NASA Is Doing Solar System Exploration All Wrong

In a recent interview, Dr. Dante Lauretta, the principal investigator of the OSIRIS-REx asteroid sample return mission said,

“Oh my goodness, September 24, 2023, that was the culmination of almost 20 years of effort by me and by an enormous team of people from all over the world to make this a success…I could hear the range command officer making the calls as they were tracking the capsule passing through the atmosphere…And no chute was called…And then, thank goodness, three minutes later, the main chute deployed, stopped that downward motion, and we came in for a soft landing.”

Twenty years of work. One shot. One failure could have killed the whole thing. Luckily, the thing completed its mission and made it back intact.

Next consider the Mars Sample Return project. The projected cost is between $8 billion and $11 billion and plagued by cost overruns. But let’s consider what it has to do:

1. Get a car-sized machine to the surface of Mars, intact.
2. Move around and pick up and store sample containers that have been left there by a previous mission.
3. Launch a rocket from the surface, carrying those containers to orbit.
4. Launch back to Earth, from Mars orbit.
5. Land safely.

But wait – all these things have been done before. Even number 2: we have off-the-shelf robots today that can do that. The most advanced is Boston Dynamics’s “Atlas”.

So if we have done all these things before, or can do them with off-the-shelf systems, why is the Mars Sample Return mission projected to cost more than $8 billion?

The reason: everything is a one-off custom design, created with the same “spacecraft-grade” processes – i.e. very custom, precisely controlled, and 100-fold higher cost.

Need a hammer? We have to design one from scratch, and we’ll make it of titanium because it’s for a spacecraft. Need a nail? We’ll put an MIT PhD on it and custom-design it, and then we’ll test it carefully for a year.

If we continue to do things that way, we will never be able to move beyond “baby steps” in exploring our solar system. We’ll spend the next century in one-off projects and endless tests to “see if we can do it” instead of actually doing it.

An Alternative

A much better way is to go commodity. Use what we have, and send lots of missions. Instead of carefully designing the perfect system, send multiple general purpose systems. Mass produce them. If we lose one, it doesn’t matter, because it was relatively cheap, and more are on their way.

Thus, instead of custom-designing a robot exactly for the task of picking up samples on the surface of Mars, use an already available multi-purpose semi-autonomous robot. If it is not “space rated” it doesn’t matter, because it will not be a single point of failure, since we will send lots of them.

Most complex systems today are over-the-wire reprogrammable. Your car’s computers are. Your smartphone is. The robots that we send to space must be. That way, if we find a problem and lose a system, we can remotely reprogram the others that are on their way, so that they avoid the problem that the previous one encountered.

Launch multiple systems in series, spaced a month apart. If a system fails, reprogram the ones that are coming.

Reuse the basic design of previous Mars landers, but commoditize it: make it mass-producible. Here’s one way to do that: challenge companies like SpaceX to take the design and create a version that they can sell back to NASA and other customers: the companies will be incentivized to find ways to make it cheaply and repeatably. That’s the model pioneered by the NASA COTS program for manned exploration. It works. Let’s use it for planetary science as well.

Use Falcon Heavy rockets. They exist, can launch very large payloads, and they are mostly reusable. When the SpaceX “Starship” becomes available, use that, because it is 100% reusable and therefore drastically cheaper with fast turnaround. (Blue Origin will have a similar system eventually, as will Relativity Space.)

And stop custom-designing the perfect system for every mission.

Today, if we want to send biologists into the field, do we custom design the vehicle that they will use? No – we obtain a Land Rover and use that. And if those biologists want to do surveillance on the forest canopy, do we custom-design a flying vehicle for that? No – we use off-the-shelf drones – many of them – and attach our equipment to those. It’s the same idea.

Go commodity, and send a fleet – not a custom-designed one-off system that represents one big and hugely expensive collection of single-points-of-failure.