25. October 2019 by Paul Fosse
During Tesla's 3rd quarter conference call yesterday, Elon Musk received a very important question regarding the widespread skepticism that Tesla will be able to roll out "Function full full self-driving" by the end of this year ( only 68 days from now ). This questioner asked him to define what "function complete" means and provide a list of functions that establish a baseline. I will summarize, quote and expand how he answered it.
First, but somewhere else in the conversation, he said that it will probably only happen to Early Access customers by the end of the year, not to regular customers. He described the problem of self-driving as 3 cases of personal use. These 3 user cases must then go through three stages of development and approval. I will give you a combination of his response to these uses with considerable comment from my experience.
3 Use Cases For Full Self Driving
1. High-speed driving. This is highway driving. Everyone who has used Autopilot and Navigation on Autopilot (when you add a destination in GPS) has seen it go from pretty good to excellent in the last couple of years. It has become much smoother and handles the weird corner cases much of the time. Many people can see that this is good enough that it may be safe not to monitor the car very soon. It won't be 100% certain, but Elon has been clear that it would be inexcusable and cost many lives if we didn't allow software to be used until it is perfect or near perfect. He has said many times that once it is 200% safer than a human (I assume that an average human being, not the world's best driver), he will push hard for it to be approved by the regulators in different regions.
2. Low speed. This means very low speed, like 5 miles per hour. This is what Smart Summon and the upcoming Smart Parking features are all about. The car is still monitored by the driver, but now from the outside of the car instead of inside the car. Elon claims that Smart Summon will be much better in a few weeks, and learns a lot from the cornerback experience with 1 million uses since it was released less than a month ago. Most will say that this is not as mature as the high-speed driving software, but they will also admit that if there was a mistake, it would probably just scratch a curb or buck a fender. An accident at this rate is unlikely to cause injury or death. [ Editor's Note: My overall thought about Smart Summon is that the biggest weakness is that it is overly cautious, but it still has to be. If I try to use it in an area that is a little busy, it often gets too much because of people passing by or curbs or other cars. People don't want to sit in a parking lot for a few minutes while a self-propelled Tesla skates forward. I hope Elon's repeated comments that doing it much more smoothly will solve most of this problem, but I wonder how it does it while continuing to be hyper cautious not to appear on the front page of the New York Times. We'll see. -Zach ]
3. Medium speed. This is driving around the city, stopping for traffic lights, stopping and looking for all the urban chaos, like pedestrians and bicycles from unexpected places. We've seen a few controlled videos that show this works, but no one outside of Tesla has been allowed to test this yet. This is expected to be released to testers with early access over the next 68 days! I guess this stage was held for two reasons.
- I think it's the toughest use case, and they didn't want to deal with it until the other two cases worked quite well. In this way, they can borrow software from the two cases to do much of this work. They need new software to recognize lights and signs (which have been running in the background for a long time). They also need logic to decide when to slow down and accelerate and take evasive maneuvers in an urban environment.
- Some of this enhanced software and image recognition requires the new Full Self Driving computer that Tesla began to insert new cars this spring. Tesla also began upgrading existing cars (for people who paid between $ 2,000 and $ 6,000, soon to $ 7,000, for FSD hardware and software upgrades). Since Tesla didn't have this in many cars until recently, it didn't make much sense to release the features. I don't think this reason is as important, but it may have been considered. If the software was ready, Tesla could have sent it to people who had early access 6 months ago, so I think the software just wasn't ready.
So these are the 3 use cases. But then each of them has to go through three levels of development.
Development Gates For Full Self Driving
I develop software every day, so when Elon described the three steps / steps / gates on the timeline, it reminded me of my day job.
1. The main functions are there. I'd say the car copes with low-speed and high-speed driving today – they just need to release the mid-speed features (or city mode Navigate on Autopilot), and this stage is complete. Elon was clear that the software does not have to work in every way and that it does not have to handle every corner house. The driver is always responsible. Elon expects this to start rolling out by December 31, 2019. It will be nearby.
2. No driver monitoring required. All the features are safe enough that Tesla feels that the driver does not need to monitor the car, but at the same time, enough data has not been collected to convince the regulators. The driver is always responsible as a result. Elon expects this in mid 2020. I expect there will be something later.
3. Regulatory authorities approve. The software no longer requires the driver to monitor the car in the area approved by the regulator. Accident liability shifts from driver to car. This is a watershed moment in history. Elon expects this in 2020. I expect there will be something later. [ Editor's Note: I think the most interesting thing here is that a specific region that wants to be first could – maybe – really be ready to jump in and approve the FSD / robotaxis basically as soon as Tesla think it's clear. I am fascinated to see which regions are first and how long it will take to approve it. I agree with Paul that it's a watershed moment and 2020 seems premature, but it's a watershed moment even in a small geographic area, because then real robotics data is starting to be collected and other bullish regions just being a little more cautious can see what happens in region # 1 in a relatively short time and then jump in. Could it be Dubai? Florida? Any areas in China? We'll see. —Zach ]
Q3 Safety Update
In addition, Tesla gave us a taste of the security numbers in the third quarter. I expect Tesla to publish a full update soon.
From Tesla's Third Quarter Shareholder Letter 2019 (Page 8):
“During the third quarter, we recorded an accident for every 4.34 million miles driven by drivers had Autopilot engaged. This is comparable to the national average for an accident for every 0.5 million miles based on NHTSA's latest US data. "
To put it in the context of the other four quarterly Tesla Safety Releases, miles per accident have ranged from 2.87 million miles in Q1 2019 (possibly higher due to winter weather in much of the United States) to 3, 34 million kilometers a year ago in Q3 2018. This is a 23% reduction in accidents per mile driven with Autopilot engaged for one year!  This means that if you could drive all the average miles someone in the US drives (13,476 miles) all on Autopilot (you can't), you would have a chance of 1 in 322 to have an accident in a year , or over a full mileage of 60 years (ages 16 to 76), you would have a 1 to 5 chance of being in an accident (which may not necessarily be serious).
Compared to accident statistics for all cars in the US, where there is an accident every 500,000 miles (which is not an apple-to-apples compared ison because it includes city driving, which you may not always use Autopilot for, but it is where the accidents are more likely). In this case, people have an accident every 37 years and will average around 2 accidents during their lifetime.
It makes driving a Tesla on Autopilot almost 9 times safer than driving an average car (with the proviso that it is not is not a fair comparison). A year ago, driving on Autopilot was almost 7 times safer than driving an average car (with the same reservation). So Tesla is making slow but significant progress.
Six months after Elon's interview with Lex Fridman, in which he radiated confidence that Tesla would win the race to make self-driving cars, Elon is still confident he can do it and he has delivered the second part of the puzzle (low speed driving). On the other hand, he has had to admit that the full Full Self Driving update with the feature is likely to be available only to customers with early access and not the entire fleet this year.
My opinion is unchanged from 6 months ago. I think Tesla can do it and I think the company will do it before anyone else, but I think it will take longer than Elon has predicted. Maybe 6 months, more likely 1 to 2 years longer, for him to get it approved by regulators. So I speculate that I believe a larger region (like a US state) will approve full self-driving by the end of 2022 or 2023.
Use the Tesla Referral Link to get 1,000 miles of free supercharging on a Tesla Model S, Model X, or Model 3 (you cannot use it on Model Y yet). Now is good for $ 100 off solar, too! Here is the link: https://ts.la/paul92237
Follow CleanTechnica on Google News.
It will make you happy and help you live in peace for the rest of your life.