Asteroid Defense is Computational — and that is great news!

Over the long march of biological and now technological evolution, we have finally reached a survival gate — we have enough computational power to model the trajectory all Near-Earth Objects (NEO’s) that could threaten life on Earth. This was not possible in the year 2000, or any time over the prior millennia. We have made a million-fold improvement in computation in just the past 20 years. So, we can see the future and predict decades in advance of an impact event and then give the NEO a nudge such that it misses Earth entirely.

It’s not like the movies, where you have an asteroid on final approach and try to blow it up somehow (that just turns a rifle into a shotgun blast); instead, you launch a rocket to rear-end it and change its velocity ever so slightly. Integrated over years, that small delta-v makes all the difference. In short, asteroid defense does not end with a bang, but merely a nudge. That is, if you know what you are doing!

The non-profit B612 (with co-fiounding astronauts Ed Lu and Rusty Schweickart) did a webinar and demo of their ADAM simulation tool for calculating asteroid orbit propagation. They gave me permission to share the unpublished work of their Asteroid Institute tech team. Here’s an unlisted video showing the sim seen here.

Rusty Schweickart, the first Lunar Module Pilot summarizes: “We live in a remarkable time in history. We can change the trajectory of the solar system, ever so slightly, and protect life on Earth”

Mapping the Final Frontier with ADAM (Asteroid Decision Analysis + Mapping):

The ADAM project runs on the Google Compute Engine to provide a cloud platform for large-scale orbital dynamics. Small errors in the initial velocity vector measurements can expand over decades to very different outcomes, especially when gravitational slingshots around the planets occur. So, they run thousands of Monte-Carlo simulations over an array of starting conditions, creating a distribution of points, as seen in the images here, some hitting Earth (red) or a near miss (green). The distribution of endpoints gives a probability of deep impact. As a heuristic patch to some insane computational complexity, we can calculate a probability for the long term, which narrows like a hurricane forecast cone to a certainly as time advances.

To reach an accuracy of a few kilometers over many decades, it’s not just the complexity of an n-body problem. They had to model effects such as the curvature of space-time due to General Relativity, the non-sphericity of the Sun, the gravitational asymmetry of the planets, moons and larger asteroids, as well as the non-isotropic thermal re-radiation from rotation of the asteroid.

So so the good news: we can do this today, and with each passing year of Moore’s Law, we can look further into the future, moving from decades to a 100 years. The further you can see, and the more precisely, the easier the nudge becomes.

For input to the model you just need a series of at least three sample points (but more is better). And we are about to get a whole lot better at that. Starting in 2022, LSST will observe ~600,000 asteroids every night, and discover new asteroids at 10X today’s rate. This will accentuate the computation-bounded problem of using this torrent of data.

There is something poetic about the computational defense of humanity. And something that rhymes with history. The Space Race of the 60s was won computationally, not by brute force heavy-lift, which would have favored the Soviets.

Survival is computational. Intelligence allows us to see the future.