Tesla and SpaceX CEO Elon Musk may be unwilling to be among the first people Mars, but he wants to get there eventually. It’s a little different with Tesla, though. He said Wednesday he’s going test the electric car’s self-driving Autopilot technology first-hand.
When asked recently about whether he’d be on the first SpaceX trips to Mars, Musk said he wouldn’t: “I’d have to have a really good succession plan because the likelihood of death is very high.” Musk was giving his long-awaited SpaceX-Mars presentation at the International Astronautical Congress. “I’d like to see my kids grow up and everything,” he said.
Musk is more confident in Tesla Autopilot. Each update to the system begins with closed-track tests. If those go well, Musk himself is one of a “very limited set of, kind of, alpha users,” he told reporters, after sharing details about a major autonomy update. “I usually have the latest software update about a day after our QA [Quality Assurance] team has it. And I’m personally testing every element of the car.”
His belief in autonomy is well-documented, but to jump inside a Tesla right after an update might be viewed as audacious, given the experimental nature of the technology. Though extremely rare — especially when compared to manual driving — Teslas do occasionally crash while on Autopilot (which is semi-autonomous driving, not fully autonomous driving). Regardless, he’s undeterred.
After the alpha tests, the update expands to “what we call an early access program, which is about a thousand customers distributed around the world,” he explained. These privileged customers are “technically savvy,” he said, and “are cognizant of potential issues.” Since they’re all over the globe, their experiences literally cover more ground than Tesla could on its own. “There are so many different environments throughout Earth that it’s just not possible for our team to cover them all,” he said.
Then, assuming those tests go well, the update goes out in “shadow mode” to to all operating Teslas. “Shadow mode means that the car’s not actually taking any action, but it is registering when it would take an action, and when it would not take an action.” This stage, it goes without saying, is invaluable. “Let’s say that somebody has an accident. You can look at the vehicle logs, and say, ‘Well, if the car had been in autonomous mode, that accident would have been avoided.’ That’s obviously a plus,” Musk said. In addition, you can study errant moves by the computer. If, while in shadow mode, the computer’s actions would’ve led to an accident, the engineers can study and correct those actions.
After enough time operating in the shadows, he explains, and after culling enough false positives and false negatives, engineers await one moment: “The point at which it’s unequivocal that turning on an autonomy feature would improve safety. That is the point at which we allow it to actually take action. Before that, we do not allow the computer to take action.”
So Musk is himself willing to trust the system to take the wheel before it’s unequivocally safer than a human. The first Mars journey’s risks, however, are too great. Let’s just hope the gods don’t like irony too much, and that they don’t judge Musk’s confidence in autonomy as hubristic.
Correction: An earlier version of this story stated that Musk wouldn’t go to Mars. He does want to visit, but perhaps not be one of the first to make the trip.