Scroll

We're going back to the Moon.

The US and China are both planning crewed landings in the next few years, and both want to build bases that last. Some of the best places to put those bases are underground, in caves called lava tubes that nobody has mapped yet.

I'm building SLAM, a rover that goes down into those caves and maps them.

The new race

A second space race

China wants astronauts on the Moon before 2030.

NASA's Artemis IV is planned for 2028: two astronauts near the south pole for about a week.

And this time the plan is to stay. NASA's Moon Base plan uses robots through 2029 to scout sites before crews show up, and China's ILRS is pitched as a long-term base with its own power and resource use. Both are headed for the south pole, where shadowed craters may hold water ice.

If people are going to live there, someone has to figure out where. I think the answer is inside the Moon's caves.

Why live in a cave · 1 of 4

Radiation

The Moon has no atmosphere and no magnetic field, so cosmic rays and solar storms hit the surface directly. Anyone living in a surface habitat takes the full dose.

A few meters of rock blocks most of it. A lava tube already has that rock overhead.

Why live in a cave · 2 of 4

Temperature

One lunar day is about 29.5 Earth days long. Near the equator the surface gets to around +120 °C in the afternoon and falls to around −170 °C during the two-week night, over and over.

Models based on orbiter data put the shaded floor of the Mare Tranquillitatis pit at about +17 °C, day and night. That means smaller heaters, smaller radiators and a lot less power just to keep things alive.

Why live in a cave · 3 of 4

Impacts and dust

Tiny meteoroids hit the surface all the time. Each impact, and each landing, kicks up sharp dust that grinds down seals, joints, lenses and radiators.

Underground, the roof takes the hits and much less dust gets in, so equipment lasts longer.

Why live in a cave · 4 of 4

The space is already there

To shield a base on the surface you either launch the shielding from Earth or have robots bury it in regolith. Both take a lot of mass, power and time.

A lava tube already has the walls and the roof, so that mass can go to life support, science or more crew instead.

But none of that helps until you know how big the cave is, how to get in and what's down there.

They exist

There are holes in the Moon

Lunar Reconnaissance Orbiter images show more than 200 pits. A few of them, out in the dark lava plains, look like collapsed skylights into lava tubes.

In 2024, researchers went back through radar data from 2010 and found a cave passage running sideways off the Mare Tranquillitatis pit. It was the first direct evidence of a cave you could actually get into.

The problem

No one has been inside one

From orbit you get a hole, a shadow and a bit of lit wall. You can't tell how far the cave goes, how high the ceiling is, whether you can drive on the floor or whether a habitat would fit.

Until someone answers those questions, nobody is going to plan a mission or design a habitat around a cave.

That's the gap SLAM is meant to fill.

My solution · SLAM

Lower a mapping rover into the cave on a tether

A surface station lands near the edge of the pit. It has the solar panels, the radio link to Earth, the data storage, the tether reel and a fixed reference point for the whole map.

Descent

The rover gets lowered over the edge on a tether that carries power, data and its weight.

The reel tracks how much line is out and how much tension is on it, so the station knows how deep the rover is and if the line gets caught.

Autonomous navigation

It drives itself with a single end-to-end neural network.

Raw data from the 3D LiDAR, IMU and wheel odometry goes into the network, and steering and speed come out. There are no hand-written driving rules in between and no one steering from Earth. The network reads the cave, keeps track of where the rover is and chooses its own path.

Keeping the data safe

Each scan goes up the tether as soon as it's taken.

If the rover gets stuck or dies, everything it mapped up to that point is already at the surface. Even a short run gives you a usable map.

The device

The rover

It's small enough to fit down a skylight and low and wide enough to handle a rubble floor. All the mapping happens on board, and the data goes up the tether.

It has four grousered wheels for climbing over rubble. Underneath is an alpha particle X-ray spectrometer that measures the elements in the ground as it drives, so the map also shows what the cave is made of.

The map

Like Street View, but for lunar caves

All the scans are combined into one 3D model of the cave. It includes the cave's shape, the path the rover took, which parts of the floor are drivable, a fixed reference frame and a confidence score for each measurement.

Here's how people would use it.

Use 1 · Select

Pick the right cave

Mission planners can compare caves by ceiling height, tightest squeeze, floor slope, rockfall risk and usable space, each with an error range.

A bad site gets ruled out on Earth, not after landing.

Use 2 · Design

Design the habitat to fit

Engineers can fit habitat modules to the real walls instead of guessing at the shape of a tube. They can plan power and comms runs from the skylight, where anchors and lights go, and an emergency exit, all before anything launches.

Use 3 · Operate

Run the base from it

Rovers and crews can navigate with the map. New scans show rockfall and other changes over time. Scientists can use it to decide where to point radar, spectrometers and radiation sensors, and to look for ice and other resources.

The map is what lets people make these decisions before anything launches.

What comes next

Then Mars

Every cave it maps is more training data and real-world testing for the whole system: the driving network, the tether handling and turning raw scans into a finished map.

By the end of the lunar missions, SLAM should be a fully trained framework that can go to other planets. Mars is the obvious next step. Orbiters have found hundreds of possible caves there, and lava tubes are some of the best places to look for signs of past life.

The Moon first, then Mars.

Before anyone can build underground, someone has to map it.

That's what SLAM is for.

Subsurface Lunar Atlas Mission