Tesla vs Mobileye: BEST SELF-DRIVING CAR TECHNOLOGY?

There are several companies currently working on autonomous vehicles, but Intel-owned Mobileye is one of the most underrated players in the game and, in my opinion, is Tesla's biggest competition, so let's take a look at the progress mobilized and also to its approach to development. fully autonomous vehicles and compare this to

tesla

's approach and

tesla

's progress, then we will first answer the question of who is likely to complete a true driverless system. I'm jonathan and welcome to cleanwise first in this conversation, I think it's important that you start with the problem, the problem is that it is very, very difficult to develop a fully autonomous vehicle and there is little to no air space.

Humans make many mistakes while

driving

and there are several of these mistakes that turn into accidents each. year, but as a society we accept a certain rate of failure on the part of humans and we accept it as normal, but when it comes to a fully autonomous vehicle and fully autonomous

driving

technology

, people are not as forgiving, there is not as much room for The question really comes down to what is an acceptable failure rate for an autonomous vehicle and how much safer than a human it has to be. Tesla is showing that they can develop a system that is much safer than a human driver, in fact their Autopilot system.

It's not yet a fully autonomous driving system, but technically a level two system comes close to about 10 times the safety of a human driver according to Tesla's Q1 2021 safety report with Autopilot enabled, Tesla vehicles They only had one accident for every 4.19 million miles driven this year. It compares to the NHTSA average of one accident per 484 thousand miles, so if Tesla is able to launch a full

self

-driving system that is 10 times safer than a human driver, that's enough, let's take a look at the number of accidents that would happen each year with different failure rates and different improvements and illustrates why I think it has to improve by a fairly large number according to the US department of energy in 2019, Americans drove about 3, 23 trillion miles with the current nhtsa rate of one accident per 484 thousand miles, that means there were probably around 6.67 million accidents in 2019 if a full

self

-driving system could improve this 10 times, which it would equate to around 667,000 accidents per year if you could ship a system that is 1,000 times safer than a human driver this number could be reduced to 6,67,000 accidents per year so obviously any improvement is good and will save lives , but the question that regulators need to answer is what is an acceptable failure rate, so I don't really have that answer, I don't really know what that answer is, but I thought it's important that we talk about how difficult this is and that it really There is very, very little room for error: what rate will regulators consider acceptable before allowing true driverless system?

Now that we have stated that the problem is difficult, let's talk about

mobileye

's approach to solving this problem and compare it to the from tesla every time you bring up the topic of full self driving and autonomy i think it's important that you have the conversation about lidar of course elon musk has made it very clear that he doesn't think lidar is necessary for his autonomous vehicle and truly Sure, however, Mobileye would disagree and believes that lidar is necessary for a truly secure system, Elon Musk has even said that lidar is nonsense and that anyone who relies on lidar is doomed to failure.

Tesla Senior Director Ai Andre Carpathi even said that referring to lidar gives a false sense of progress and is ultimately a crutch Now, of course, Tesla doesn't use lidar, but they currently use radar, a forward-facing radar in their vehicles; However, according to this March 12 tweet from Elon Musk, with the upcoming significant release of their full self-driving beta, they will be using pure vision and not even using radar. Tesla is really going to push the limits of a camera-based vision system and Elon Musk seems very confident that they will be able to release a system that is safe using only camera vision, so Tesla goes so far as to say that the lidar and radar are not necessary,

mobileye

on the other hand is tackling this problem from both sides.

They believe that it is important to solve the problem of camera-based vision system, so what they are doing is proposing something of a hybrid model, they are separately developing a camera-only subsystem and at the same time they are developing a radar-only subsystem lidar and before they ship their final product, they will combine these two subsystems accordingly, which will allow them to solve the really difficult problem of a camera-based vision system and everything that entails, all the neural networks involved, allows them solve that problem and then add lidar and radar as a layer of redundancy for added security now at first glance this hybrid system and this additional layer of security that mobileye is adding to their system sounds reasonable however I think there is a big fatal flaw that could be a big problem for mobileye in the future with this approach and we'll talk about that in just a minute while moving on to the actual mobilized systems that they are shipping right now, I think it's important to realize that mobileye is one of the big leaders when it comes to advanced driver assistance systems and according to a recent presentation from them in 2020, they shipped 19.3 million Their systems are also impressive: Mobileye has developed what they call monitoring, which is their camera system that doesn't need radars and should actually ship to its first customers later this year, so Mobileye has gone much further than most people and believe that a camera-based vision system is enough for a level 2 8 system, however, they do not believe it is sufficient for a truly driverless vehicle.

They do not believe that a high enough level of safety can be achieved just with camera vase vision because of this for their next generation full autonomous driving product, they are developing their own lidar system called frequency modulated continuous wave lidar As well as developing their own lidar system, they also want to improve the radar systems in their vehicles and move to imaging radar sensors that are much more accurate and give them better data, so although this mobileye hybrid approach will put them ahead of most competitors that rely too much on lidar and radar, I think mobileye has already limited itself to saying that a camera-based vision system is not enough.

I think this will prevent them from making a real breakthrough when it comes to a camera-based vision system, and after all, they won't push the limits of a system like this when you think something is impossible for you. It really is impossible now, of course, there are many examples in science of things that we thought were impossible at one time and that we now know are not, but I think it's important that I step back for a minute and give an example just to show. how limiting. a belief is that something is not possible until 1954 many experts thought that it was impossible for a human to run a mile in less than four minutes, however, in may 1954 an athlete named roger bannister demonstrated it by running less than four minutes in a mile that was possible, the interesting thing is that his world record of less than four minutes did not stand for long and shortly after several people beat that record and nowadays it is common for a high-level athlete to be able to run a mile in less than of four minutes, which was once thought to be impossible, is now commonplace.

I think this will also happen because once Tesla proves that a complete self-driving system can be shipped with just cameras, I will prove to the world that this is possible and then other companies will be able to solve this too and it will prove once and for all that the Lidar and maybe not even radar are necessary for fully autonomous driving, so now I'd like to step back for a minute and talk about why this is important. Why does it even matter if Tesla and Mobileye can offer a system with similar functionality, but Tesla uses cameras and Mobileye uses LIDAR and radar cameras?

What is the difference? Because? It doesn't matter if they work the same in the end, the big reason why this is important is because of the cost of these sensors. It's no secret that lidar is much more expensive than radar, and radar and lidar sensors are much more expensive than cameras alone. Cameras are very cheap, so let's say the year is now 2023, both Mobileye and Tesla have a system that works very similarly, has an excellent safety record and are both approved by regulators, let's say GM is coming and They want to buy a system from these two companies.

They want to license a full self-driving system from one of these two companies when it comes to choosing which system they want to use. Do you think GM will want to implement a system that uses only cameras and is very economical? I think they will want to pay more for their vehicles to install lidar and radar sensors. Of course, a company will choose the cheapest, simplest and easiest to implement option. Yes, they may still be able to have robo-taxis because the economy works differently for that, except for the sheer scale of shipping a consumer product and licensing it to other brands like they do now.

Mobileye's current business model, which will actually fail when competitors come up with a simpler solution, another really important piece. The entire conversation about autonomous vehicles and self-driving revolves around data. The company with the most and

best

information will have a big advantage in the race toward fully autonomous driving. Now Mobilelite has a big advantage over most other competitors besides Tesla when it comes to data, Mobileye currently has six major OEM partners that allow them to get some basic data from the sensors they have in their vehicles and according to them , can capture data from around 8 million kilometers of roads per day.

However, when you dive into this data collection, what they are collecting is a very low bandwidth of data and it has nothing to do with recording events, but it has to do with them mapping the world now. mobileye has made it very clear that they are not building HD maps. HD maps have a lot of detail that is not really necessary for a fully autonomous vehicle. They are building what they call autonomous vehicle maps or AV maps as the vehicles with their sensors move around the world they are capturing. data like drivable space lane line curbs, then they use these maps as another layer for their autonomous vehicles and their adas systems now in the past, like this tweet from october 2020 from elon musk, he talked about how they don't use high resolution maps For their full self-driving program, they mentioned the fact that their system is capable of driving in places they haven't even seen once, while Tesla doesn't use HD maps, as Elon Musk has made clear.

I think they have some kind of basic framework map. that is very similar to mobilized av maps which give the basic space limits for driving and some of the basic features of a road you are about to approach and while you may not need to have this for driving I think They do have it for areas where we have done it before as it increases the accuracy of their system, so when it comes to the amount of data that is collected between the two companies, Mobileye has an advantage over many others companies, but of course Tesla is not able to gain an advantage.

A lot more data as they have a larger fleet and they can get more quality data because they are the ones actually making the car and they can get much higher bandwidth data from their vehicles plus more and better data. The Tesla collection can also run what they call shadow mode on their fleet, which is a background process that doesn't actually control the vehicle, but they can still test certain predictions of their system in real-world environments. I think this is one of the most underrated aspects. of tesla's approach to solving the problem of full self-driving, so now that we've laid all that out, I think it's now important that we answer the question of who will win, who will probably get the full self-driving feature first when the Recent Q4 2020 Investor Call Elon Musk said quote regarding full autonomous driving.

I am very confident that the car will be able to drive itself with better than human reliability this year. Note that he didn't say regulators would approve it or that it would be many thousands of times safer than a human driver. He just said it would be safer than a human driver this year and when you look at some of theexamples of this Tesla full self-driving beta program, like this example from James. Locke, you can see how impressive their current beta is. Of course, Tesla's beta system is not perfect and still requires driver intervention. However, it shows what is necessary with your systems.

It doesn't even use lidar. Plus, Elon Musk seems really impressed. the next version, version 9.0, Elon Musk is so impressed with version 9 of the beta that he believes that sometime in May or even June they will be able to release a large version of this complete, self-contained beta. Now it is, of course. It is important that we talk about the elephant in the room and that is the fact that Elon Musk has promised that they would have full features for fully autonomous driving for several years and of course he has not met those goals, yet with all the progress What are we watching.

Right now, with their full autonomous driving beta program, I think we're really getting close and while it may not be on the exact timeline that Elon Musk believes in, I do think it will get there in the future. In the not too distant future we will learn a lot more about this at a possible AI day that Tesla could hold in July of this year, now in regards to the mobilized projections and when they believe they will have a complete system in a presentation. Recently, they talked about how they want to get to autonomous vehicles at the consumer level sometime in the 2020s.

As far as the robotaxi fleet goes, they hope to have an early writing program sometime this year and have a commercial driverless pilot program in Israel in 2022. They also hope to launch this in other parts of the world in the coming years, also with a recent association they have had. They hope to have 35,000 autonomous delivery vehicles on the roads by 2028 and should begin commercial operations with this partnership. in 2023. In conclusion, mobileye is doing many things right and I see them as a leader in autonomous vehicle

technology

. I think they can achieve complete autonomy and with Intel as the parent company they have resources that no other company has.

I think Tesla will solve what is impossible for many experts and have a complete autonomous driving system that uses only cameras with much greater safety than a human driver in the not too distant future, but not really within five to ten years. It's like that. It doesn't matter who got there first, but it does matter who has a system that is the easiest and least expensive for an automaker to implement in their vehicle, that is the company that will be successful and I believe that company will be successful. be tesla thank you very much for watching this video.

I hope you enjoyed it and learned something too. If you are not already subscribed to this channel, please consider subscribing and if you subscribe, by clicking the bell icon you will be notified when new videos are posted. Also, if you liked the video, consider clicking the Like button because it also helps other people find the video. I also wanted to take a moment to thank the Patreon supporters who support me every month and help me do this. possible content, a special thanks to my performance followers and also to the other followers who appear on the screen.

If you'd like to learn more about the Patreon community I've created, I'll put a link in the description below, thank you very much. a lot