For the first time, the unmanned logistics vehicle has been reduced to the level of 100,000 yuan, and it is devoted to both technology research and development and application.

  Autonomous driving enters a new stage of "two-legged" competition

  Our reporter Sun Qiru

  Before the ultimate goal of "full driverless" autonomous driving is mature and implemented, unmanned delivery vehicles will undoubtedly become an important landing and monetization method for autonomous driving technology.

This year, stimulated by factors such as the special needs scenarios of the epidemic, low-speed unmanned logistics delivery vehicles are accelerating their landing as "AI couriers".

Recently, unmanned delivery vehicles have also accelerated the application process through price drops.

At the same time, Momo Zhixing completed hundreds of millions of A+ rounds of financing, Leading Technology completed tens of millions of yuan in Pre-A rounds of financing, and Pony.ai completed D rounds of financing... With the recent frequent financing of the domestic autonomous driving industry, the funds have continued to increase. With the influx, the research and development of various enterprises has accelerated, and the application scenarios of autonomous driving technologies such as assisted driving, unmanned logistics vehicles, and unmanned mining vehicles have been implemented one after another. Two legs" competition new stage.

  Unmanned logistics vehicles "lightly loaded" on the road

  Dressed in a Chinese New Year red color scheme, with a long tail raised, on the eve of the Spring Festival of the Year of the Tiger in 2022, on the streets of Shunyi, Beijing, a "little tiger" with a cartoon appearance and an AI technology core sent goods for the New Year of the Tiger to the residents of Shunyi.

This "little tiger" for automatic delivery is an unmanned delivery vehicle "little devil camel".

  I bought fresh food and snacks in the supermarket, and when I received a text message notification to pick up the goods, I found that the "courier" was an autonomous delivery vehicle.

Such scenes are no longer uncommon for citizens living and working in Yizhuang and Shunyi in Beijing.

With the changes in the epidemic situation in many parts of the country, many batches of automatic delivery vehicles have also been deployed in communities and distribution points in Shanghai and other places, reducing the pressure on community volunteers, couriers and other staff.

  Taking the epidemic prevention and control in the past two years as an opportunity, unmanned delivery vehicles have accelerated the landing.

In Beijing, Shunyi, Yizhuang, Haidian and other areas alone, there are self-driving terminal logistics vehicles from many companies such as JD.com, Haomo Zhixing, Neolithic, and White Rhino.

Take Wumart as an example. By the end of this year, all Wumart stores in Shunyi will use the end-of-life automatic delivery vehicle "Little Magic Camel 2.0".

  In recent years, with the rapid growth of e-commerce, intra-city services and other formats, the logistics industry has faced many difficulties in terminal distribution: such as express delivery, catering, fresh food, medicines and other distribution items, there are many types of distribution items, and distribution in communities, commercial areas, office areas, etc. The scene is complex, the distribution personnel are highly mobile, and it is difficult to hire.

In April last year, the Ministry of Human Resources and Social Security released the ranking of the “most short-sufficiency” industries in the first quarter of the year that recruited more than job-seekers, and couriers ranked eighth among the 100 occupations on the list.

  "Automatic distribution of terminal logistics is on the eve of the outbreak, and leading customers have begun to deploy large-scale scenarios." said Zhang Kai, chairman of Maomo Zhixing.

In order to meet the rapidly growing order demand for terminal logistics automatic distribution vehicles, the terminal logistics automatic distribution vehicle factory of Momo Zhixing in Baoding has been fully upgraded, covering an area of ​​10,000 square meters, and can achieve an annual output of 10,000 terminal logistics automatic distribution vehicles. The production capacity target of vehicles has become the world's largest flexible manufacturing base for terminal logistics automatic distribution vehicles.

  With the acceleration of mass production, the cost of automatic distribution vehicles has also accelerated.

On April 19, Haomo Zhixing released "Little Magic Camel 2.0", the bicycle price is 128,800 yuan, becoming China's first 100,000 yuan terminal logistics automatic distribution vehicle. The vehicle is expected to be launched on the market in May this year.

  "In Beijing, the labor cost of a courier is about 100,000 to 120,000 yuan a year. While improving stability and reducing costs through technical means, the price of 128,800 yuan is also based on business logic. Unmanned logistics The car must 'wait for others'," said Gu Weihao, CEO of Haomo Zhixing.

  "AI old driver" car skills are still improving

  Compared with the unmanned logistics vehicles that carry goods, the public and the industry still have "higher and stronger" expectations for the technical capabilities such as safety of the self-driving passenger vehicles that carry people before they are fully on the road.

  On April 19, Momo announced the upgrade of its autonomous driving data intelligence system MANA. At present, MANA has completed nearly 200,000 hours of learning, and the virtual driving experience is equivalent to 20,000 years of human drivers.

Over time, MANA will make the perception of the autonomous driving system more accurate and the cognitive decision-making more human-like.

  One of the manifestations of the improvement of cognitive decision-making ability is that the AI ​​​​old driver learns to read traffic lights.

  The traffic lights in different places vary greatly, such as horizontal, vertical, 3 lights, 5 lights... Even the traffic lights in the same area will have various real-time states such as digital countdown and flashing.

Gu Weihao said frankly that the most difficult thing is to determine which road and which lane a certain light is on.

  In the past, many autonomous driving companies could greatly reduce these difficulties in identifying traffic lights for autonomous vehicles with the assistance of high-precision maps.

What if there is no high-precision map?

In order to ensure the self-driving car's ability to "stand alone", it must improve its ability to recognize traffic lights and lanes without the assistance of high-precision maps.

According to Gu Weihao, the team created an original "dual-stream" perception model, which decomposes the traffic light detection and road condition binding problem into two channels, thus realizing traffic light recognition in the state of light dependence on high-precision maps in the daily passenger car test.

In addition, its heavily invested Transformer model is used for lane line recognition, which realizes the automation of labeling through the two stages of "target coarse localization" and "attribute fine estimation", and realizes multi-sensor fusion lane line recognition on urban roads. There are no unrecognized lane lines".

  However, the relevant person in charge of Maomo revealed that the coping ability of unmanned vehicles in some extreme scenarios still needs to continue to be improved, such as dealing with jamming at extreme close distances, dealing with small obstacles such as kittens that suddenly jump out, etc.

  In order to make up for the shortcomings of vehicle perception, industry companies such as Baidu proposed that in addition to bicycle intelligence, “smart vehicles and smart roads” should be realized through vehicle-road collaboration, so as to solve the “ghost probes” and unknown obstacles faced by unmanned vehicles. The scene problems such as the sudden intrusion of objects.

  Double breakthroughs in technology and policies are required for batches on the road

  In the past two years, not only Beijing, but also Suzhou, Shanghai, Shenzhen, Hefei and other places have successively introduced policies related to autonomous driving, opened up traffic scenarios in pilot areas, etc., to accelerate the implementation of autonomous driving technology.

For the industry, in addition to technical strength and technical route, at this stage, capital is also quite concerned about the progress of commercialization and mass production of the autonomous driving industry.

  For the general public, apart from taking self-driving taxis in some leading areas and picking up express delivery from unmanned delivery vehicles, the most direct way for them to connect with self-driving technology may be to buy a car equipped with an assisted driving system. .

The reporter learned that this year, a number of cars equipped with the domestically produced urban pilot assisted driving system will be launched soon.

  In commercial segments such as assisted driving and logistics, the autonomous driving industry has begun to see "returning money" while investing a lot in "burning money".

  However, whether it is L4-level autonomous driving or low-speed unmanned logistics vehicles, the double breakthrough of technology and policy is required for mass adoption.

"Taking Beijing as an example, in Shunyi, Yizhuang and other areas, after publicity and notification in advance, unmanned logistics vehicles can already go on the public road, but in Beijing's fifth ring road with complex traffic, due to the density of vehicles, safety, etc. Considering that, it may not be possible to open end unmanned logistics vehicles on the road in the short term." Hou Jun, COO of Maomo Zhixing, said that with the maturity of technology and the implementation of applications, batches on the road will eventually be realized.