Joanne Pransky, associate editor of Industrial Robot, talked recently with Mitchell Weiss, chief technology officer of Seegrid Corp. Weiss has experience with developing industrial automation systems, building successful startups, and protecting IP.
Weiss has a Bachelor of Science from the Massachusetts Institute of Technology and a graduate certificate in intellectual property (IP) from Northeastern University. He has taught at Pennsylvania State University and the University of Pennsylvania and lectured at MIT.
In addition, Weiss holds 24 patents, has served as an expert witness in IP lawsuits, and is a member of national committees on safety standards for automatic guided industrial vehicles.
His first job was at Unimate, the world’s first robot company. Weiss has been an executive at Brooks Automation, PRI Automation, ProgramMation Inc., and United States Robots Inc. His current company, Seegrid, is a 2017 RBR50 member that makes vision-guided robots and vehicle-control systems.
This interview is available free to Robotics Business Review readers until April 30, 2017. Here’s a preview:
Pransky: You’ve worked for many robotic companies, you co-founded your own, and you’ve worked in academia. What position has been your favorite position, and why?
Weiss: While it’s always fun to be the boss, there are really two things that get me up in the morning. One is getting to design a machine from the system level.
The other one that gets me up is solving customer problems — applications of the machines.
I spend a lot of my time on the road and have done so since my first real job out of college, which was at Unimation. I was the first applications engineer on the Puma robot. …
I’ve been in factories that make transformers, potato chips, semiconductor chips, clothing, and cars. I tell young people who are interested in manufacturing that you want to be an applications engineer [AE] because you will learn more about manufacturing in that job in less time than you will anywhere else.
At this point, my jobs at Seegrid consist of the CTO job, which is the systems and future side. There is also the application-support job, in which I get to go out with our AEs. We wander a factory.
I tell them that within a couple hours, you need to be able to say, “That, that, that, and that is being done wrong, and this is being done right.” You have to very quickly eyeball where the problems are and where the benefits are, and you learn a lot about business.
Pransky: How long did it take Seegrid to commercialize evidence and stereo vision technology, and what have been some of Seegrid’s challenges as a company?
Weiss: One of the big challenges was taking something out of a research lab and turning it into a repeatable, producible, manufacturable product.
You can have an evidence grid that’s fixed to the world, which is your basic map that the vehicle’s traveling through. You could also have an evidence grid that’s fixed to the vehicle that tells you about what’s changing in its surroundings.
Seegrid’s sensors are built using stereo cameras that utilize imagers for things like you see in automobile blind-spot detection or in cell phones. Until just a couple of years ago, we were still fighting with how to produce stereo cameras in volume reliably and sustainably and at a low enough price.
Pransky: Are you licensing your technology IP in other products not owned by Seegrid?
Weiss: We did sell our navigation suite instantiated in hardware to other companies, for example to Raymond Corp.’s Courier line of industrial trucks.
Our experience tells us, and our belief is, industrial environments change too much [to rely on one thing]. For example, lighting conditions such as whether we’re working in a facility in the morning or at night, and the dock doors are open or the sky lights are lit up, or features in the environment change.
These dynamic variables must be taken into consideration, and we think our stereo vision processing combined with evidence grids … provides the best way to get information into a good, robust 3D model about the environment.
The No. 1 benefit that we bring to companies with VGVs [vision-guided vehicles] is reducing the cost of operating a forklift, of which over 70% or 80% is the driver.
Additionally, the No. 1 killer in industrial accidents is forklifts. Every three days, somebody is killed by or on a forklift. It’s over 100 people a year, typically.
Pransky: What advice do you have for robotics startups regarding patents and IP?
Weiss: Patents are obtained for a couple of reasons. One is that they’re important to your early investors who want to see that at least you have something novel. That was my strategy at the last couple of companies I was at, one of which had no patents, so I set up the whole [IP] process and took us to around 50 patents.
While you need to take your basic ideas and protect them to demonstrate to your investors that there’s some value there, the patent may not ultimately protect you from keeping your competitors out of the business. Should you get into a misunderstanding with a competitive or other company, you now have a stack of assets to buffer your risk because you can trade that stack off.
Patents potentially also have some value as technology you can license out. If you look at the phone wars and the camera wars that have gone on, it all ended up being trading assets.
I think America’s probably the best place to have patents. We file in the U.S. Then, we file a Patent Cooperation Treaty [PCT], which covers a lot of other countries. Then we file in the European Union at the European Patent Office.
Under the PCT, we get Canada and Hong Kong, and we’re selective with which countries, because you have to pay fees for every single country. And those fees are ongoing every three, sometimes seven, years.
In China, we have very little expectations of what our patents will do for us, and we’re very limited in what we patent there.
More on Vision-Guided Vehicles and IP:
- MobilEye Is Intel’s Latest Self-Driving Car Buy, Gamble
- Autonomous Cars Accelerate Toward the Future
- Connected Cars Yield Useful Data for Analysis Through Xevo
- Robotics, AI, and Automation Transform the Workplace
- Machine Vision Investments Eye Safety, New Apps
- U.S. Manufacturing, Logistics Look to New Robotics Roadmap
- Seegrid Scores $12 Million From Giant Eagle
- AGVs Get an ETA With Seegrid’s Subway Platform
- Researchers Tighten Focus on Robot Vision
- Intellectual Property Considerations for the Robotics Industry
Pransky: What do you think Ph.D. and Master’s of Engineering students should do while in school to prepare for the commercial side of robots?
Weiss: I think they need to learn about the domain they want to work in. I think that’s a huge missing piece of what I see. If you want to be in industrial robots, you better learn about manufacturing. That’s what drives the needs.
I think the biggest problem academic researchers and engineers make is that they find a great solution, but they don’t know what the problem is.
In the class that I was involved teaching at MIT, the classic design course, the first thing we talked to the students about is the problem definition. The problem definition is so critically important to good design.
In fact, I argued that I’m not a particularly clever inventor, because the answer is told to me by looking at the problem clearly. There’s only one or two ways to solve that thing.
Regardless of the problem the customer is facing, whether it’s flexibility, usability or how to put a peg in a hole or whatever, there are only so many solutions to it. But if you come up with a gizmo and then don’t figure out what applications are there for it, you’re not going to be all that successful.
Additionally, I think the universities and programs are very siloed. They’re getting better at it than they were when we were students, but I still don’t think they’re experiential and broadly based enough. I think that when you get to the master’s and the Ph.D. level, it even gets narrower.
More Essential Interviews:
- Gianmarco Veruggio, Telerobotics and ‘Roboethics’ Pioneer
- Rob Buckingham, UK Atomic Energy Authority Director
- Jacob Rosen, Surgical Robotics Pioneer
- William ‘Red’ Whittaker, Field Robotics Pioneer and Entrepreneur
- Aldo Zini, President and CEO of Aethon
- Steve Cousins, CEO of Savioke
- Yulun Wang, Founder of InTouch Health
- Robert Ambrose, NASA Researcher
- Helen Greiner, CEO of CyPhy Works
- Esben Østergaard, CTO of Universal Robots
- Rodney Brooks, Founder of Rethink Robotics
- Nest’s Yoky Matsuoka
- iRobot’s Paolo Pirjanian
- When Bill Townsend Met Burton Doo
- Mark Tilden, Inventor of BEAM Robotics
- Rich Walker, Shadow Robot Company
- Russ Angold, Co-Founder, Ekso Bionics