**Hiro Ono:** There are many reasons that this is challenging, but I would say the greatest challenge is the uncertainty, the unknowns. We’ve sent 10 spacecraft to Mars, knowing a lot beforehand about the regions where we landed and explored. But the best resolution we have on Enceladus is about 6 meters (20 feet) per pixel and only around very limited regions. We don’t know, for example, what the surface topography is or what the geometry of the vents is. We still don’t know how strong the jets are. It’s really, really hard to design an explorer without knowing what the environment is.
That’s why EELS is so different from rovers, in that its snakelike design gives it many ways of moving around. If something unexpected happens, it can change the mode of locomotion and figure out the best way to interact with the environment.
**SAA:** There are two things going on here. One is the actual physical design of this object, but there’s also the brain behind this technology. EELS uses artificial intelligence to figure out how to move around its environment based on the conditions it encounters. Hiro, you’ve spoken before about the idea of “Space Exploration 3.0.” Can you talk a little bit about this framework for understanding space exploration and how this technology represents a new phase forward?
**HO:** The way we explore planetary surfaces has changed over time. In the ’60s, NASA sent a bunch of spacecraft to the Moon as a precursor to Apollo. It was basically trial and error to learn what worked. And that was a reasonable way to learn how to get to the Moon because the Moon is nearby. You can get there in three days. So we could call that “Robotic Space Exploration 1.0”: trial and error.
Mars is six to seven months away, and the launch window opens up only every 26 months. Now you cannot do trial and error and wish for the best, right? You need to be sure that what you’re going to do will work, so we changed the mode.
We got more cautious; we incrementally learned about the environment and refined our capabilities, starting with flybys, orbiters, landers, and then rovers. Then, we knew enough to build Curiosity and Perseverance — super complex robots — and now we are designing an even more complex Mars Sample Return campaign. That’s “Robotic Space Exploration 2.0.” And that’s how we were so successful on Mars in the past few decades.
But we can’t extrapolate that model to the outer Solar System, simply because it takes much too long to get to those destinations. So that’s where “Robotic Space Exploration 3.0” comes in. It involves a more intelligent, more adaptive robot capable of diving into a highly unknown environment — a robot that can learn by itself \[and\] adapt by itself to robustly explore the environment.