ARK Invest founder and CIO Cathie Wood and analyst Tasha Keeney featured Elon Musk in a recent episode of the firm’s weekly For Your Innovation podcast, where the Tesla CEO explained the rationale behind the electric car maker’s decision to develop a custom-built hardware solution for future Autopilot improvements and Full Self-Driving features.
During his conversation with the ARK Invest CIO and analyst, Musk reiterated a point he made last year, stating that everyone who purchased Full Self-Driving would be getting Tesla’s Hardware 3 upgrade for free. Musk also explained that as neural networks used for Autopilot and Full Self-Driving features get more advanced, it becomes more challenging to optimize software according to a given hardware’s limitations — something that Tesla is already employing in its current NVIDIA hardware. Musk was ultimately optimistic about the improvements offered by Hardware 3, stating that the custom-built solution has not “tapped out” yet despite intensive tests.
“If you want to have a complex neural network, you need to have a combination of software and hardware. And your software needs to be that much better in order to compensate for hardware’s (limitations). Sort of like, you have video games and how they’ve progressed — it’s a combination of software and hardware. No amount of clever software could produce a video game on old hardware that you have today. It doesn’t matter, you know. It’s the same thing with neural nets.
“So right now, we can process on the order of 100 frames a second and we really need to do a lot of work in terms of cropping the frames, and sort of bending the pixels, and not going to full resolution on all cameras, that kind of thing with the current hardware. We’re at full frames, full resolution with the Tesla hardware. All cameras, at full resolution, full frames, and it still hasn’t tapped out.”
Echoing Tesla’s statement back in 2016, Musk added that he believes Full Self-Driving is still attainable with the current NVIDIA hardware installed on the company’s electric cars, though the company would have to employ “tricks” to effectively manage the hardware’s computing power. Nevertheless, Musk maintained his point that autonomous driving systems would likely improve safety on the road. For example, even if Tesla ends up using its NVIDIA hardware for autonomous driving, the results would still be safer than a human driver.
“I do think that we can achieve full autonomy with the NVIDIA hardware, but it’s a much harder software problem. So really, you have to try to budget your compute and do all sorts of tricks to manage how you use your compute. So it’s a harder software problem. With hardware that’s 2,000% better, you don’t have to do that constant budgeting, and so the software problem is much easier. So I think with the current hardware and a lot of effort, we could get to Full Self Driving with maybe being like 50-100% safer than a person, but with Hardware 3, I think it’s probably like a thousand percent safer.”
Elon Musk also added that Tesla’s rollout of its upcoming Full Self-Driving features would ultimately be up to regulators. In this sense, Musk stated that he believes regulators would likely understand the case for autonomous driving and its advantages, provided that Tesla shows ample data. Considering the data that Tesla has gathered so far, Musk is quite optimistic.
“I think they will understand data, so if we show you know, billions of miles with a given safety level, then they will appreciate that. It’s not like saying, ‘Hey, we have this really fast computer and everything’s gonna work.’ It’s like, ‘Well, you know, that’s just a statement.’ But if you got hard data, billions of miles, and you could show the accident rates and intervention rates, and that it’s essentially unsafe if you don’t have Autopilot on, which I think is really unequivocal at this point. No matter how you slice the data, it is unequivocal at this point that it’s safer to have Autopilot on.”
It should be noted that Tesla’s Enhanced Autopilot is not an autonomous driving suite. Nevertheless, the driver-assist feature is already a notable safety system. In Q4 2018 alone, Tesla’s Quarterly Vehicle Safety Report noted that the company registered one accident for every 2.91 million miles driven while Autopilot was engaged, while one accident was recorded for every 1.58 million miles driven while Autopilot was deactivated. In comparison, the NHTSA’s most recent data shows that in the United States, there is an automobile crash every 436,000 miles.
Elon Musk’s recent segment in ARK Invest’s FYI podcast could be accessed here.