Robots & Humans Working Together
The vision driving Cognicept is to bring robots out of the lab.
Meet Michael Sayre
Co-founder, CEO @ Cognicept Systems
Singapore
Michael Sayre has two decades of experience with robotic and telerobotic systems. Having been both an R&D engineer focused on R&D, and a product manager focused on deployment, Michael has a deep understanding of the capabilities and limitations of robotic systems.
He drove the development and deployment strategies of warehouse systems, automation, and technology for RedMart and developed robotic and telerobotic systems for military and subsea applications.
What inspired you to pursue a career in robotics?
I have always been intrigued by machines, even as a child. One of my earliest jobs was working in a repair shop, fixing electronics like VCRs, slot machines, and even a piece of radar equipment once.
This leads naturally to an interest in robotics, which combines many disciplines such as mechanics, electronics, and computer science.
I also used to do many manual jobs in my youth, and I always felt like lugging heavy objects or doing mundane, repetitive tasks was a waste of a good human.
I started my career in telerobotics for deep ocean applications. When machine intelligence started making huge leaps in the early 2010s, I shifted away from telerobotics to focus more on autonomous robotics.
Explain Cognicept and the vision of Robots and Humans working together?
Before founding Cognicept, I was involved in deploying robotics technologies in an e-commerce warehouse and saw that the technology didn’t work in the field the way it works in the lab.
This is because we can’t make the world like the lab without massive spending on infrastructure and process changes.
This investment is never desirable and often impossible. Rather than a steady march to greatness, the past of robotics and AI is littered with examples of mundane practicalities preventing robots from delivering on the purpose and promise of technology.
If you want to know what the future really holds, it’s not the youtube highlight reels that matter – it’s the out-takes that should have our attention.
The vision driving Cognicept is to bring robots out of the lab and into the wild of warehouses, sidewalks, and hospitals and enable labor to be delivered via an internet connection.
Creating the human/robot intervention operating system will enable robots to do any manual job a human can do and many jobs humans cannot do. Why are factories so often located far from the markets that consume the items they produce?
It is because we want to go where the labor is less expensive. But what if you could buy labor in Hanoi for a factory in California?
Our technology empowers anyone to drop a robot in place of a human worker and hire someone anywhere in the world to guide it.
While some applications are limited due to internet communications latency, the widespread adoption of 5g will drastically improve the performance of our teleoperation technology and expand the value we can deliver.
Robots are guided by machine intelligence. This intelligence gets “confused” in unstructured environments and applications with too much variability.
Cognicept has built systems to allow these robots to “dial out” to human intelligence on demand.
Operators see these intervention requests, connect to the robot, assess the circumstances that caused the edge case, see the robot's environment and intention, and provide guidance and control to overcome the situation that caused the failure, getting the robot back to normal operation.
The adoption of smart robotics has lagged behind expectations, and these edge case failures are the reason. Cognicept will be instrumental in the coming robotics revolution.
Describe the Smart+ telerobotic technology and remote robot pilots?
At Cognicept, we provide hybrid robotics AI with our Smart+ telerobotic intervention technology and remote robot pilots.
We make it easier and more economical to deploy robots by bridging the gap between the capabilities of the technology and the needs of your customers. Our supervised autonomy tools and services make unpredictable applications reliable and expand the potential of your robots.
These tools are also useful during prototyping and testing during robotics development and will save a new team from developing these tools internally.
We make remote intervention software for robotic systems that enables humans to guide and operate any robot anywhere in the world.
When a robot gets confused, it dials out using our system. Our operator can then connect to the device, see what the robot sees, and guide it intuitively through the confusing scenario.
Getting it back to normal operation. This allows users to push robots into previously impossible applications by overcoming AI's limitations and allowing them to improve the reliability of already deployed systems.
How do you see the role of robots in a post-pandemic work world?
With the age wave cresting and labor shortages emerging in developed economies, the need for labor-saving and labor replacing solutions is extremely urgent.
However, the widespread deployment of robotics has been stymied due to many technical and economic challenges. And now, we are dealing with the emergence of a global pandemic.
This has precipitated a myriad of countermeasures such as travel restrictions, operational restrictions for businesses, and disruptions in global supply chains and labor availability.
Robotics are uniquely positioned to remedy many of these challenges. At Cognicept, we have even been involved in the emergence of new robotics applications such as remote disinfection and quarantine supply delivery.
Using robots for these applications allows us to avoid the spread of disease, as robots cannot become infected.
Explain one challenge robotics is facing for their use and development?
Considering all these forces, why don’t we see more robotics adoption? Traditionally, commercial applications of robotic systems have been limited to repetitive tasks in tightly controlled workspaces.
While robots have been reliable, precise, and cost-effective, they have not been competent in handling unpredictable inputs or situations. Robots are great for dull and exacting tasks, they don’t get bored, and they don’t get sloppy.
They're also great for anything dirty. Robots can handle heat, cold, and exposure to many things that normally hurt or kill a human.
Finally, robots are great for dangerous work. They don’t get scared, and if something goes wrong and destroys the robot, you can buy a new one.
No next of kin to inform and nobody to bury. But robots don’t understand some fundamental things about the world that people take for granted.
Recent advances in AI and machine intelligence, yielded largely by advances in GPU hardware, have greatly enhanced robotics capabilities. However, it is not uncommon for AI-based systems to have untenable failure rates.
These “edge case” failures often prevent robot users from taking advantage of autonomous systems. Human-in-the-loop systems have emerged as a leading solution to manage these edge cases.
However, due to connection latency and limitations in the operator interfaces, these solutions have been unwieldy, slow, and require high-skill operators.