Autonomous Driving Spring: Challenges and Innovations in Vehicle Software

The era of autonomous driving is beginning, with Waymo leading the way, as Tesla aims to transition to an AI company with robot taxis.

### The Future of Autonomous Driving: Tesla, Waymo, and the Changing Landscape of the Automotive Industry
* 🚗 Elon Musk compares non-FSD cars to "**feature phones**" and FSD cars to "**smartphones**", emphasizing the importance of autonomous driving technology.
* 🤖 Waymo is leading the way in autonomous driving, with **eight years of testing in Phoenix** and now launching **commercial robo-taxi services in San Francisco**.
* 📊 Tesla's FSD approach differs from Waymo's, focusing on **in-house development** and using only **cameras for vision**, without **LiDAR**.
* 🏭 Volkswagen's struggles highlight the challenges of developing vehicle software, leading them to partner with **Rivian for software development**.
* 🌐 China is preparing for the era where electric car competitiveness will be based on **software and autonomous driving capabilities**, not just vehicle design.
* 🧠 The US is moving towards an era where **autonomous driving without LiDAR** is becoming more accepted, with **Tesla leading the way**.
* 🛣 The future of autonomous driving includes challenges such as **regulatory approval**, **societal acceptance**, and **data prediction for efficient service**.
* 🌱 The automotive industry is entering a new era, with **software and autonomous driving capabilities** becoming increasingly important for vehicle competitiveness.

Elon Musk said that the car that does not have Full Self-Driving (FSD) is like a feature phone, and the car with FSD is like a smartphone. Now, even in China, where only the outer shell of electric cars was competitive, they are preparing for the era where competitiveness ends. Due to the Volkswagen and Rivian investment incident, chaos broke out in the German automotive industry. How difficult is it to develop shooting software? Volkswagen, a large company, had to swallow its pride and invest with Rivian, right? How much did they invest? Volkswagen, who has not invested in software until now, only has six developers. However, after trying for about 3 to 4 years, they found out that software was not coming out internally. Clearly, there has been a winter of autonomous driving in the last 4 to 5 years, but the long winter is ending, and spring is beginning. Now, we need to take a closer look.

The fact is that now there is something called Sura. This is the self-driving autonomous driving accident that Amazon acquired in 2020. Here, it's called Waymo as well, and they are currently testing in several cities. They haven't commercialized it yet, but it can be said that they have evolved a lot here as well. Another thing Waymo is doing is collaborating with Uber, which is quite important. If Uber is like ours, it's like Coupang Eats or Baedal Minjok, and they are currently testing. So, when people order food delivery, robot taxis deliver it. Of course, there are people inside, but there are also simple autonomous things. Another interesting thing is that in June of this year, they received a large investment of $200 million from Yara. Yara is a startup in the United States. What Yara does is trucking, so they transport goods.

When it comes to autonomous driving trucks, it may be more feasible to commercialize freight trucks because they don't go anywhere other than the predetermined routine. For example, autonomous taxis don't know where to go, they can go home, go to a promised place, or meet someone, but freight trucks, delivery trucks mainly, for instance, delivering goods only to this mart, only to Lotte Mart. If the route is fixed, after learning it hundreds, thousands, tens of thousands, or even millions of times, the likelihood of accidents here is lower because all the variables are known. Especially in the case of the United States, when you look at highways, Amazon trucks, and other large vehicles keep coming and going from city to city on the same route. Since they always travel the same route, they can travel more safely.

It will be like that, right? The rain is starting to come down a bit, so to speak, and here in the United States, there are companies that now make tractors. In cases like this, the farmland in the United States is so huge that it's unimaginable. Since there is a lower chance of accidents when driving alone in places like this, this is also a factor. Nowadays, many startups are being established, and military vehicles can also be good. Because rather than needing special equipment like forklifts or robots to handle the necessary supplies for the military, it is more efficient to just load the goods onto American military vehicles and transport them to their destination. There could be enemy attacks on the way, right? It is possible to transport military supplies without casualties. In this aspect, the United States is considered to be somewhat ahead because autonomous driving technology is being incorporated a lot. And now, cruises.

In the same case, as you all know, it was a GM disaster. From what I see, it was a self-driving car, so humans were driving remotely from behind. The CEO stepped down, resigned, and many people were restructured. Now, GM has decided to invest again and start testing again. So, we can say that it has reached a certain stage. So, to summarize, currently, Waymo is leading the way. In fact, eight years ago, when they tested in Phoenix, initially, there was a backup driver, right? Yes, yes, yes. With the backup driver, they put passengers in, then removed the backup driver, and remote drivers took over. Then, they eventually went fully autonomous. Now, they have reached a point where they go to San Francisco, where initially there were backup or remote drivers, and then they tested fully autonomously, involving some people.

We have been testing robo-taxis with some passengers, but starting this month, it will truly become a commercial service. It's the first time. It took eight years step by step to get here, and now we are starting. While we might be able to shorten the time, since there is a lot of data, it will still take time, right? Yes. So, Tesla's approach and Waymo's approach, as I mentioned, are different. Tesla only uses cameras, but there are two markets for Tesla. One is making their own vehicles smart, so Elon Musk said, If a car can't do full self-driving, it's like a feature phone, but if it can, it's like a smartphone. Who would use a feature phone when smartphones are available? That's why FSD.

I will add to the appeal of basic Tesla vehicles. This part is now important, and with FSD, revenue is generated. This revenue needs to be reflected in the financial statements. This is AI revenue. That's how we transition to an AI company and eventually start robot taxis. I think it will take quite some time for robot taxis to be widely accepted. There is societal resistance to robot taxis, as seen in the Waymo incident, which is not good for Tesla. Even if it's a Waymo incident, it's not good for Tesla either. It deteriorates the societal perception of Robo-taxis overall. That's why progress can be very slow, and then there's the second point. As I mentioned earlier, there is a societal system that needs to change. Taxi drivers need to disappear, which is why taxi driver unions oppose it. Therefore, it's not easy to get in. If there are political elections and so on, and thirdly, there are data prediction apps like Woebot.

The reason why we do it with Uber is because the customer data on Uber is important. Even taxi drivers, if they work the same hours, there are people who earn more money. It's because they know the routes well, where customers get on and off frequently, and where they can pick up more customers. There are people who have know-how, right? This is now intuitive intelligence. But we need to prove this with data to know where to dispatch the cars. For example, Tesla started Robo-taxi services. If you call from the app and it says, Arriving in 15 minutes, you wouldn't want to ride it. You would just call a regular taxi, right? Whether it's a reporter who came to cover a story or someone commuting, if they say it's 15 minutes here and 30 minutes there, then you really have to be able to get to a place within 3 to 4 minutes, maximum 5 minutes.

In advance, but this is something that may need to know customers' mobility data, so I think Tesla could also cooperate with Uber. Therefore, it's not completely ruled out. In other words, it's possible that among Uber's vehicles, people can choose Waymo, and they can also choose Tesla's Robo-taxis. In this way, there is a possibility of creating a business model. However, since Waymo lacks scalability, it needs to expand the market as a robot taxi itself. Tesla can also expand its Robo-taxi, but first, the goal is to increase the adoption rate of FSD. The next bigger goal is to gradually purchase licensing for FSD from other companies. To go that far, as I mentioned earlier, it involves technological stages, political and social stages, and then data, which means transportation data, customer data.

Because it is necessary, it may take quite a long time until the robot taxis start generating revenue in the market. However, now, the U.S. has finally come out of a somewhat long winter, and it is clear that spring has begun. We used to refer to the 1980s and 90s as the AI winter, but in recent four to five years, there has definitely been a winter of autonomous driving. The spring of autonomous driving is starting from May this year, and by July, finally, in San Francisco, Waymo has been recognized by authorities for the level of safety as a robot taxi that anyone can ride. In this regard, Waymo is leading the way now, and following that, robot taxis will come. With this, it seems that we are gradually entering the era of autonomous driving services at a faster pace. This is a boom in what kind of technology.

It's often said that you can't do it alone, you have to compete together to develop. If Waymo starts commercial service alone, it might be a small news, but with Tesla continuing to advance like this, both will have a positive effect on each other. As you mentioned so far, in the case of Waymo, they need to modify the cars, install lidars, install radars, and there are many things to add to existing cars. So, it seems difficult for other companies to apply it. When you talked about Tesla before, there were comments like, is it easy for other companies to bring FSD and apply it? This also seems challenging, right? For that to happen, it needs to be installed. There are currently 3 hardware cameras, and from this year, hardware 4 is being installed, and then.

Starting next year, Hardware 4 will become Hardware 5. This time, they changed the name to AI 5. Just by saying AI 5, of course, everyone thinks Windows will run. So, like what was installed in the past in 3 and was called Intel Inside, it will now be Tesla Inside. For that to happen, eight cameras and two high-end chips like Hardware 3 will be included. This is like the M1 chip, saying it has the performance of 50 chips. The Apple M1 chip, these are high-end chips. Two of these boards will be placed in Hardware 3 and Hardware 4, and from next year, AI 5 will be released. All of these components need to be installed, and because of this, prices will increase. The vehicle prices will go up, with eight cameras. All images captured by these eight cameras in all vehicles will be sent to Tesla's data center.

It's going there, so we are teaching and that's why it's not just about having a camera. It must have hardware 3 and hardware 4, commonly referred to as on-device AI here. With our own cameras, even when the internet is not connected, autonomous driving for FSD to operate requires these hardware devices. Therefore, these need to be sold together. So, it's clear that some additional devices are needed, but it's not too difficult to apply like the way it was before. If you make some changes, it can be applied to existing electric vehicles to some extent. Therefore, it is possible to license it. They are talking about this now. Why can't other cars do it? They talk about this. Tesla has been making significant investments in this since this year. Also, they have been testing this without LiDAR, and the FSD version was initially quite gradual.

It rose, but it is currently rising exponentially. Another thing is, Tesla designed the vehicle itself with the ultimate goal of autonomous driving from the beginning, making these things possible. When it comes to making software for cars, what's important is not only controlling the car but also very important. Controlling a car is not just about having a steering wheel, handle, brake, and accelerator pedals. However, what Tesla ultimately wants is, in the current car operating system, there is a controller in each part, and the brake, instead of being one component, is a combination of several components, forming a brake system component. In this component cluster, there is a controller. If this controller is present in an iPhone, for example, when updating the iPhone or Galaxy operating system, they all get updated together. So suddenly,

To improve certain functions, the middleware needs to be integrated into one stack so that it can be fully controlled. In order to do that, the controller should be produced 100% by the manufacturer. Currently, when it comes to Cybertruck, Tesla manufactures about 85% of the controllers. By designing them, Tesla has gained control over 85% and will have 100% control in Model 2 and RoboTaxi. This is how cars are born for the first time. This is something Waymo cannot do. In fact, Waymo achieves autonomous driving by controlling the gas pedal, brake pedal, and steering wheel, but lacks control over components. However, Tesla is moving towards creating RoboTaxis, and in order to do that, the controller needs to be fully controlled. This part is something that Model Y and Model 3 cannot currently achieve.

So, creating these kinds of things, for example, look at the Cybertruck. Cars have a lot of wires for electricity. In the Model S, the wires go in about 2km. But in the Cybertruck, it's 100m. Why is that? It's because they connected the controller with Ethernet. But now, the most important things like the brake system or gas pedal, especially in the case of a Cybertruck, instead of using wires for the steering wheel, it moves through signals. Now, for things like this, they connected it with Ethernet just in case, but they actually connected it with wires. In reality, Tesla is making tremendous innovations internally, although not so noticeable to us. This ultimately leads to the birth of a car and symbolically shows how much this will eventually lead to the birth of a car.

If it's difficult, this time, Rivian's stock price went up due to the joint venture investment case with Volkswagen. How much did they invest? What did they decide to do? They decided to make software for vehicles. Here, you should pay attention to how difficult it is to make software for vehicles. Even though it's important whether Volkswagen, a big company, will team up with Rivian, it doesn't mean that Volkswagen hasn't invested in software until now. No, they have. They created a subsidiary called Car.R which was established in 2020, and there are 6,000 developers working there. However, after trying for 3-4 years, within the legacy organization,

The software didn't come out. That's why there must be a huge amount of manpower, and then really smart people gathering, right? But now, those 6 people gave up and invested in Ria. That's right. It caused a stir in the German media. That's why even the German political circles are surprised. It's a very big car company in Germany. Audi is here, Porsche is here, Volkswagen is here, and there are also trucks like Scania, buses like MAN, and various companies like Skoda, Seat, Bugatti, Lamborghini. Volkswagen has all of them, really. This company employed 6,000 developers and has been working since 2020, but they are giving up on this. What's going on? Germany seems so incompetent. When they did this, many reports came out in Germany, and Audi keeps contacting them.

Let's say this time, even in our hybrid vehicles, you guys should create the software. Porsche contacted us and asked us to try something on the Taycan this time. Continuously, requests keep coming in to add software to the products we are currently releasing. That's why we are falling behind. If we start from scratch, if we design a car from the beginning, a car designed as an electric vehicle from the beginning, and evolve like Tesla, taking control of the controller from 30% to 60%, 80%, and up to 100%, we can create a new concept of a vehicle while evolving the software. We need to have a medium to long-term plan to evolve the software here, focusing only on this. If we keep doing various things that already exist, our strength will be dispersed. It's like saying our DNA has internal combustion DNA. We keep being influenced.

Without even leaving Germany and moving to the United States, other executives in the Volkswagen Group boldly decided to create a joint venture to develop software, preventing other executives from contacting them. This is about developing software that goes into vehicles. So, the car infotainment system we have now, where we enjoy entertainment, play games, watch Netflix, and even include karaoke like in China, is completely different from vehicle software. They are creating an operating system to control all components, even if fewer parts are needed for electric cars. This emphasizes the importance of software innovation. In terms of global revenue, formally ranking second in the world after Toyota, this second-ranking company is also giving up on this. So, you need to pay attention to these points. Hyundai and Kia have achieved the highest sales since their establishment.

What is happening is all because of electric cars. But these days, Kia is not there, so I can't buy it. In the U.S. or Europe, has Hyundai ever had a time when they couldn't sell cars? To the extent that they sell so many electric cars in the U.S. and Europe, Hyundai and Kia are benefiting from the overall sales. So, electric cars have significant meaning. Really, Hyundai and Kia sell the most cars on their founding day. So, what's next? From the perspective of Hyundai and Kia, electric cars have come in handy. This is a bit hardware-related. So, here, there are two things. One is vehicle software. So, nowadays, with SDV, software-driven, like basic, and such concepts are being incorporated. So, they are evolving vehicle software or autonomous driving like FSD.

You have to create. The reason why Tesla is not competitive in China is that in China, Tesla is just a pure electric car. Currently, Tesla has two functions. One is the autopilot function, which changes lanes on the highway, exits, and so on. When passing through military bases, it doesn't work. It doesn't work near airports, and it doesn't work near major government facilities. If you go to places like these, it doesn't work. So, it does work, but the operational segments are too short and disconnect too often. However, how did they solve this? This time, they are using the Baidu mapping service in China and placing related servers in China. Some of you may think that Chinese people are also bad or something like that, but the same goes for Korea. The reason why mapping services in Korea are not working well is because the servers are not located in Korea. In our country, military facilities are.

Many, so we manage services in these areas very well. It's not that China has blocked it, but China will likely be more strict. So, we have decided to use Baidu's maps and are currently installing servers next to the gigafactory in Shanghai. Therefore, testing is being done in Shanghai, and once this testing is completed, Full Self-Driving (FSD) can be properly released nationwide in China. Ultimately, for Tesla to be competitive in China, FSD functionality must be available. This means that China is preparing for the era where the competitiveness of electric cars will no longer be based solely on the vehicle's exterior. So, is Hyundai Kia preparing for this shift as well? In other words, transitioning to a new era where electric cars are not just about the exterior, but about integrating software and autonomous driving to sell vehicle services.

The market is evolving towards autonomous cars. So, are Korean companies preparing for this? Korea is signing up for this. It is necessary to consider this aspect for the Korean economy. Lastly, I have a question. Most of the robot taxis that were tested with permits in the past had LiDAR installed. But now, companies like Tesla are using cameras for vision. Even in this case, there is no problem with getting permits from the U.S. government or similar authorities. FSD is already permitted. Of course, the U.S. National Highway Traffic Safety Administration is always monitoring closely. However, things are evolving. Currently, what is being acknowledged is that China's Xiaopeng Motors had LiDAR, but now they are starting to switch to vision-only without LiDAR. They are starting to imitate Tesla's end-to-end approach, moving away from LiDAR, showing that it is possible to do without it.

The level of technology has advanced. For example, what was considered impossible in 2020 is now achievable through current vision principles in 2024. In this aspect, it is clear that we are moving towards an era where riding without radar is possible. I firmly believe that now, like the Chinese government, is experimenting with three things: traditional radar, radar-equipped robot taxis, and services like vision FSD, which are technologies like autonomous driving that can be attached to existing vehicles. The Chinese government, for instance, is incorporating things that can emit signals on the road, allowing vehicles to communicate, thus creating a road that enables autonomous driving to work well here. They are testing various ways to create such roads. By testing various methods, they are likely to choose those who can survive in the market. In addition, Chinese car companies, for instance, currently have 16...

We are conducting tests such as autonomous driving in 19 cities in China, so it is being carried out on a very large scale right now. In the case of China, it is uncertain which technology will win in these areas, so I can't guarantee it. China is almost aggressively pushing for quantity, while in the United States, there are actually two things going on. One is the Waymo method, and the other is the Tesla method. Waymo was thought to be the only one that would win for a long time, but Tesla has been slowly catching up and is now suddenly making a leap, chasing after Waymo. In these aspects, as of now, of course, autonomous driving is ultimately aimed at FSD, which is about Level 4. So, I am currently observing progress up to Level 4. Yes, what is currently applied with FSD requires the driver to hold the steering wheel.

First of all, now it's at level 3. No, it's not level 3, it's actually officially FSD level 2. From level 3 onwards, if an accident occurs, it's not the driver's responsibility but the manufacturer's responsibility. Because of the manufacturer's responsibility, companies that have reached level 3 now are Waymo and then Mercedes-Benz, which are operating at level 3 on restricted roads in the United States. However, even at level 3, there are limitations. It doesn't work when it's raining, snowing, or at high speeds, meaning it can't drive over a certain speed limit. So, level 3 is progressing with these restrictions. If there is an accident in the level 3 section, Mercedes-Benz will take responsibility for the accident. That's why FSD always comes first. It's FSD that is supervised by a human under supervision. So, technically speaking, experts evaluate that FSD 12.4 is at about level 3 or level 4.

The moment you remove the super visa, the responsibility just goes straight to Tesla. That's why human supervision is mandatory. However, in the case of a robocar like a RoboTaxi, when the super visa is removed, if an accident occurs, Tesla will take full responsibility. If they can demonstrate the same level of safety from an external perspective, even if it says super visa, people will likely stop paying attention and may end up driving with their hands off. So, in order to build trust, RoboTaxis may still be necessary. Currently, officially, it's at level 2. But if RoboTaxis are permitted to operate without human drivers, Tesla could be recognized for achieving level 3 or 4. Once recognized, it will have marketing effects.

I also want to use FSD for my car. I will take responsibility, but since there are no accidents with robot taxis and they are running so well, I think I can use them too. Users will have this kind of thought, and YouTubers are just filming without any concern. Actually, that's not legal. So, if there is an accident, the responsibility lies with the individual. Ultimately, robot taxis aim to bring in revenue through robot taxis to Tesla in a short period of time and take all the revenue from the taxi industry in the U.S. That's not right. Initially, it will have a marketing effect. Waymo is going to taxis anyway, and they're not going to sell it. As you mentioned to the general public, when Tesla's robot taxi is running, among all car makers, Tesla is at level 3-4.

There is an obligation to do, but the technology itself is becoming a car equipped with technology that can walk alone. I want to live like this. That's right, isn't it? Yes. So, the most important thing is that now, in the case of autopilot, it can only be used on highways, but people will use it for commuting tomorrow. It's not like you can drive drunk or anything like that, but if you can say, I can use this comfortably in my commuting vehicle, you can drive more comfortably and move around. This increases the adoption rate of FSD. You have to pay for FSD every month, but which car company has made money by doing that every month? As the number of people paying for it increases, and the most important thing is, we can ask the question in reverse. Why did China approve FSD?

"테슬라는 로보택시로 돈 벌려는 게 아니다" (강정수 박사)

Technology(기술), Trend(트렌드), Transformation(혁신)에 관해 
꼭 알아야 할 지식.  

일하는 사람들의 지적 동반자 티타임즈TV. 

www.ttimes.co.kr

Contact
주소  서울시 종로구 청계천로 11(서린동,청계한국빌딩)
e-mail  ttimes6000@gmail.com
Tel  02-724-7718

티타임즈TV

The Evolution of Autonomous Driving Technology

Elon Musk likens cars without Full Self-Driving (FSD) to feature phones, highlighting the importance of autonomous technology.

The partnership between Volkswagen and Rivian has caused disruption in the German automotive industry, as Volkswagen, despite its size, had to invest in external software development.

Autonomous driving technology is advancing, with companies like Waymo and Sura making significant progress and receiving substantial investments.

The potential for autonomous driving is particularly high in freight trucks, tractors, and military vehicles, leading to a growing number of startups in the industry.

The Evolution of Self-Driving Cars: From Testing to Commercial Service

GM faced a setback with its self-driving car, leading to leadership changes and restructuring, but is now reinvesting in testing.

Waymo has taken a gradual approach to autonomous driving, starting with backup drivers and now launching a commercial robo-taxi service.

The transition to widespread use of robotaxis may be slow due to societal resistance and the need to change the existing taxi driver system, while the use of data prediction apps is crucial for the success of robo-taxi services.

The Future of Autonomous Driving: Tesla, Waymo, and the Era of Robot Taxis

Tesla and Uber could potentially collaborate, allowing customers to choose between Waymo and Tesla's Robo-taxis within the Uber fleet.

Waymo has been recognized for its safety as a robot taxi in San Francisco, leading the way for the era of autonomous driving services.

Tesla is advancing its Full Self-Driving (FSD) technology, with plans to increase adoption rates and introduce new hardware, but the required installations may lead to increased vehicle prices.

Tesla's Innovations in Car Design for Autonomous Driving

Tesla's on-device AI hardware 3 and 4 are now essential for autonomous driving with FSD, requiring them to be sold together and making it possible to license the technology for existing electric vehicles.

The company's goal of achieving full control over car components, from the brake system to the controller, sets it apart from competitors like Waymo.

Tesla's internal innovations, including the use of Ethernet and signal-based systems, are revolutionizing car design for the future.

Volkswagen's Joint Venture Investment with Rivian for Vehicle Software Development

Volkswagen made a significant investment in a joint venture with Rivian to develop software for vehicles, after struggling with their own subsidiary's software development for years.

The decision to focus on evolving vehicle software is a bold move to take control of the controller from 30% to 100% and create a new vehicle concept.

This joint venture aims to create an operating system to control all components of the vehicles, emphasizing the growing importance of software innovation in the automotive industry.

The Impact of Electric Cars on Hyundai, Kia, and Tesla

Hyundai and Kia are benefiting from the increasing sales of electric cars in the U.

Both companies are also evolving their vehicle software and autonomous driving capabilities to adapt to the shifting market trends.

Meanwhile, Tesla is making efforts to improve its Full Self-Driving (FSD) functionality in China by using the Baidu mapping service and installing servers in the country, in order to remain competitive in the Chinese market.

The Advancement of Autonomous Driving Technology in China and the United States

Technology has progressed to the point where radar-free driving is now possible, with the Chinese government experimenting with traditional radar, radar-equipped robot taxis, and vision FSD.

Chinese car companies are conducting large-scale autonomous driving tests in 19 cities, while the United States is pursuing the Waymo and Tesla methods, with Tesla rapidly catching up to Waymo.

The current focus is on achieving Full Self-Driving (FSD) at level 3 or 4, with Tesla potentially being recognized for level 3 or 4 once RoboTaxis are allowed to operate without human drivers.

The Legal and Ethical Implications of Tesla's Full Self-Driving (FSD) Technology

Using Tesla's FSD for personal use raises legal responsibilities in the event of accidents.

Tesla's goal to dominate the taxi industry revenue through robot taxis raises ethical concerns.

The increasing adoption of FSD for daily commuting raises questions about its regulation and the company's profit motives.