قالب وردپرس درنا توس
Home / Business / Can Tesla stay one step ahead of the industry and deliver full self-driving cars next year?

Can Tesla stay one step ahead of the industry and deliver full self-driving cars next year?



23rd April 2019 by Paul Fosse


In this article I will examine the allegations that Tesla made on his April 22 Tesla Autonomy Day. Although I'm a small investor In Tesla and undoubtedly a fan of the company and its cars, I will try to be as objective as possible and to show where Tesla's allegations are or are undisputed, and where they are unproven and require a vote of confidence. In order to have a Robotaxi, Tesla must have the following puzzle pieces:

1. Cars. Tesla claims to have built about 500,000 vehicles that have the required sensor array (each Tesla built since October 2016 has 8 cameras, 12 ultrasonic sensors and 1 radar) and these cars can use the upgraded FSD computer (Full Self Driving) received. Tesla plans to produce another 500,000 cars next year. There should be a million cars available to compete with Uber and Lyft at the touch of a button.

Of course, Uber and Lyft have registered more than a million cars on their platforms (estimated at over 2 million only at Uber). Not all these millions of Tesla cars are willing to let a stranger ride for money. On the other hand, Uber and Lyft riders only drive a few hours a day, while the Teslas can run for up to 24 hours as they do not need a driver.

It seems clear that this Tesla does not have the balance to inflict significant damage on Uber and / or Lyft when it offers services for the first time. On the other hand, investors are trying to anticipate the future with their investments, and if they believe the story of Tesla to be credible will do much harm to Uber and Lyft – if Tesla can scale in the next five years without a driver to the driver It is obvious that the company will have much lower costs than Uber and Lyft if there is no access to millions of self-driving cars.

. 2 Redundancy. Tesla needs cars that can accelerate, brake and steer with electric motors. Elon and the team mentioned that they are fully redundant when braking and steering (acceleration was not mentioned), so a power steering motor and a motor with engine brakes can fail and still steer and stop safely. I would think you would want to stop and fix the problem, not with the single steering and brake motor. Although this is possible with gas cars, most people claim that controlling a car is a little easier with electric cars. This point is not really denied by the critics of the company.

. 3 Electric cars against gas or diesel cars. You can certainly build a self-propelled gas or diesel car (if you can figure that out naturally, of course), but it is undisputed that electric cars have much lower fuel costs (about a quarter of the cost in most areas). If you only drive a few miles a day, this tends to affect the higher purchase price of the electric car. If you drive a lot, as in 24 hours a day, to maximize your income, the lower cost of an electric car will be very high. Tesla is the only manufacturer of electric cars in the US that has a significant size. It seems that automakers are electrifying throughout the industry, but it is widely disputed how fast this will happen and even if so.

Elon claimed that her Model 3 engine and bodywork can travel a million miles and that their batteries can go from 300,000 to 500,000, but that's not proven. At the presentation, Elon claimed that next year, a new battery pack would come on the market, which was designed for more charging cycles and would last a million kilometers. This is not proven, but Elons record with these claims is excellent. He has always delivered the promised battery power, but not always in the promised timeframe. It is believed that maintenance costs for Tesla vehicles are much lower than for gas cars, and although this benefit is sometimes controversial, the evidence for this is quite strong.

. 4 Sensor array. Does Tesla have the right sensors?

Nobody disputes that cameras, ultrasonic sensors and radar are very useful, but almost everyone thinks Lidar is necessary. I wrote about it here. CleanTechnica has also addressed it here and here. The problem is that while Lidar makes it easier to find the safe areas to drive in, it provides a 3D map of the room with no artificial intelligence (only one laser illuminates and measures the time it takes to jump back) can) not work in bad weather and also helps with many other problems you need to solve to drive yourself. The lasers do not help with stop signs, traffic lights or the detection of bicycles or pedestrians or cars or the prediction of the future behavior of these three devices. Lidar does not help to read road markings or signs or other aids used worldwide to support billions of human drivers.

Lidar is great if you want to build a car into a science project and run around on the street, running the road and not encounter any fixed objects in perfect weather. Then you do not need fancy software, you can only tell the car where the stationary objects are and find a way around them.

As you can see clearly (pun intended), Lidar does not help with all of the issues involved in complex urban environments – namely, moving people, bicycles, animals, cars, and trucks controlled by humans or animals who do unpredictable things in any weather. For that you need some intelligence, either human or artificial.

This is Anthony Levandowski speaking above.

5. Intelligence to understand the environment of the car. To help you understand the road in the future, Tesla demands that you need modest processor and graphics processing power as well as enormous multiplication and additional power for linear algebra. As I wrote almost a year ago, Tesla has explored what the industry has to offer to meet the demands of the computer, and no one has been found working on a chip that meets the performance requirements (in particular, the ability to perform edit single image simultaneously) instead of stacking 256 images – at very low power). Using too much power will significantly affect the range of the vehicle.

Elon hired a top team of digital equipment, Intel, Apple and AMD to create a custom chip. With little CPU and graphics requirements, they licensed existing designs and simply put them on the chip. However, because they had unique requirements for high-performance, very low-power multiplication and addition operations, and could not find acceptable solutions to a license, they developed a very simple processor with very high performance. In the semiconductor industry, it is well known that you can make a chip faster for an operation if you do not have to work with a complex instruction set.

This is also displayed during cryptocurrency mining. When you're ready to develop a chip for mining, you can do things much faster and with less power than using CPUs or GPUs for math. The reason why no currency has its own chip is designed to cost each chip a fair amount, and it is difficult to predict which crypto currencies will be used enough to repay the initial chip design cost. Of course, this is not a problem – if this chip solves the complete problem of self-driving, there is no great demand.

I've heard critics claim it's unlikely that Tesla could design a chip that is better than the "experts" of Intel and AMD, but I find their plan workable for several reasons:

  • They have first-rate Talents from the industry set.
  • They only tailored the parts of the chip they had unique requirements for. They licensed proven (but not leading) designs for the CPU and graphics engines. This project would have been much riskier if the entire chip had been individually designed.
  • You manufacture the chip in a Samsung factory. Elon may love vertical integration, but he's smart enough to realize that building a 14-nanometer lithography process that switches to a 10-nanometer process is a headache that they did not have to worry about.

. 6 You will need a lot of training data.

There is no doubt that Tesla drives many more cars with cameras than any other player in the world combined. It is debatable whether they can afford the mobile phone charges to send the data back to the mothership. If this is not the case (and they are likely to be able to return only a small portion of the data), then they select the right sample to get the edge cases they need for the safety of the cars. They use the driver's retreat to decide what routine video they do not need, and what's special about the image recognition software they're dealing with.

. 7 Image recognition and depth perception.

Elon made it clear in the presentation that Andrej Karpathy was not only a doctoral student and professor of artificial intelligence education at Stanford, but developed the very popular one there he taught image recognition and is arguably the best expert the world when training neural networks for image recognition. I think few would deny that Andrej is a top expert, but many (myself too) are not convinced that image recognition can progress as fast as Elon claims. I've read a lot of articles about it, and that all sounds plausible, but it's such a big leap in performance that I doubt whether it's going to make so much progress in such a short time.

I will say that one example where my skepticism was misguided was Alexa's natural language ability. I had seen PC products for 30 years that were said to perform speech recognition, and they all needed a lot of training to get disappointing results. Suddenly Alexa (and I've heard that Google also has a good one) solved the problem and it seems to understand what I'm saying pretty well. It still seems pretty stupid to do complex tasks but does a good job doing simple tasks.

Tesla has a great team, but this problem is incredibly difficult. This is really the area Tesla just has to prove it works because the world will not trust them no matter what they say.

. 8 When you drive the car, you can see which objects are there and where they are going. This is not too difficult with the exception of the chicken game that the drivers play when changing lanes. Tesla has to prove that they can find a way to be assertive enough to get into a busy lane without causing a minor accident. This is difficult for humans and also for computers.

Conclusion

I was impressed by the Tesla Autonomy Day, impressed by Tesla's strategy and enthusiasm, but unconvinced that they will make it the next moment of the year. In my 35-year career in software development, I have seen many examples of a project that I expect to take four years to complete with excellent leadership and programming skills in a year's time. I have also seen that several projects that could have been completed in one year were discontinued after several incomplete years. This is usually due to a leadership role that had a great vision but did not have the necessary talent. Too complex development processes have also brought some projects to a standstill, but I do not think this will be a problem for Tesla. Elon has been developing commercial software since he was 12 years old – he will not allow this project to be destroyed by a bad process.

My opinion is that they can do it, but I really do not know if they will not make it next year or later. As Yogi Berra said, "Forecasts are difficult, especially in relation to the future."

If you would like to use my Tesla referral link to get on a Tesla Model S, Model X, or 1,000 miles free charge Model 3, here is the link: https://ts.la/paul92237 ( if someone else helped you, please use his code instead of mine). I recommend buying before the Full Self Driving (FSD) price goes up on May 1st, if you believe in Tesla's ability to get it up and running soon.


Tags: Andrej Karpathy, Elon Musk, Pete Bannon, Tesla, Tesla Autonomy Day, Tesla Autopilot, Tesla Full Auto Driver


About the Author

Paul Fosse A software engineer for over 30 years years. First the EDI software was developed, then data warehouse systems were developed. Incidentally, I also had the opportunity to start a software consulting firm and operate portfolio management. In 2010, I was interested in electric cars because gas became expensive. In 2015, I started reading CleanTechnica and was interested in Solar, mainly because it threatened my oil and gas investments. Follow me on Twitter @ atj721 Tesla Investor. Tesla reference code: https://ts.la/paul92237[19659042((((sid){varjsfjs=dgetElementsByTagName(s)[0]; if (d.getElementById (id)) return; js = d.createElement (s); js.id = id; js.src = "http://connect.facebook.net/en_US/all.js#xfbml=1"; fjs.parentNode.insertBefore (js, fjs);} (Document, & # 39; script & # 39 ;, & # 39; facebook-jssdk & # 39;)))
Source link