Mention the word "cost", there is a car company that is extremely sensitive to costs, or a car company that is extremely urgent to reduce costs, that is, it is indispensable, known as a car passed down for three generations, millions of kilometers without major repairs, full of craftsman spirit toyota motor.
On April 7, Toyota Motor's subsidiary Woven Planet said that it would use a single visual solution to develop autonomous driving in assisted driving and higher-level autonomous driving projects. That is, in the absence of expensive sensors such as lidar, data is collected through relatively low-cost cameras to promote autonomous driving technology. At this point, Toyota has also joined tesla and other companies to oppose the use of lidar in self-driving technology.
At present, the autonomous driving technology for automotive applications is mainly divided into two camps: one is the lidar faction, which advocates lidar as the leading, with millimeter-wave radar, cameras, etc., to achieve multi-sensor fusion and improve the safety of automatic driving; one is the pure visual faction, which tends to use low-cost cameras, supplemented by artificial intelligence algorithms, to reduce costs.
Toyota This is... Blatantly anti-water?
Obviously, Toyota is like Tesla, using pure visual methods to achieve automatic driving, Toyota this time can be described as a big anti-water, lidar faction was originally proposed by Toyota, Nvidia, Google, Volkswagen, Daimler, Bosch, ZF, Waymo, Cruise, Uber and mainland Baidu, PONY AI and other autonomous driving technology research and development companies are surrounded by this camp. Lidar companies believe that safety is the first logic of automatic driving, and tend to the fusion of high-cost lidar, millimeter-wave radar, ultrasonic radar, cameras and other sensors to cope with multiple application scenarios of automatic driving. The advantages of multi-sensor fusion technology are very obvious that the computing power is strong, the flexibility is high, and it can have higher redundancy; but its disadvantage is also very prominent, that is, the high cost of technology application and the large demand for chip computing power, which cannot be installed on all levels of models at this stage.
Toyota's self-proposed multi-sensor fusion camp against water this time is really a bit unexpected. Toyota, which used to install seven or eight lidar on the test vehicle, actually chose to give up, what is worth thinking about behind this?
First of all, "cost", camera costs less, this is indeed a fact. 2 million pixel camera average at 150 yuan, 5 million pixel camera price is about 300 yuan, 8 million pixel camera single cost of more than 500 yuan, and a single cheapest double wedge prism lidar also needs more than 1200 yuan, Toyota well, can give you a 2 million pixel camera is good, even if you install 8, only with a lidar almost the price. For Toyota, which resolutely implements TPS, how to reduce costs and how to save more money from between the teeth is really an important factor to consider?
Is this determination to perfectly interpret the "spirit of craftsmanship"?
As we all know, Toyota has an extremely shrewd way of building cars - Toyota Production System (Toyota Production System), referred to as TPS, which is also sought after and learned by many companies. TPS covers various concepts of production management such as timeliness, automation, Kanban mode, standard operation, and lean. Because of the complete production mode of TPS, Toyota can have a good control of the process, supply chain, production capacity, quality, inventory, etc., and at the same time can control the cost of car manufacturing very well, coupled with its own certain product strength and excellent marketing, toyota also makes Toyota sit firmly in the position of the world's most profitable car company in 2020.
Well, what is the control of TPS, toyota itself said, toyota products screened to the consumer side, and did not do what it said in "process, production capacity, quality, inventory", but cut corners, deliberately lower production capacity to engage in price increases, quality reputation by water army washing the ground, inventory a lot, that also has to increase the price, so it can also leave a car passed on for three generations, people are still walking the reputation?
Toyota's current vehicle cost is not high, you think about it, it is now the second year of the 2020s, Toyota's new Highlander, with a 4200 rpm or more to output the maximum torque of the natural suction engine, with a set of Toyota proud, until now do not give Chinese consumers to solve the oil increase emulsification THS hybrid system, the battery to you with nickel-metal hydride, lithium battery so "dangerous" so expensive how can you use, the lowest model of the whole car does not have a camera, reversing image can not have, There is no front and rear parking radar, fabric cloth seats, plastic steering wheel, hard plastic interior of the whole car, seat manual adjustment; perfect interpretation of the can plastic never leather, can be manually never automatic, can save the province, can be used on the line of "craftsman spirit", such a car if there is a configuration of lidar, is it too uncoordinated?
Toyota's pure visual approach can indeed significantly reduce costs, because there is no more sensor blessing, lack of perception redundancy, can only leave more pressure to the developers of the algorithm platform.
Imagine that a Toyota's small-displacement naturally aspirated engine sells for less than 5,000 yuan, the cost is conservatively cut by more than half, and the cost price of a lidar is almost close to the price of the engine. Now toyota is equipped with reversing images, reversing radar auto parts city, maybe you can add this business in the future, install lidar, to help Toyota automatically improve redundancy.
Secondly, Toyota's such a big fanfare into the embrace of pure vision may be influenced by Tesla, Musk is definitely a loyal supporter of pure vision systems, Tesla began to produce in the middle of last year began to gradually cancel millimeter wave radar, replaced by a pure visual system with only cameras. Musk has said tesla's FSD self-driving technology is smarter than human drivers. Many people think tesla is saving costs, but Musk responded that humans drive and control cars through pure vision systems.
Will Toyota's self-learning ability be stronger than Tesla's?
In fact, Tesla has been collecting data through the cameras on more than 1 million cars on the market, allowing the algorithm platform to fully self-learn, allowing the owner to become a "tool person" to help Tesla develop autonomous driving technology, pure vision to perceive environmental information, and based on the original video data of Tesla owners, through the neural network to build a real-world three-dimensional vector space, to achieve the car's behavior and path planning, and generate control signals to transmit to the actuator. Then, through the perfect data closed-loop system and simulation platform to achieve continuous iteration of autonomous driving capabilities.
Now, if Toyota adopts the same approach, Tesla's more than one million owners are really a dime a dozen here in Toyota, and its annual global sales of more than 7 million vehicles can become the best source of data. Toyota's global popularity and penetration are much higher than Tesla's, and if Toyota really rolls out like this, Toyota will have a considerable amount of data that can make its autopilot system perform sufficient deep learning.
And "we need a lot of data. It's not enough to collect a small amount of data from a small batch of very expensive self-driving cars," Woven Planet believes that equipping vehicles with expensive sensors is not a sustainable approach, and the large amount of data collected by cameras has great flexibility to expand richer capabilities at a certain hardware cost.
In addition, Toyota in the field of research on automatic driving, can be described as obscure, Toyota in 2018 set up an autonomous driving department, but did not disclose too much about the degree of self-development of autonomous driving, Toyota invested in Uber's divested automatic driving division ATG, to develop autonomous vehicle technology. In China, Toyota has invested in smart driving company Momenta and self-driving startup Xiaoma Zhixing. But Toyota's entry time is too late, and there is no achievement that can be brought to the table in recent years, and most traditional car companies are no different, so it is also a shortcut to directly use pure vision and cross the complex process of multi-sensor fusion?
Toyota to the current entry-level model, has been equipped with L2 level auxiliary driving, even the lowest three-cylinder Corolla also has L2 level driving assistance, as for why the low-end Highlander does not, ask Toyota, do not understand.
Toyota invests in pure vision, and traditional manufacturers all come to see it.
As one of the world's largest traditional car companies, Toyota has not been a pioneer in the road of intelligent and electrification transformation, and the retaliatory release of more than ten new energy vehicles at the end of last year has made the industry more and more unclear. Toyota's blatant anti-water this time has turned to the automatic driving of pure visual solutions, which is bound to affect other traditional car companies on the road of transformation. At this stage, the scheme of lidar + millimeter wave radar + camera + high-precision map is more secure, can provide more adequate perception of redundancy, so that the car itself is more aware of the world to face, but the corresponding cost is too high, the traditional car companies to take the entry model can not afford such a cost rise, at this point, Tesla will often adjust the price according to cost changes, and Toyota will? If you really have so many sensors, can you still buy 80,000 Vios and 120,000 Corollas?
So I believe that after Toyota uses pure visual solutions, there will be many traditional car companies to join it, and it is currently said that lidar is a shortcut to achieve automatic driving, but after a period of time, the pure visual solution that is more in line with the logic of the human eye will not have the upper hand and become a new shortcut, and it is necessary to see how much breakthrough this technology can have under the efforts of more car companies.
I think that for the traditional car companies that have not yet made much progress in the field of automatic driving, the use of pure visual solutions to complete automatic driving, although at this stage will be blamed, suspected, but also a quick realization of automatic driving research and development on the car program, after all, there are ready-made living examples, the cost is not high, the difficulty of pushing is low, the need for technology market can be bought, many owners who buy joint venture models may not have much hope for the intelligence of joint venture models, after all, it may be a map brand, or even a "high-level" And a low-cost intelligent driving system can bring advanced enjoyment, will definitely make the owner satisfied, and the need for few sensors, not much need for post-maintenance replacement, owners may also be happy to do so, said out "we this Toyota, and Tesla is a technology." Car companies and car owners are willing, so why not?