The team likes to joke that ARTEMIS really stands for “A Robot That Exceeds Messi In Soccer.” (Image: RoMeLa at UCLA)

Not to be confused with any Lunar missions, ARTEMIS, the Advanced Robotic Technology for Enhanced Mobility and Improved Stability, was designed by researchers at UCLA’s Robotics and Mechanisms Laboratory (RoMeLa) as a general-purpose humanoid robot, with a focus on bipedal locomotion over uneven terrain.

Standing 4-foot-8 and weighing in at 85 pounds, it’s capable of walking on rough and unstable surfaces, as well as running and jumping. ARTEMIS is even able to remain steady when pushed or shoved.

During lab tests, ARTEMIS was clocked walking at 2.1 meters per second — making it the world’s fastest walking humanoid robot, the researchers noted. It’s also believed to be the first humanoid robot designed in an academic setting that is capable of running — and only the third such machine to do so overall.

ARTEMIS’ major innovation is that its actuators were custom-designed to behave like biological muscles; they’re springy and force-controlled, as opposed to the rigid, position-controlled actuators found on most robots. Another major ARTEMIS advantage is that its actuators are electrically driven, rather than controlled by hydraulics. As a result, it makes less noise, operates more efficiently than robots with hydraulic actuators, and is cleaner because hydraulic systems are notorious for leaking.

“That is the key behind its excellent balance while walking on uneven terrain and its ability to run — getting both feet off the ground while in motion,” said professor and RoMeLa director Dennis Hong. “This is a first-of-its-kind robot.”

ARTEMIS’ ability to respond and adapt to its senses comes from its system of sensors and actuators — it has custom-designed force sensors on each foot and an orientation unit and cameras in its head to help it perceive its surroundings.

ARTEMIS is scheduled to travel in July to France to take part in the soccer competition of the 2023 RoboCup — an international scientific meeting where robots demonstrate capabilities across a range of categories. To prepare ARTEMIS for the RoboCup, the team has been testing it around campus.

“We’re very excited to take ARTEMIS out for field testing here at UCLA and we see this as an opportunity to promote science, technology, engineering, and mathematics to a much wider audience,” said Hong.

It won’t be the group’s first RoboCup rodeo — RoMeLa has previously won the RoboCup five times throughout its two-plus decades of building humanoid robots. The team is hoping that ARTEMIS — named after the Greek goddess of the hunt, wild animals, chastity, and childbirth — will bring home trophy number six.

Here is a Tech Briefs interview, edited for length and clarity, with Hong.

Tech Briefs: What exactly inspired the research?

Hong: Let me start with the bigger picture; this is a true story. When I was seven years old, I watched Star Wars, the first one, for the first time at the theater when I was a kid, and I was completely mesmerized by the two droids — R2D2, the one that looks like a trashcan, and C3PO, the humanoid robot. On my way back home in the car, I told my mom and dad, ‘I’m going to become a robot scientist when I grow up,’ and I followed my dream and I’m here today. I think, subconsciously, that’s why I started working with humanoid robots.

We’ve created more than 40 different types of walking, climbing, jumping, rolling, chemically actuated, wheel-leg, hybrid robots, all different crazy stuff. But if you look at our robots carefully, you can actually see R2D2 and C3PO in there. So many of our robots have novel locomotion, completely new ways of moving with three legs, wheeling, hybrid, all different things.

Now talking about humanoid robots in particular, we’ve been doing such research for the past 20 years or so. We have a robot called Charlie, which is considered the United States’ very first full-size humanoid robot. ARTEMIS is our latest, greatest humanoid robot during the past 20 years of research.

I have a dream; in the future, I would like to be living with robots in our house — just like in the science fiction movies. Wouldn’t it be cool for the robots to be doing the dishes, taking out the trash, and all those kinds of things? I claim that the robot needs to have human shape and human size for that. The reason is this environment that we’re living in right now is designed for humans — your stairs are a certain size and shape for humans to walk up, or your door handle is at a certain height for humans to open.

Tech Briefs: What have been some of the biggest technical challenges you faced throughout the 20 years of making robots?

Hong: It might sound like a generic answer, but it’s really everything. You need better sensors, actuators, power, and sources; like the batteries only last for 30 to 50 minutes, how can you use them in real life? But to answer your question, I’ll focus on the key breakthrough technology that enabled ARTEMIS to perform so well.

The actuators are probably the most important breakthrough that we made. Actuators are devices that make things move. For humans or animals, these are muscles; for robots, typically, these are electric motors.

So, 99.9 percent of the electric robots that exist today use servo motors, which is basically a big motor with a bigger geartrain in the front to decrease the speed. These are great for precise, rigid motions … but they’re not the right type of actuators for leg locomotion. What we really needed were actuators that behave like biological muscles, which means something that’s more compliant or has springiness, and not only for position but also something that can control the force or pull control.

So, we needed those actuators, and we built our custom actuators that mimic the behavior of a biological muscle. I think that was the biggest technical challenge.

Tech Briefs: The team has been testing the robot with regular walks and soccer playing. How did that go and what did it entail?

Hong: Testing means everything. It’s not that you just build a robot, switch it on, and it starts walking. ARTEMIS fell thousands of times before it was able to take its first step. You test it, failure; you figure out what went wrong; we have this iterative process; so, it’s testing, testing, testing, and even now, although it’s walking very stably, you try to push the limit.

We shove it, push it, kick it; we make it walk outdoors without tethers. In robotics, that’s a huge, bold statement — making your robots walk outdoors without tethers. That’s the biggest thing. Every three weeks, we bring ARTEMIS outside on campus and it walks around. Students love it; they take selfies. ARTEMIS is really a celebrity.

There’s a competition called RoboCup — an international autonomous robot soccer competition where robots play soccer against each other. Universities and research labs from all around the world gather every year; it’s actually a real soccer match.

The cool thing about this RoboCup competition is its ultimate goal is, by the year 2050, to have a team of robots play soccer against the human World Cup champions and win. We’re five-time world champions.

But just because it can walk doesn’t mean that it can play autonomous soccer; so, we’re testing the perception system. In other words, the robot needs to figure out localization.

And even though it works well in the lab, that doesn’t mean that it’s going to work well outdoors. Once you take it out of the lab, everything changes — e.g., sunlight, glare, wind, objects, debris. ARTEMIS needs to be able enough to handle all these things. So that’s some of the testing that we’re doing.

Tech Briefs: Do you have any advice for engineers aiming to bring their ideas to fruition?

Hong: The first 10 years, things fall down, they break, things don’t work. We could have given up, but we did not because we truly believe in why we’re doing what we’re doing, so I think that’s very important. But one of the really important lessons that I teach our students in our lab is not to fear failure. Of course, failure is not a good thing, but everybody fails, especially in robotics. It's not going to work the first time all the time. But if you give up, then failure is the end. But if you learn from your mistakes, then that becomes the stepping stone for success.

In our lab, we have this culture where we embrace failure. As a matter of fact, if a student hasn’t failed at all, that’s not a good thing because that means that the student hasn’t pushed the limit. So, it’s OK to fail as long as you learn from the mistakes.