This visual object intelligence platform is making industrial robots smarter
Imagine being able to order your phone, car or even your salad — all personalised to your taste, directly from a ‘Universal’ Factory in your neighbourhood.
This is exactly what Bengaluru-based startup(Cybernetics Laboratory) hopes to make possible one day. Founded by Gokul NA and Nikhil Ramaswamy in 2019, the visual object intelligence startup says it automates the manual labour involved in the manufacturing processes, opening up possibilities for production.
This technological feat in manufacturing, if reached, will be akin to how computers have changed our lives by simplifying data processing.
“We are working on what is called the holy grail of robotics -- a visually enriched robotic arm that can pick, orient and place objects like humans do. This will be achieved by a combination of innovation in hardware and deep research in ML & Vision,” says Gokul.
Typically, industrial robots need customisations to suit production requirements.
Co-founders of CynLr (L:R) Nikhil Ramaswamy and Gokul NA
“Our vision is to simplify the engineering complexity and costly customisations that manufacturing faces today, resulting in most automation being infeasible. We envision a universal factory, which doesn’t have to be rebuilt every time the design of the parts they produce change.
Universal factories is a futuristic concept where factories can morph into producing a wide variety of goods with simple reprogramming or retraining of robots and machines.
“We at CynLr envision that universal factories will become feasible with our visually intelligent robot arms as replacement for traditional part-specific robot installations,” the co-founder adds.
According to the startup founders, universal factories will make today's product-specific factories obsolete.
Today, due to all the custom infrastructure required to produce goods, a factory can produce only a small set of pre-envisioned variants of a specific product in an assembly line. For example, a Mahindra XUV 500 manufacturing line cannot immediately scale to produce a Mahindra 700 as all the automation and systems in the line is purpose-built for the components and dimensions that go into an XUV 500.
“At universal factories, assembly lines of visually intelligent robots can rapidly morph to produce various types of components or products on demand, factories can adapt easily for changing product designs and variants, or maybe even produce entirely different products altogether. A smartphone assembly line today, a healthcare PPE line tomorrow, a rocket engine nozzle line day-after,” Gokul explains.
The team at CynLr
CynLr’s robotic arm is being built as a standardised unit between different objects, different orientations, and different tasks that can be replicated quickly. This makes the assembly line more universal, where businesses can repurpose the line for different use cases, rather than having to change the whole set up even for a dimension change in the dimension of the object being handled.
“We have a working prototype already where a robotic arm can “see” and is able to use our propriety, built from the ground-up, visual object intelligence engine to pick, orient and place objects,” the co-founder says.
Co-founders Gokul and Nikhil began their careers at National Instruments (NI) in 2011. Gokul says the stint at NI gave them an opportunity to learn more about the limitations and impact of vision in industrial automation.
At NI, the duo observed that only three out of every 10 attempted problems in Machine Vision are successfully solved. Machine Vision involves both technology and methods used to help computers “see”.
“We quickly realised that the existing paradigm of vision was built with the assumption that identification is the sole utility. This fails utterly when it comes to manipulation or movement of objects,” Gokul says.
Gokul NA, Co-founder & CTO, CynLr
“On deeper analysis, we realised that computer vision as a domain has oversimplified the vastness of all tasks into a simple identification problem, followed by a blind manipulation of a Machine or Robotic Arm. And every manufacturing plant that used machine vision to solve their coordination challenges has failed terribly,” the co-founder explains.
Gokul says that while attempts had been made to solve the problem of object manipulation, it was done so by using methods best suited for object identification. However, object manipulation and object identification were not treated as two separate problems.
So, Gokul and Nikhil developed an approach to solve this, but then realised that no hardware platform existed to universally apply his approach and solve every object manipulation problem. This is when they quit NI in 2015 to start their consulting business Vyuti to test and prove this approach.
In 2015, the duo bootstrapped Vyuti to understand the market better by providing their expertise in vision to solve long-standing unsolved manufacturing problems. Having been working together on this problem for more than a decade now, the duo say they have solved more than 30 such real-life problems with a 100 percent success rate.
Boosted by this validation, in 2019, they raised their seed funding to work on the hardware and founded Cybernetics Laboratory (CynLr) to evolve their approach into a vision technology platform.
From the product standpoint, the startup says it is in the pre-launch phase.
“We have pilots and POCs (proof of concepts) that are paid in nature going on with both Indian customers and global customers. In India, it's largely the auto component manufacturers and the auto OEMs that we are having engagements with,” Gokul says.
Among pilot customers, the startup says, it has roughly a 75-25 split in terms of Indian and global (Germany, Italy, USA) clients.
“However, when we start delivering large scale commercialisation in 18-24 months, we anticipate this split to reverse,” he adds.
CynLr says its OEM partner is Ace Designs. It declined to disclose the name of its other clients, citing non-disclosure agreements.
The core team
The startup has a team of 20 members. An electronics engineer from BITS Pilani, Nikhil is the CEO of CynLr and looks after management, sales and business operations. Gokul, who did his engineering from Amrita School of Engineering is the CTO at CynLr and is focused on product, technology and brand.
Nikhil says by the end of the year they would scale to more than 50 full-time employees.
YS Design team
The road ahead
While there are numerous use cases for universal factories, the starting point could be the automation of manual tasks performed by humans, the co-founder says. As per a McKinsey report, this market alone is worth more than a trillion dollars in just the US.
Gokul is optimistic that the startup will unlock more value once it moves towards what it calls as its vision of universal factories.
“Think of this from the App Store lens. Once Google and Apple created App Stores to allow third-party developers to create applications for their OS, a huge value was created. The numbers have bypassed the wildest imagination of the companies or even the business analysts,” he explains.
However, as its current milestone, the startup is focusing on creating solutions for robotics and automation markets for manufacturing and warehousing problems. As per its internal estimates, the market is worth $130 billion.
CynLr’s immediate focus is on hiring.
“We are looking for people across the board. Further, we plan to expand to the US market this year,” states Nikhil.
The robotic startup plans to build capacity to address the current pipeline of customers and deliver 100 robots annually.
CynLr claims to have raised total funding of $5.25 million so far. In 2019, the startup raised Rs 5.5 crore (about $720,000) in a seed funding round. The investment came in from Speciale Invest, Arali Ventures, growX ventures, CIIE Initiatives, and investor Dr. Vijay Kedia. In April 2022, it raised $4.5 million in pre-Series A funding round led by Speciale Invest and growX Ventures.
Speaking of competition, Gokul adds, “We are a B2B (business-to-business), multi-disciplinary and diverse application company. The competitors are contextual. There are no direct competitors per se.”