by Paul J. Heney, Editorial Director
UCSD Robotics guru Thomas Bewley talks about how technology will evolve in the coming years—and how engineers can better lead the way.
If you spot Thomas Bewley at a trade show, chances are there will be a crowd around him. In addition to being a gregarious, approachable person, Bewley is eager to talk about the multitude of little robotic friends you’ll likely see scurrying around his feet. And people want to see them, pick them up and play with them.
One of those robots is quite recognizable—MiP was one of the big toy hits of Christmas 2014, won the 2015 Innovative Toy Of The Year award, and was featured on the cover of this magazine 15 months ago. Bewley, an engineering professor at the University of California San Diego, leads the UCSD Flow Control & Coordinated Robotics Labs, which developed MiP and other robotic technologies, and licenses with high-tech toymaker WowWee.
For our annual leadership issue, Design World sat down with Bewley to talk with him about what’s next in robotics, what challenges he sees facing engineers who want to grow into leadership positions, and where engineering education is headed.
In 2004, Bewley gave a final exam about a self-righting Segway-like vehicle that could hop. It had some internal plunger type mechanism inside and the students developed control algorithms for stabilizing the Segway-like motion and for stabilizing the Pogo stick-like motion.
“I thought they would solve these control problems as final exam questions and that would be it, we would have solved the problem,” he said. “But three of the students came up to me after that exam and said they actually wanted to build it—so that was the start of the robotics lab in the mechanical engineering department at UCSD. We started with some naïve designs and quickly refined from there.”
Along the way, they patented a few designs for managing the energy flow and the hopping mechanism.
“Hopping is interesting from a dynamics and control perspective—but is not energetically practical in a 1 g environment. It’s more practical on the moon; take a look at the Apollo astronauts, they hop to go from point to point,” Bewley said. “In a 1 g environment, it’s more efficient to roll. We quickly moved toward focusing more exclusively on different types of rolling. We built hacked-up toys, we built small, Segway-like vehicles.”
Bewley and his team were focusing on making the vehicles smaller with more modern technology. Then, through the tech transfer office at UCSD, he approached a couple toy companies and struck out.
“We had a design that was in, what I called the drawer of shame, for a couple of years. A design that was one of these little toy-grade Segway-like vehicles that could pick up and throw a ping pong ball, and it did a wimpy little throw. Then, a talented masters student, about the time that low-cost 3D printing was starting to become available, re-designed the whole thing and designed a mechanism such that when you drive over a ping pong ball, just by the motion of the wheel, pinching the ball between the moving wheel and the body would pick up the ping pong ball and put it in a frame.”
“We had a little cage that could hold a few ping pong balls, and then one by one, you could lay back and throw it in a lacrosse-style motion. That prototype was called iFling. We put together a little video on it and Gizmodo picked it up, and in a single weekend, we had like 30,000 views and it caught some attention. We went back to one of these toy companies, WowWee. We resonated with them and shared with them some of our ideas, and we decided to make these low cost Segway-like vehicles. The WowWee people are very good at embedding character and a mode in which you interact with such a toy.”
It was also a matter of timing for his team. Low-energy Bluetooth (BLE) was just becoming available, so users could engage with a low-cost robotic vehicle with their smartphone or tablet.
Not all fun and games
Bewley stressed that taking a development like MiP and making it a reality was difficult at times.
“We spent a lot of time working with them on a bunch of questions about designing for low-cost manufacturing—how can we shave off a quarter here and 50 cents there?” Bewley said. “Those questions are kind of tedious, so it’s not all glitz and fun—some of it is arduous. Teaching somebody who doesn’t speak your language the control theory that you know, so you can put the thing together [was a challenge]. And then you really have to whittle it down to its essence so you can minimize the materials cost.”
He said that he has no regrets and described the process as fun and interesting. But he also noted that, at first, they failed to resonate with the toy business because his team simply didn’t understand the industry.
“We were building these big things out of aluminum that were expensive and could hurt someone. It was really, to be fair, before 3D printing was readily available. So now, it’s easier to do these things,” Bewley said. “The three most important parts of that business are cost, cost and cost. You have to have a compelling play-spec, but you can’t have 18 motors. You have to get the most enjoyable play from the least number of motors and it has to be robust. Three-year-olds will throw these things across the room in temper tantrums, and so these things are drop tested from a significant height, hundreds of times … and they have to survive that. The robustness test is severe.
“In MiP you’ll notice that there is just the right amount of play in the wheels, there’s load bearing right on the output shaft. So if it takes an impact, you’re not putting load directly on the gears or the motor shaft. That sort of thing is what the FAB facilities that WowWee works with do exquisitely well—figuring out the right amount of shock absorption so this thing can handle the abuses of a kid.”
Bewley is now working on packaging a similar robot in an educational way, keeping the robustness of the base. He doesn’t want to simply bolt wheels on the output shaft of the motor the way they started, as that design is very fragile. And the last thing he wants is to frustrate students he’s trying to teach.
The coming tech
One of the major projects that the lab is working on right now is autonomous stair climbing. With the emergence of advanced smartphone-grade technology, Bewley said they can do autonomous scene recognition better.
“We want to be able to say: Here’s a staircase, figure out the rise and run of the staircase and your pose in front of it,” Bewley said.
He envisions a robot that would drive in a Segway-like motion up to a staircase, then employ a third motor to lift the vehicle up a post that goes through the center of the vehicle. As the two wheels and the body rise up to the top of the post, it uses the wheels as reaction wheels to stay balanced. Once it gets to the top, it turns the wheels one way, the body reacts the other way, it leans onto the obstacle and then does an end-over-end—sort of an inverse slinky maneuver.
He said that a firefighter could unleash a robot like this if, say, there was a fire at a big box store.
“The first firefighter on the scene could release a handful of little vehicles of this type at the front door and say, ‘Map it. Give me situational awareness. I want to know a map of where all the aisles are. I want to know where the people are. Where the hot spots are and where any dangerous fumes might be,’” said Bewley. “Send a bunch of vehicles and they solve all the small problems that are needed from there. Collision avoidance and resource management, figuring out there’s a stairway up to the 2nd floor, so maybe move more vehicles up there.”
They are also working hard on autonomous deployment of under-door camera systems. These are tactical systems that are deployed by marines and police forces to figure out what’s going on before you go into a room. Bewley describes them as being like a stack of credit cards that has a small cellphone-grade camera—which can be slid under any standard door.
“When trying to clear a building, one of the important things is to make sure that there isn’t anyone in the offices and that’s one thing you can do, is you can shove something under the door and you can sniff for things and you can spray the room with a light from a small LED and see what you see back there,” he said. “Another technology that can be deployed with an under-door system like that, besides a camera, is an electronic nose. There is a development that’s using a nano-technology so some small carbon nanotubes with some detectors placed beneath that can sniff out different chemicals. [This technology] has a certain signature if exposed to butane or propane or methane or other VOCs. You expose it to each of these VOCs or carbon monoxide or whatever it is that you’re worried about, and you get a fingerprint for that chemical of how the surface responds. Then if you go into an area where there’s a mixture of these compounds, you can say, ‘Okay, I see 0.3% of CO. I see trace amounts of butane,’ or whatever it is. It’s a technology that, once mass-marketed, can be as inexpensive as the accelerometer and gyros that are in every smartphone today.”
The crossover to industry
Bewley feels that the industrial side of things will continue to benefit greatly from a spin up of consumer grade technology.
“Accels and gyros and GPS are cheap because they’ve been mass marketed for smartphones and cars—and because of that, the industrial space can directly benefit for self-driving harvesters and combines and things like that,” he said. “As electronic noses become mass produced for wearable technologies, they will be directly useful in monitoring industrial environments, for example, to make sure that the atmosphere is safe. I think we will see a significant use of little mobile vehicles running around in security situations, especially once they can efficiently climb up and down stairs.
“I keep on coming back to steps because they are pervasive in environments built for humans, which most of the environments that we have in our target market are. Steps and closed doors are the two main problems, and we’re well on our way to solving the steps problem with small vehicles. So far, our approach for addressing closed doors is just to peek underneath. It takes a much bigger vehicle to actually get up to a doorknob and manipulate a doorknob or get a skeleton key or something.”
Precision agriculture will be another huge growth industry because of these coming technological advances, said Bewley.
“It’s a multi-billion dollar market where people are trying to figure out how to address the precision delivery of fertilizer, of pesticide, and of water to plants individually, on a day-to-day basis,” he said. “Eventually, it will be the robotic harvesting of soft fruits like strawberries and oranges.”
Engineers as leaders and teammates
Bewley also touched on the challenges of having engineers move into leadership positions—many engineering students are taught the technical side but not always the people skills to succeed as managers or visionaries.
“A lot of those [people] skills, you learn in your job,” he said. “Formal education is only a starting point of your actual education. You see inspiring leaders throughout your career and you’ll learn leadership skills throughout your career. I’m very much an advocate of professional education where these components come in. Your education is only just starting by the time you get to your last degree.”
Bewley said that what is outdated in the university space is the departmental structure.
“Having a mechanical, an aerospace department, an electrical department, a CS department, that are all separate, is historical and unnatural in the day that everything is interdisciplinary,” he said. “You need to focus on the subsystems of the complicated project that you’re building, but as you move toward a leadership role and a system development, you need to have an understanding of the complexities and the state-of-the-art in the other component disciplines to know the right questions to ask of your team who’s putting it together. Interdisciplinary education is very important.
“I think senior engineers with good people skills can make effective managers, because they understand, especially in a technology-heavy field like robotics, what key components that are available and emerging need to be developed to get a certain system together. You understand the component technologies that need to come together to make things … I find many university labs are focusing on narrow research questions. Make one-off prototypes that never make an impact in industry or sit on the shelf for a long time until someone else, or some other team, picks it up and figures out how to make it relevant in the market place,” Bewley said.
Bewley stressed that what management can provide that the engineer doesn’t always see is the market scope and market reality in terms of price points that work.
“Sometimes engineers get so focused on their build, their technology, that they don’t have a larger view toward what the market needs right now. This is something that I’ve learned to appreciate, working with my collaborators … that sense toward the several different aspects of what works in the marketplace and for the supply chain, so it can be delivered to the marketplace in large numbers, effectively,” he said.
“When we started [MiP], I thought, yeah, sure, we can make a Kickstarter project that sold a thousand units. But how do we sell a million?” he said. “That’s a completely, completely different thing. Working with a team that has the supply-chain connections that can provide volume and availability of a quality product, is really a team-focused experience. I think that engineers need to appreciate that their skills take them so far, and if they really want to make a large impact, reaching out to teams with complementary skills is absolutely necessary.”