How this Indian robotic startup adds 'human instincts' to robotic arms

Prasanth Aby Thomas, DIGITIMES, Bangalore 0

Gokul N A, founder, CynLr. Credit: CynLr.

Robotic arms have been a part of various industries for quite some time now. However, most of them still cannot work independently, having to be spoon-fed each process separately. CynLr, a startup based in Bangalore, India aims to bridge this gap by developing layers of perception that extend from 'eye' to 'fingers'.

This encompasses seeing and understanding the environment, recognizing objects in their myriad forms, and intelligently manipulating them based on their properties.

Speaking to Digitimes Asia recently, Gokul N A, founder of CynLr, pointed out that the company aims for training-less systems with basic instincts for object manipulation, akin to human capabilities. This approach negates the need for extensive programming for each new object or scenario, enabling the system to handle objects it has never encountered before.

"Our work starts where conventional robotics ends—transforming robotic arms from mere programmable machines into entities capable of learning, adapting, and making decisions based on their 'perception' of the environment," Gokul said.

What it is

Unlike traditional robotic arms, CynLr's focus isn't on the hardware itself but rather on the intelligent and manipulation stack that integrates with these systems. They start with vision technology, enabling our robots to adapt and manipulate objects in ways traditional robots or even COBOTS currently can't.

"Robotic arms are essentially electromechanical systems with no innate intelligence," Gokul said. "They operate through pre-recorded motions, lacking awareness of what they're interacting with. Collaborative robots (COBOTS) offer improvements by working safely around humans without requiring protective barriers, but they still don't understand their surroundings or how to adapt to changes."

CynLr's technology is designed to enhance the ability of robots to comprehend and engage with their surroundings in a complex way. This capability extends to recognizing and adjusting to various object orientations and conditions, where existing robotic systems often falter.

The strategy employed by CynLr incorporates sophisticated vision and sensor systems. These systems transcend pattern recognition, encompassing an understanding of depth, motion, and additional essential factors in real time, offering a more nuanced interaction between robots and their environment.

Why this is relevant

The robotics market, valued at $48 billion in 2019, is complex and fragmented, according to data from the IFR. Robotic arm sales made up $16 billion, while $32 billion went into environment customization for robots. This has limited robotic arms to specific tasks like welding, painting, and palletizing, with welding comprising over 63 percent of sales. Deploying a standard automation solution typically requires 12 to 24 months and costs $150,000 to $250,000.

"CynLr aims to revolutionize this scenario by transforming complex manufacturing lines, which are currently dependent on specialized machines, into versatile, universally adaptable factories," Gokul said. "These factories would utilize simple, dexterous robots that operate without requiring environmental customization, thus removing the need for specialized machinery. This innovation promises to grant factories unparalleled flexibility to adapt to market-driven design changes, enabling easy repurposing and reconfiguration without needing customization, all driven by CynLr's advanced 'Object Intelligence'."

Traditional vision systems in robotics have been limited by their reliance on static pattern and color recognition, struggling with variations in object appearance due to changes in perspective or lighting. Reflective objects pose a particular challenge, as the perceived color can change dramatically based on the environment.

CynLr overcomes these limitations by employing advanced algorithms that mimic human perceptual abilities, enabling robots to interact with objects more naturally and intuitively.

"Current robotic arms are highly precise but lack flexibility, requiring objects to be presented in exactly the right position and orientation," Gokul said. "Our technology, by contrast, allows robots to understand and adjust to variations, much like a human would. This capability extends to recognizing and manipulating objects never encountered before, offering adaptability unseen in traditional automation."

Looking to expand East

CynLr's robotics system is an intricate assembly of around 400 parts sourced from international markets.

The technologies employed are sourced globally, with a supply chain extending across China, South Korea, Taiwan, Japan, Germany, France, Israel, Denmark, Canada, and Switzerland. CynLr has established a hardware design center in Switzerland and a business development office in the US.

"To enhance its supply chain in sensing technologies and to leverage superior manufacturing capabilities, CynLr is actively engaging with several Chinese partners," Gokul said. "These partnerships focus on producing robotic hands, arms, joints, skin sensors, optics, and specialized PCBs. CynLr is looking to expand its operations further east."