Tesla released its blog post about the Autopilot’s new radar processing capability today and CEO Elon Musk made himself available for a 1-hour long press conference during which he explained in more details the upgrade and took a few questions from reporters.
The talk was more technical than usual and Musk gave a few great examples for applications of the Autopilot’s new radar processing technology. And since it’s technical and going in details about the Autopilot systems, including a lot of details unannounced or unconfirmed until now, I think it’s worth making a transcript of his talk for future reference.
I will publish the transcript in parts as I complete them. I’ll try to have everything on Electrek by Monday night. We are starting with Musk’s opening statement and then the Q&A:
Elon Musk – Tesla CEO:
I assume everyone received the blog piece that I wrote. I’ll just give it a basic overview. I think the real exciting thing is that we are making a much more effective use of radar. Where previously the use of the radar in the car was as a supplementary system to the vision system. Essentially because we weren’t confident that we could resolve false positives – like with radar we think that it should brake but it shouldn’t.
But after a lot of analysis and getting some upgraded drivers from our supplier for the radar to expose more raw functionality we now believe that we can combine that with fleet learning and almost entirely eliminate the false positive – the false braking events – and enable the car to initiate braking no matter what the object is as long as it is not large and fluffy.
But anything large, or metallic, or dense, we are confident that the radar system in the car will be able to detect that and initiate a braking event. Both when the Autopilot is active and when it is not active. When the Autopilot is not active, it will operate in an emergency braking mode. In that case, it’s more likely to mitigate the impact speed because if Autosteer it doesn’t know if the driver is actually going to get out-of-the-way of an obstacle or not. So it will only brake at the very last second.
If Autosteer is turned on, the car computer knows what its probable path is and whether it will actually turn in time or not. And so it will be a much more comfortable braking experience as opposed to the last-minute and in that case, we think mostly likely we will be able to brake to a complete stop instead of simply mitigating the impact velocity.
We think it will probably work better with Autopilot on than off. It’s quite and exciting thing.
The exciting thing is that even if the vision system doesn’t recognize what the object because it could be a very strange-looking vehicle, it could be a multi-car pileup, it could be a truck crossing the road, it really could be anything – an alien spaceship, a pile of junk metal that fell off the back a truck. It actually doesn’t matter what the object is it just knows that there’s something dense that it is going to hit – and it should not hit that.
It doesn’t need to know what that thing is – while a vision system really needs to know what the thing is. It is what I think will be a very dramatic improvement in the safety of the vehicle and entirely through software – no additional sensors are needed.
It’s a lot of software and quite a complicated job to fit that software on to the available computer hardware in the vehicle. It was a challenging software problem, but one that we were confident we could solve. And something that quite uniquely Tesla is able to solve because of the fact that we can use fleet learning to have all the Tesla cars out there effectively give us the geolocations of where all the false alarm occurs and what the shape of that object that causes the false alarm. That we know that at a particular position at a particular street or highway, that if you see a radar object of a following shape – don’t worry it’s just a road sign or a bridge or it could be a Christmas decoration that somebody put across the street.
As soon as a few cars pass that point, it will geo-coded into out system via fleet learning system that it is now being generalized to create a list of exceptions. iT applies to radar as well to the width of a road so that you know where the lines should be, you know which road marking to pay attention too, you know when markings may disappear and you may need to go to GPS navigation for a few seconds, you know when to listen to the ultrasonic.
We are essentially generalizing what was a more simpler geocoded kind of blacklist of thing to ignore to a more generalized list of exceptions based on particular geo-coded locations – with now early braking based on radar being one of the applications.
That’s very exciting because it means that all the cars that have been built today that have radar, which is everything for the past 2 years, will receive this upgrade just by over-the-air software. It’s really I think something far beyond what people expect.
You can read part 2 here: Transcript: Elon Musk’s press conference about Tesla Autopilot under v8.0 update [Part 2]