Tim Hornyak

Sep 6th 2017

AI in the Space Travel Drivers’ Seat


How will artificial intelligence influence the future of space travel? If HAL 9000, the AI from “2001: A Space Odyssey,” were a space explorer, where would it go and what would it look for?

For decades, we Earthlings have sent probes out into the vast unknown of space to study, explore and document some of the great mysteries of the cosmos. These spacefaring robots have featured varying degrees of autonomy, though most have been teleoperated by people on their home planet. Recent probes, however, have shown that self-directed robots can do amazing things. The Cassini probe, for instance, was launched in 1997 to explore Saturn. It has been studying the ringed planet since 2004, but this year it has been executing its Grand Finale sequence of dives between Saturn and its innermost ring before it burns up in the Saturn atmosphere.

AI and the Future of Space Travel

Robotic space travel will evolve even further with AI technologies, such as machine learning and deep learning. In a recent paper published in Science Robotics, researchers argue that robotic spacecrafts will make more and more of their own decisions in deep space, improving mission efficiency and even accomplishing “otherwise impossible observations, such as responding to a short-lived plume at a comet millions of miles from Earth.”

Steve Chien and Kiri Wagstaff of NASA’s Jet Propulsion Laboratory (JPL) at the California Institute of Technology provide an overview of how space probes have developed over the years. In the future, they could be exploring remote parts of our solar system, such as Europa, one of the moons of Jupiter.

“With rapid innovation in space robotics, the possibility of exploring neighboring systems far, far away (such as Alpha Centauri, a voyage of approximately 60 years by robot), is on its way to becoming a reality,” the authors wrote in the paper.

“The goal is for AI to be more like a smart assistant collaborating with the scientist and less like programming assembly code,” Chien, a senior research scientist on autonomous space systems, was quoted as saying. “It allows scientists to focus on the ‘thinking’ things — analyzing and interpreting data — while robotic explorers search out features of interest.”

Evolving the Exploration Game

Aside from Cassini, there are a number of examples of how AI has helped drive robotic space travel and exploration. NASA’s Autonomous Sciencecraft Experiment on the Earth Observing-1 satellite features machine learning and pattern recognition functions to perform tasks such as analyzing images to detect phenomena that were not present in prior observations. The Opportunity and Curiosity rovers have used AI systems to autonomously identify Martian dust devils as well as targets (and fine-tuning those targets) for chemistry analysis using a laser; NASA’s Mars 2020 rover will also feature an AI system. The Mars 2022 Orbiter, a next-generation communications satellite that Northrop Grumman is working on, may also feature autonomous systems.

AI-driven space travel and exploration missions will require the ability to use the probe’s own observations to create models of what is “normal” for a given environment, such as a remote moon. When sensor data record something extraordinary, the spacecraft will have to detect that phenomenon, evaluate its significance and perhaps decide to devote more resources to it to learn more.

While artificial intelligence can save time as well as energy and storage resources for probes that are far from Earth, researchers don’t envision human operators being out of a job anytime soon.

“For the foreseeable future, there’s a strong role for high-level human direction,” says Wagstaff, a principal data scientist with JPL’s machine learning group. “But AI is an observational tool that allows us to study science that we couldn’t get otherwise.”