This tech could be the secret ingredient to the mass adoption of robots

From the simple Roomba to cute room service robots in hotels that make puppy noises, smart robots are all around us now. But the success of the industry depends heavily on near-zero failure rates, and the high cost of development to reach those standards means that many organizations can’t fully realize the benefits of robotics.

Just ask Michael Sayre, a robotics engineer and product manager who had seen several projects falter because a solution couldn’t be found. The experience was frustrating enough for him to take a leap of faith and solve this problem with his own hands.

Cognicept founders (L to R): Ruchit Rami, Michael Sayre, Alok Pathak / Photo credit: Cognicept Systems

“Walking away from a good job and forgoing an income for an unknown amount of time was personally very difficult, as I have a son and family to take care of, school fees, and a mortgage to pay,” says the CEO of Cognicept Systems, who previously worked at Hope Technik and Redmart.

Sayre adds, “But the more I thought about it, and after speaking with many of my connections in the industry, I realized there is a major gap in the way robotics are deployed, and that this gap is pervasive across almost all smart robotics deployments.”

Putting humans back in the driver’s seat

Working primarily with industrial and service robots, Cognicept uses “human-in-the-loop,” an error-handling system that aims to cut failure rates to near zero.

“Our software sits on the robot and listens for error alerts. When they arise, it dials into one of our human operators, who can see what the robot sees, and can control it remotely through an internet connection,” says Sayre.

“If you have even a 10 percent failure rate for certain applications, that can be a really big problem,” he continues. “Take self-driving cars for example: Why don’t we already have those all over the roads? It’s because they aren’t perfect yet. They’re really close, but it’s still not good enough. The cost of failure is so high that there needs to be perfection.”

Having been a robotics engineer for most of his career, Sayre got the idea for Cognicept while working on a warehouse robotics project. “I thought it would be really good if there was a way to remotely intervene in these edge cases. Essentially, what we’re doing is stepping in when these robots get confused.”

He believes his product can increase the adoption of robots in various industries. “For example, a business attempting to deploy a robot may get a 25 percent failure rate, which isn’t good enough.” But if human intervention can mitigate error in the short run and improve a robot’s machine learning algorithms in the long run, the robot can exceed 99 percent reliability and become acceptable to deploy, explains Sayre.

Phantom Auto and Mavenoid are some of the startups in the same space as Cognicept. However, Sayre doesn’t consider Phantom Auto a direct competitor as it focuses on self-driving cars. As for Mavenoid, it’s more concerned with immediate troubleshooting rather than improving the long-term reliability of robots, he claims.

A rapidly growing market

On top of an existing market of warehouse and hotel robots, Sayre points to the projected growth of the smart robot industry to illustrate its potential. “The growth rate of robotics is, conservatively, 20 percent year-on-year. Some higher-end projections even put that figure at 50 percent. According to an International Federation of Robotics report I read, by 2022, we should be seeing about 2.25 million smart robots in the global marketplace.”

Cognicept aims to capture 10 percent of that global market. “Based on what we charge, that represents US$500 million in yearly revenues,” points out Sayre. Cognicept has a subscription model, and it’s also exploring flat annual billing that would cover all of a company’s robots.

Technical support for robots