The future of humans is on Mars. Between SpaceX, Boeing, NASA, and every other national space program, we’re going to Mars. With this comes a problem: flying to Mars is relatively easy, but landing a large payload on the surface of another planet is orders of magnitude more difficult. Mars, in particular, is tricky: it has just enough atmosphere that you need to design around it, but not enough where we can use only parachutes to bring several tons down to the surface. On top of this, we’ll need to land our habitats and Tesla Roadsters inside a very small landing ellipse. Landing on Mars is hard and the brightest minds are working on it.
At this year’s Hackaday Superconference, we learned how hard landing on Mars is from Ara Kourchians (you may know him as [Arko]) and Steve Collins, engineers at the Jet Propulsion Laboratory in beautiful Pasadena. For the last few years, they’ve been working on COBALT, a technology demonstrator on how to use machine vision, fancy IMUs, and a host of sensors to land autonomously on alien worlds. You can check out the video of their Supercon talk below.
There are a few methods that have been used to land on Mars over the years. The first successful landing, Viking, in 1976, simply dropped the lander off at the top of the atmosphere with the hope of not landing on top of a gigantic boulder or on the side of a cliff. Curiosity, the car-sized rover that’s been going strong for half a decade, was a little more complex. The entry vehicle had an offset mass, and as the lander was plunging through the atmosphere, the computer could roll around its center of mass, imparting a little offset to its trajectory. This is also how the Apollo modules came back from the moon, and proof you can fly a brick, provided it doesn’t have a homogenous density.
But there are Mars rovers being built right now. The Mars 2020 rover is currently being assembled, and with that new landing techniques are needed to put the rover next to interesting geological formations. For Mars 2020, this means having the lander take pictures of the landing area during its descent through the atmosphere, compare those to maps created by one of the many Mars orbiters, and have the lander figure out if it’s going to land on a pile of rocks. If the lander senses it’s going to land in a dangerous area, it can divert its landing site a few hundred meters away towards safer terrain.
COBALT — the CoOperative Blending of Autonomous Landing Technologies — is a project to improve this technology. Eventually, we’re going to want to land on even more dangerous terrain on Mars or even Europa. These are challenging environments, and we don’t even have high-resolution maps of Europa. We probably won’t have high-resolution maps of Europa until we try to land there.
To manage this, COBALT is a payload package loaded up with LIDAR, cameras, IMUs, and a beefy computer providing real-time sensing for where a rocket will land. The COBALT team actually got a chance to test their payload out last spring in the Mojave aboard a Masten Xodiac rocket. This rocket shot upward, turned down its engine, then moved off to the side and landed on a pad a few hundred meters downrange.
This test was a complete success. You can check out a few videos of the test from the Armstrong Flight Research Center in the Mojave where the rocket goes up, figures out where it is, and directs the engine to a precise landing point.
There will be a lot of ways we’re going to land on Mars. SpaceX is going all-in with lifting bodies and offset centers of mass. Boeing will probably go Thrust or Bust. Who knows what China and India will do. We will eventually get there, though, and when it comes to worlds other than Mars or the moon, this is probably what we’ll be using.