What are the challenges of building a robot that operates millions of kilometers away on another planet? It has always fascinated me how NASA Jet Propulsion Laboratory (JPL) repeatedly builds cutting-edge autonomous robots and gets them to work beyond Earth, especially since the conditions of other planets are hard to recreate here. I got a taste of how much real-world testing matters at ETH Zurich, building an autonomous underwater drone swarm in the focus project SWARM and a high-speed, high-agility drone in my joint Bachelor’s thesis at the Autonomous Systems Lab (ASL). Many key problems only appeared once the full system was tested under real conditions. I spent three months in JPL’s Aerial Robotic Systems group this summer, where part of my work was contributing to a testing and integration campaign.

Ingenuity, the first helicopter on Mars, flew itself, but its waypoints and landing areas were planned on Earth. Future rotorcraft will need to find safe landing sites among rocks and slopes on their own, even in an emergency. With no GPS on Mars, pose is estimated with a single camera, a laser range finder and an Inertial Measurement Unit (IMU). As the drone moves, images from different positions form a stereo pair, producing a depth map that is used to find landing sites.

My task was to help bring the autonomy stack from working in simulation to working on our research hexacopter. Real-world tests revealed issues simulation didn’t anticipate, sometimes seemingly small ones: the dim lab light caused image noise that degraded the depth map, so we removed the neutral-density (ND) filter in these conditions. By the end, the hexacopter found landing sites on outdoor flights. Thanks to my drone-flying experience, I could pilot many of the tests. From a Swiss perspective, Pasadena comes close to a desert. Some days hit 39°C and we survived cold water and a fan. But working on something so interesting lets you forget the heat. Quite a shift from my last tests at ETH: −7°C and snow.

I also upgraded research drones and assessed whether a 4K rolling-shutter camera could complement the proven low-resolution global-shutter navigation camera to let a future Mars helicopter fly higher at the same ground resolution. Working with some of the same sensors and computer models expected to fly on Mars one day was inspiring.

On the weekends, together with other interns, I explored the beautiful landscape nearby. After an exchange semester in Singapore, LA was quite a change, to say the least: cars, streets and parking lots everywhere. On a drive to Orange County, Google Maps could barely fit all ten lanes on the screen. Healthy food was tricky, yet I already miss the unlimited soda fountains and In-N-Out. I was also surprised by how well Waymo’s autonomous taxis worked: they drove us around so quickly and confidently that it almost scared us.

I loved the mix of fascinating work and exploring a new country. I’d encourage anyone with the chance to work or study abroad to just go for it.

A big thanks to Roland Brockers and Preston Rogers for this opportunity and their guidance, and to ETH Zurich’s StudioX for their financial support.