The Robots that Could Beat Optimus to Space
Before we get to our column, we should acknowledge OpenAI’s latest chip deal, this time with Broadcom, announced Monday morning. The Information first reported the two companies were in talks in July of last year but a lot has happened since then. Most obviously, OpenAI has also struck deals for supplies of AI chips with both Nvidia and AMD. We’ll write about the implications on this in the days and weeks to come…
Last week, my colleague Theo reported on the setbacks afflicting Tesla’s Optimus robots. Designing a humanoid robot involves some thorny engineering problems, especially when it comes to their five-fingered hands. Those issues could delay Elon Musk’s plans for Optimus to walk on Mars next year. Making matters worse, the robot has been designed for indoor use on Earth, where temperatures are much warmer and the air is less dusty.
But other robot-makers are also targeting space, and since they’re not wed to the humanoid form factor, they could potentially get there before Tesla.
For example, four-year-old startup Inversion recently announced Arc, its flagship delivery vehicle, which is sort of like a self-driving space Waymo. Arc is designed to drop off a trunk's worth of materials, starting with military supplies, to any place on Earth in an hour by traveling through orbit, which is faster than flying cargo on a plane through the atmosphere. Inversion already flew an earlier prototype this year on SpaceX’s Transporter-12 ride share mission, and aims to send Arc to space next year.
Since Arc has to rapidly adjust its flight path and land in remote areas, it relies on “autonomy, machine learning, computer vision and AI to make decisions and land precisely when there are no humans around to control it,” said Justin Fiaschetti, Inversion’s co-founder and CEO.
Startups also say that robots should help save precious hours of astronaut labor. Currently, when a load of cargo arrives on the International Space Station, it takes a team of four astronauts seven days to unload it and seven more days to repack the vehicle with waste and other materials to send back to Earth, according to Ethan Barajas, co-founder and CEO of one-year-old Icarus Robotics.
Icarus is building robots to unload supplies at space stations, starting with a year of testing on the International Space Station in 2027. Icarus’ robots have two arms with grippers on the end mounted on a drone base (you don’t need legs to walk around in space!).
“One hour of astronaut time costs $130,000,” Barajas. That’s time they could be spending on more valuable tasks, such as running scientific experiments.
Another key advantage over astronauts is that “robots are fine on one-way missions,” said Robert Ambrose, who formerly ran NASA’s software, robotics and simulation division. In fact, the Voyager spacecraft that recently passed Pluto can autonomously detect and respond to issues, making it a bit of a robot itself, he said.
(To be sure, it’s hard to imagine robots becoming capable enough to put astronauts out of work any time soon.)
Eventually, Ambrose expects people to buy one-way tickets to the moon, and then Mars, but in the meantime, only robots will sign up for those trips. “What we really ought to do is send robots first, and get some momentum going, get some deliveries there and get equipment in place,” he said. That way, when humans arrive, everything will be set up for them.
Other robotic space efforts are already underway. For example, two-year-old startup Crest Robotics recently unveiled a spiderlike robot intended for building moon bases. If Optimus does make it to space in a few years, it might find itself in good company.