then populates a series of dashboards that help direct production staff on where to focus their attention. It highlights bottleneck locations and, more importantly, why they exist. This way significant volumes of actionable data on manual activities are guiding production decisions in the manufacturing industry.

The firm said many multi-billion dollar, global manufacturing companies have already contracted with Drishti and deployed the cameras on their factory floors. 

Drishti is one of the very few Silicon Valley (U.S.)-based companies whose core research and development (R&D) happens in Bengaluru. It has 15 patents pending, all related to technology developed  in Bengaluru. Drishti partners with three of the top six global auto OEMs (original equipment manufacturers), two of the top five auto tier-1s and two of the top three electronics contract manufacturers globally. The firm said the technology and solutions are a first in manufacturing and possibly in the world.

"AI is at a very nascent stage and to find youngsters who can bring innovative thoughts to the table is refreshing. Bengaluru has been an amazing place to network and interact with like-minded members. They can hold a place of their own, anywhere in the world,” says Chaudhury. “Their ability to think laterally is the key differentiator and I am happy to say, Drishti has been the hub for such thoughts and therefore innovations. We look forward to growing this family of innovators."

Drishti’s product’s key use cases include defect reduction, training acceleration, process monitoring and optimization. The other use cases include remote visibility and collaboration and new product ramping.

Drishti said its chief differentiator is its focus on people and not machines. It creates data from human activities as opposed to conventional Industry 4.0 technologies which are more focused on robots, automation and machine-generated datasets. The firm said it empowers people on the line instead of displacing them, promoting job retention and satisfaction.

Drishti has raised a total of $37 million in funding over two rounds with their latest being Series B financing round led by venture capital firm Sozo Ventures in 2020. Drishti has a global operational presence, including the US, UK, Mexico and India.