Here’s how the SPHERE drone did it despite all those difficulties: first, it zoomed around the station’s Japanese module using its 12 gas thrusters, recording everything in sight with two cameras. Once it was done learning distances between objects after that two-eyed run, it was able to navigate on its own without bumping into things with only a single camera open. Before all these, though, the team tested their learning software on a quadcopter in sets they built at the Delft University of Technology.
Dario Izzo, the project coordinator from the European Space Agency, said his team have been working towards this goal for five years:
“It was very exciting to see a drone in space learning using cutting-edge artificial intelligence methods for the very first time.
At ESA, in particular in our team, we’ve been working towards the goal for the last five years. In space applications, machine learning is not considered a reliable approach to autonomy: a ‘bad’ learning approach may result in a catastrophic failure of the entire mission.”
They recently presented their results at the International Astronautical Congress in Mexico. You can read their paper if you want to learn every detail about the experiment, or watch the video below for a more visual explanation.