When it comes to new energy, who comes to mind first? Is it a solar cell?" What is a solar cell?", which is an interesting question. What exactly is a solar cell? What are the types of solar cells? How are they related? Tesla, BYD, Wei Xiaoli?
I believe that most people will think of Tesla or BYD, these two companies are giants of the domestic and even overseas new energy market (today we will talk about cars, not Wuling Hongguang MINI!). Since entering the domestic market, Tesla's sales in China have been very strong, and rising stars have regarded Tesla as a competitor, hoping to one day pull Tesla down the ladder and compete for the status of a new energy brother. At this time, do we have a reason to pay attention to Tesla? If Tesla really becomes the leader in China's new energy vehicle field, what impact will it have on the domestic automotive industry? How should we view this phenomenon? However, just as everyone was discussing which car brand Tesla's biggest competitor was, a piece of news surprised everyone.
There has been a saying for years: "It's not necessarily your peers who kill you, it's probably your crossover!" "This is actually a misreading of cross-borders." Because for a long time in the past, people often regarded cross-border as a derogatory term. In fact, cross-border itself is a positive way of thinking and behavior. This is also very suitable for Tesla!
In March this year, at the GTC Spring Conference held by NVIDIA, NVIDIA released its own Olympian Intelligent Driving System Chip, taking a big step in the field of intelligent driving. Huang Jenxun, founder of NVIDIA, said: "Today we are pleased to announce that BYD, as the world's second largest electric vehicle manufacturer, will start producing cars equipped with the DRIVEOrin computing platform in the first half of 2023. This is also the news that BYD officially announced its cooperation with Nvidia. In addition to BYD, many domestic new energy brands have also chosen NVIDIA's automatic driving solutions, such as Weilai, Xiaopeng, Ideal, Weima, Gaohe Automobile, etc. It can be said that it occupies half of the domestic new energy! So, how powerful is this new battery? What kind of surprises can it bring us? Let's take a look. First, let's take a look at what an electric vehicle is. Electric vehicles are also called electric vehicles. Today Xiaobian will talk to you about Tesla and Nvidia Autopilot, who will laugh to the end!
Since the successful completion and operation of Tesla's Shanghai Gigafactory, "Big Catfish" has disrupted the domestic new energy market and injected vitality into the original "backwater" energy market. On the first anniversary of Tesla's listing, its stock price hit a record high. For a time, Tesla became the object of everyone's hot discussion. In this era of so-called "black technology", what kind of charm does Tesla have? But it's also always in the public eye, with frequent accidents and growing sales in stark contrast, with many people asking why Tesla, which is as rough as a furry house inside, sells so well. This also starts with Tesla's intelligent driving system!
Tesla FSD
Some people say that the first half of the automotive revolution is electrification, and the second half is intelligent. Obviously, the current development of new energy is moving from electrification and intelligence, which also means that more new technologies and products will appear in the near future to replace traditional energy. So, what are the possibilities of electric vehicles in the future? The answer is undoubtedly autonomous driving. Can driverless cars replace human driving? Intelligent driving solutions gathered, Tesla's automatic driving is undoubtedly the leader in the development of intelligent electric vehicles, many people do not agree, but the fact is that Tesla's FSD is currently the world's only fully automated driving system system, of course, only the United States Tesla is used, the rest are assisted driving.
Pure vision, intelligence, Tesla's self-driving solution is different from other brands. Tesla's self-driving approach is also seen by many as an irresponsible change to save costs. In this regard, Musk is also vague, no, he explains, that the pure vision scheme is to imitate the principles of the human visual system, constantly capturing images of the real world seen through 8 cameras, processing them through a complex perceptual neural network architecture in a powerful in-vehicle system to create a three-dimensional vector space of the real world. Because it's a huge project, like the world we see ourselves, we can clearly identify traffic lights, pedestrians, traffic, zebra crossings, traffic signs, and so on. In the real world, the roads in Germany are very complicated, and we humans are very good at driving on complex roads through continuous Tesla learning and driving accumulation. So why is Tesla doing this? The answer is simple, because these cameras are all learned! Why is that? What do they do? Can they really see things like people? Are they also learning?
AI learning; from above we can see a picture book that looks like a schoolboy, as well as the drawing of a three-dimensional object. In the picture on the left, Tesla Autopilot appeared before 2021, while Tesla re-planned its autopilot program 21 years later. Until 2021, Tesla will identify objects frame by frame by tagging video objects frame by frame and uploading them to a database. This method not only takes into account low values, but also causes the labeling of some scene objects to be missing. Today, thanks to powerful algorithms and chips, from 2D image recognition to 4D, every Tesla on the road is constantly learning, learning the driving habits of drivers, learning the behavior of car owners when recognizing objects, and more than 1 million Teslas run around the world every day, constantly learning. By early 2020, Tesla had developed a new technology, deep convolutional neural networks, that automatically identify all features in an image and extract the semantic information contained therein to accomplish complex tasks. This process is done entirely by computer automation. In short, from manual import to intelligent self-learning! Therefore, tesla's autonomous driving is infinitely closely related to human driving habits.
Super-powered computers – Dojo; Tesla Autopilot will achieve high growth throughout 2021. Also thanks to the Penguin Dojo incident a few days ago, "Dafaku Fast Three Data Processing, we finally know the answer: Tesla Motors collects more than 1 million videos around the world every day, and spends nearly 10 seconds or even more Dafa Fast 60 regular time on these videos every day to label, which is undoubtedly a very tricky thing for Tesla! Just last year's Tesla AI Day, Dojo's chipset architecture diagram debuted. At present, Do Jo's work is mainly divided into two parts: the first part is feature extraction of pictures, and the second part is the use of machine vision technology to achieve the recognition and detection of biometric information such as faces and irises. The first part is face recognition. It was specially developed to be able to monitor large amounts of video data for labeling and training. Without Dojo Tesla would just stop at 2D image analysis because it doesn't have enough brain support, and Dojo is equivalent to a super brain that is constantly learning, constantly recording and learning what the naked eye sees.
When many people buy Tesla, they will think of Tesla's leading intelligent driving technology and enjoy the changes brought by technology, not the Tesla brand! In the United States, one company has launched a concept model called "autonomous driving.". Not only does it have all the features of a traditional car, but it can also achieve driverlessness through voice control. Fully autonomous Tesla is operating across the U.S., and it has tested the reliability of FSDs time and time again. Full automatic driving is not supported abroad, but it is also in a leading position in the field of domestic assisted driving, so when everyone is thinking who will be Tesla's mortal enemy, the emergence of NVIDIA has broken the calm on the surface of new energy vehicles.
Nvidia appeared
NVIDIA believes that everyone is familiar with autonomous driving technology, and I will not go into details today.
After watching last month's GTC2022 conference, many people lamented that this should be the most powerful autonomous driving, look at 2026! However, Nvidia's future self-driving version, hyperion9, is still very strong. From the appearance point of view, "Hyperion" is no different from ordinary mobile phones, except that one screen is a little larger and two cameras are smaller. But it has more room to innovate. It's like a homemade domestic mobile phone, stacked very high.
Equipped with ultra-luxury hyperion9 platform, there are 14 cameras, 9 mmwave radar, 3 lidar, 20 ultrasonic radar, 3 cameras and 1 mmwave radar in the car. Tesla was stunned. But it won't be until 2026 to launch a first-class model with Nvidia! Is this also the case with Tesla's mass production version in China? Can Tesla really do that?" The word "Tesla" sounds lofty. In fact, this is a false proposition. It's just a gimmick. Luxury configurations and smart platforms like Tesla certainly require super brains to handle.
At the press conference, NVIDIA also released the Atlan chip, which has a target computing power of 14 times that of Tesla FSD; But Tesla FSD is also about to become a thing of the past, more than 3 times that of NVIDIA's current mainstream chip Orin. But to say the least, it won't be online until 2026! That being said, don't you think there will be plans to use NVIDIA New Energy in China this year and next? "Yes!" That's what we're going to talk about today. New energy vehicle technology is to improve the performance of the on-board battery to achieve the purpose of extending the mileage and reducing fuel consumption, thereby improving the overall fuel economy and emissions of the vehicle. It was announced at the meeting that in March this year, Orin chips have been mass-produced, such as 4 chips on the ET7 of Weilai, making the hash rate of ET7 reach 1,000 tons of 1000TOPS. BYD will also start production of electric vehicles equipped with the Olympian Orin chip DRIVEHyperion system in the first half of 2023. It's not as powerful as Atlan four years later, but right now it's the only one that can match Tesla's, with 254 trillion computing power per second, as well as the ability to distinguish between peripheral interfaces and high bandwidth. This is undoubtedly good news for China's local independent brand car companies! In addition, there are also a number of new energy companies in China to develop self-developed model definition software on NVIDIA's cooperation platform!
In addition to car brands, even Baidu has begun to cooperate with NVIDIA. As early as November 9, Baidu announced that its third-generation autonomous driving platform, AUC's Sanshen, is equipped with Orin chips and uses Nvidia. Realize the "three fresh" automatic driving computing platform to cover the global automatic driving scenario. This means that in the future, driverless cars will increasingly adopt this technology. At present, the AUC system has entered the mass production phase and is planned to achieve global deployment in 2018. furthermore. Using the Orin chip Baidu autonomous driving will improve performance and safety.
Nvidia's ambitions
In the context of speculation about whether Tesla's visual collection data will endanger the security of national road information, Nvidia said it will establish Drivemap, a high-precision map platform spanning 500,000 miles of roads in Europe and the United States, in 2024 to improve the reliability and safety of autonomous driving. This result was developed by the famous British car manufacturer Infineon Group (NXP) and is currently being tested, and it is expected to be in mass production and market within the next 5 years. The technology is provided by platform developer DeepMap, which has been working on high-definition maps for the past few years and was acquired by NVIDIA last year. In the next step, the company will combine the original DeepMap digital camera crowdsourced image technology with 3D lidar data collected by vehicles equipped with NVIDIA's self-driving platform to draw a resolution map with higher accuracy and artificial intelligence crowdsourced AI crowdsourced map resolution map autonomous driving map!
Having said all that, do you think Tesla and NVIDIA are more dangerous?
IOS and Android in the world of autonomous driving
In fact, from this perspective, Tesla's FSD and Nvidia's Hyperion are very similar to Apple's IOS and Android, which are developing their own systems around Tesla. On the other hand, NVIDIA provides systems for other car brands. NVIDIA's move has largely encouraged those who might have chosen Tesla smart driving to consider buying domestic new energy vehicles with the same price, technology and interior. If these factors are not taken into account, it is likely to be eliminated by the market. This is a loss for Tesla, but it is a rare opportunity for Chinese automakers. Just like Android, Android's brand can also be developed through its own tonality, so I believe that our self-developed new energy brand will create an autonomous driving that is more suitable for domestic roads and usage habits on the basis of NVIDIA. This is a big blow to Tesla's sales, so Tesla's biggest opponent should be NVIDIA!