Why AI won’t work in cars? Let’s look at the real problem

liu, tempo Date: 2021-09-17 09:51:55 From:ozmca.com
Views:40 Reply:0

The invasion of AI is not only in the field of media, but also in automobiles, one of the biggest application scenarios of AI. If the AI automatic driving of L4 and L5 can be popularized to 100,000 level cars, then the driver can also be replaced, which is also the real realization of automatic driving that we often say.


However, it seems to be a bit distant now. Tesla, which is the best in automatic driving at present, has heard the news that its Autopilot accidentally caused a rear-end collision and hit a passer-by. For this reason, The proud Tesla added the word “assist” before its “Autopilot” to remind the car owners who are eager to try.


However, the l2-L2.5 level of autonomous driving technology has been produced in mass production, most of which is still limited to active parking or driver assistance system, and there are “old drivers disdain to use, novice dare not use” embarrassment. Who exactly can answer the “why should I pay for something THAT I don’t want to use and is unreliable” question? This is probably called technological iteration, just like 4G to 5G, traditional cars will be eliminated by smart cars. Where do you buy feature phones now?


Ways to embrace AI

Since smart cars are irreversible, all that remains is how to hug. The traditional car is much more closed than the mobile phone industry, but since the emergence of Waymo and Tesla two disrupters, the global smart car steps are bigger. This industry players, generally speaking, there are several patterns in the play, represented by Waymo and Tesla automated driving mode of radical and closed, from L2 directly to L4, the road has a large team or customer base, a day for automatic driving system to provide the traffic information and driving behavior data, to the background of autopilot system for training and iteration.


However, this kind of game traditional car companies are disdained, they prefer to be with the same level of people alliance. Ford’s alliance with Volkswagen, BMW’s alliance with Daimler-Benz, and GM’s Cruise have also attracted Honda and SoftBank. And the third way, that is, Internet companies, enter the intelligent car industry the fastest and most widely spread mode – open platform, with money to burn out the circle of friends.


 automatic driving


BAT+H uses money to burn friends

BAT in the field of has entered the car and the big travel + H in the semi-finals, open to all car companies use most of the system, let all people sharing system, every car companies to provide data for system, the system of the ecological building more perfect and powerful, so system and feedback to participants, let them become more competitive in their respective fields. According to the comprehensive collation of the above data according to the public data, baidu and Alibaba have the deepest penetration in the automobile industry at present, followed by Tencent, while Huawei has technological advantages in the underlying architecture and Internet of vehicles communication.


Let’s take Baidu as an example. If Alibaba, Tencent and Huawei enter the field of intelligent vehicles for the sake of moat, Baidu is in urgent need of finding a profit point, so it is the most motivated to expand its circle of automotive friends. According to incomplete statistics, there are 14 traditional auto companies cooperating with Alibaba, 8 with Tencent and 20 with Baidu.


In addition, Baidu’s circle of automotive friends includes both traditional giants such as BMW, Daimler and Volvo and new forces such as WeIMar and Ideal Automobile. Besides, Baidu participated in the early venture capital of NIO and Weimar. AI (artificial intelligence) and autonomous driving were tactically determined early on. As early as 2017, in the autonomous driving field alone, Baidu launched the Apollo fund “Double Hundred Plan”, saying that it will invest 10 billion in three years to complete the investment of more than 100 projects, which is the largest autonomous driving special fund in the world.


Now it seems that Baidu’s layout in the car industry is indeed leading, but the most important is to throw money back; The reason behind it has a lot to do with missing out on the mobile Internet era. In the past two years, no matter baidu’s medical bidding ranking, which has been criticized by the whole people, or its market value falling below 40 billion, it fell behind BAT and was overtaken by JINGdong and Meituan.


Baidu, which once enjoyed unlimited popularity in PC, now loses the power of rapid growth and its advantage in the search field will disappear. In this way, Baidu is compelled to look for a new wind and bet on AI. Car travel, intelligent transportation and other fields are one of the main scenes of AI realization, and of course, it attaches great importance to them.


The question is, is Baidu profitable in the auto industry? At least in the area of AI, where the hype is high, it is far from profitable. At the baidu AI developer conference earlier, Robin Li spent a long time introducing baidu AI’s achievements in the automotive field, and invited Li Shufu to the scene through AI. Such action is seen as Robin Li regards the automotive field as a very important profit point for Baidu AI, even if it is not “ALL IN AI”, it is deepening IN this temporarily unprofitable field.


The real question: What magic can 2,000 yuan do?

Some experts say that the coming L3 level of automatic driving is the water ridge of intelligent vehicles, enterprises invest more in technology, capital, more moat, and only focus on a certain segment of the field and have enough leading position, it is possible to continue to obtain the blood transfusion of funds in the war of burning money.


Baidu’s Apollo or AI programs are advanced in this regard and fit well with what experts call a “technological moat.” But it is well known that a new technology is expensive until it is widely used. Statistics show that at present, for a joint venture car of more than 100,000 yuan, the hardware and software cost that the manufacturer really invests in intelligent network connection and intelligent driving is only 2,000 yuan.


At present, within the range of 2000 yuan, BAT+H layout is mostly the software at the bottom of the vehicle. If more sensors are involved in the project, such as limited automatic driving and complex automatic parking scheme, it will only stay at the level of demonstration, or installed in luxury cars and high-equipped cars, which is restricted by laws and regulations. But it’s more a game of technology and cost.


In AI developers conference earlier baidu, robin li shows the independent parking solutions, according to the scene of a video display, from the inside of the building science and technology park road to baidu baidu, self-driving cars can be implemented by the user mobile phone remote call vehicles, vehicle automated driving from the basement to the front of the building, and parking library automatic queuing at the destination.


According to baidu’s research, there are 270 million cars in China at present, and the gap of parking space is as high as 60 million. Meanwhile, 44% of the utilization rate of parking space is very low, and the search for parking space takes up 30% of the driving time. Baidu’s solution can solve this pain point. In addition, we have reached cooperation with weema, Great Wall, Ola, Volkswagen, Hyundai, Lifan, GAC, Jaguar Land Rover, Gofun Travel, Panda Travel, Xiaerrena Car rental, Parking simple and other enterprises.


Yes, the technology of the new scheme is very high, and a small range of L3 level of automatic driving, the work of road perception is completed by the car, the owner does not need to pay attention to the road, to achieve the liberation of eyes; At the same time, the market demand is great, but it is probably impossible for Baidu and his friends to reduce the cost before it is popularized. Therefore, Robin Li said in a pun: “As you can see, there will still be all kinds of unexpected difficulties on the way forward of AI.”


If the cost of AI on a bike can’t come down, then it’s a good idea to try intelligent transportation. According to Li, Baidu’s large-scale deployment of intelligent traffic lights in Baoding has reduced the traffic delay time of local residents by 20-30 percent during morning and evening peak hours. However, li did not disclose whether or how much the project is profitable.

Leave a comment

You must Register or Login to post a comment.