April 26, 2016      

Joanne Pransky, associate editor of Industrial Robot, recently interviewed Aldo Zini, president and CEO of Aethon Inc. While working on his BS in industrial engineering at the University of Pittsburgh, Zini became interested in using healthcare automation to make hospitals more efficient.

Zini later obtained a master’s in public management (health systems IT) from Carnegie Mellon University and became vice president of sales and marketing at Automated Healthcare. The company’s ROBOT-Rx was the first robotic medication-dispensing system, and it was acquired by McKesson Corp. for $67 million.

At TechRx, Zini was senior vice president of sales and marketing. NDC Corp. later bought that pharmacy software provider for more than $200 million.

In addition to developing methodologies to quantify the value of hospital automation, Zini owns several patents around medication dispensing technology. Pittsburgh-based Aethon’s TUG is a mobile autonomous robot with more than 450 installs worldwide.

This interview is available free to Robotics Business Review readers until May 31, 2016. Here’s a preview:

Pransky: How has the TUG developed over the years? Could you describe the different versions or major iterations of the TUG since its beginnings in 2002?

Zini: When we started the development, our goal was to make an affordable, reliable, and safe autonomous mobile robot. Back in 2002/2003, that had not been accomplished yet. There were some mobile robots that existed back then, but they were very expensive, costing hundreds of thousands of dollars to build; were technically complex; and really weren’t viable products.

Aldo Zini and Aethon's TUG

Aldo Zini with Aethon’s TUG robot.

Our goal was to try to simplify and make a robot that was affordable, meaning that it had a good return on investment [ROI]. That was our initial goal, and we set out to develop the robot. There was a lot that we didn’t know and a lot that we had to learn. We probably really didn’t have the first commercial version of the robot until about 2004.

That robot, while it worked OK, had a number of limitations. Some of those limitations had to do with the technology because laser scanners were very expensive back then, and the cost prohibited us from using them. We relied on other sensor modalities — a lidar scanner — that wasn’t as accurate and wasn’t as reliable as a laser scanner.

While we were able to build an affordable robot, it wasn’t as reliable and maybe not even as safe as we would have liked it to be back then, especially for a structured environment such as a hospital.

Pransky: When and what were the modifications to your second iteration?

Zini: There were two modifications we made. One had to do with just making it more reliable and safe. The other was adding functions and features that the market was telling us they wanted to see in a robot.

On the reliability and safety part, we continued to make the software more robust by developing a modified version of simultaneous localization and mapping (SLAM) with pattern-matching technology. We increased the number of sonars.

As the cost of laser scanners and other component robotic parts starting coming down over time, we were able to integrate more robust sensor modalities along with better navigational software. Those two things really enabled us to raise the bar to another level where the technology became very reliable and very safe.

Even to this day, there are not too many people that have an autonomous mobile robot. It was a very difficult engineering problem we were trying to solve, because this robot had to travel in a hospital — everywhere in the hospital — open doors, ride elevators, and do it 30, 40, 50, sometimes 80, 90 times a day, and do it alongside patients, visitors, doctors, nurses, clinicians, everybody.

We couldn’t afford one mistake. To get it to that level of reliability, accuracy and safety with that many people overseeing it, it literally took us about another four years, to at least 2008, maybe even 2009, to implement the technology.

Pransky: What new developments for TUGs are you planning for? And what new challenges are you facing?

Zini: We’ve been adding new functionalities to our robots over the last two or three years. One of the challenges we had to overcome is elevators. Elevators are very unique around the world, and every elevator is different. We had to develop our own elevator software and our own elevator controllers that were very simple to install and very inexpensive to operate, install, and maintain. A lot of work went into that.

A more recent capability that we’d been asked to add and is becoming very important in the industrial world is the ability to automatically pick up and drop off a cart or container without any infrastructure.

TUG robot with racks

The TUG robot has an exchange base platform to carry a multitude of racks, carts, or bins for use in healthcare or industrial facilities.

In other words, think about our mobile robot advancing toward a cart somewhere, finding the cart, picking it up automatically, taking it somewhere where it needs to be delivered, and dropping it off automatically, without any human intervention and without any tracks, beacons or wires. This is all autonomous.

That is another huge technology leap, because having the robot just go from Point A to Point B was hard. But having it go from Point A to Point B to have to pick something up automatically, take it somewhere and drop it off automatically, that was a whole level of complexity which had never been done.

We’ve been able to successfully implement it in several sites, including some very large manufacturing facilities, and I think we’re the only ones in the world right now with that capability. That was a big accomplishment, although we’re still limited to some extent on how we do it. We’re continuing to improve and perfect the technology.

Pransky: What is the biggest mistake/greatest lesson you’ve learned?

Zini: I would say that probably — and I think this is a mistake a lot of people make — we think we know more than we do. Until you actually get your technology being used in a real-world application, you never really have it done yet.

You can work on something in your lab and expend a ton of money and time on it night and day, but until you actually get it implemented somewhere where people are actually using it, you really don’t know “what you don’t know.”

I think one of the mistakes we probably made early on is we thought we had certain things done, and we all kind of high-fived and said, “Boy, we’ve got that done.” And then when we actually put it in use, it didn’t really work as we thought it would. We had to go back to the drawing board. But that’s the lesson you learn.

I think getting real users — in our case, nurses, doctors and technicians — to use the technology is where you really start understanding the features, functions, and what you need to provide in order to make it a viable, valuable product. That’s one of the things I’ve learned in my career. You can’t bypass that step, and you can’t push it too fast, either.

Pransky: What do you think Ph.D. and Master of Engineering students should be doing while in school to prepare them best for the commercial side of robotics?

Zini: When I got my master’s at Carnegie Mellon, I had the opportunity to meet and do some projects with some companies. I learned more from that than I did, I think, from the actual college courses. I would recommend that people who are getting their master’s and Ph.D.s really look for opportunities to work on projects with companies.

There are a number of companies that do internships, and it’s important to get out in the real world and actually see how businesses operate. I see a lot of engineers [who] don’t understand the business side of things. They’re very focused on the technology, and want to develop and work on it.

Here at Aethon, I’ve always said, “The technology is important, but we want to build a business.” To build a business, you’ve got to think of: Is it affordable? Is it reliable? Is it safe? Does it solve a problem? What problem are you actually solving? Does it provide a good ROI? How do you market it? How do you sell it? How do you have a very methodical process, etc.?

I’ve seen some very good robotic engineers with great ideas, but they didn’t know how to turn them into a business. They were too focused on just the technology. To the extent that you can get out in the real world, work with companies, and get that experience, I think that’s what I would recommend.

Click here to read the entire interview.

More Essential Interviews: