Wen | the low-speed driverless newsroom
In the field of terminal distribution, Ali relies on a complete e-commerce and logistics system to launch an unmanned vehicle "Little Wild Donkey" for the "last three kilometers" express delivery, and successfully moved towards large-scale commercialization, using 350 vehicles in more than 200 colleges and universities across the country.
However, on Friday, the "little wild donkey" unfortunately had an accident at a university.
Inside Henan University, the unmanned car "Little Wild Donkey" drove directly into an undried cement field that had not been buried for a long time, and was firmly immersed in it, causing students to laugh and watch.
Some people questioned: "On this technology, do you still want to grab jobs with couriers?" ”
It is reported that this is not the first time that the "little wild donkey" has encountered embarrassment, and there have been "immobile" in other parks before, such as the front is blocked by bicycles or road construction, etc., and can only wait in situ for rescue, and too many people next to it will be "frozen" when they are watching.
The problem of the long-tail scenario of autonomous driving
"Little Wild Donkey" is designed and developed by the Autonomous Driving Laboratory of Ali Damo Academy. In terms of hardware, the "little wild donkey" is equipped with a deeply customized 32-line lidar in front and rear, and there are 6 cameras around the car to form a surround view scheme, as well as millimeter wave radar, inertial guidance and other sensors. In terms of software, "Little Wild Donkey" integrates the most cutting-edge artificial intelligence and automatic driving technology of Damo Academy, has human-like cognitive intelligence, the brain emergency response speed reaches 7 times that of humans, and also designs a five-fold redundant safety system to ensure the safety of unmanned vehicle operation.
Thanks to the cooperation of multiple sensors, the "little wild donkey" has the ability to actively avoid obstacles, but it still "overturns" in this wet cement field.
Some analysts believe that although the "little wild donkey" is driving in a relatively closed campus, the actual situation is far more complicated than what the naked eye sees. Unmanned vehicles operate in closed places, and most of the driving routes and stops are planned in advance, which also means that this cement land that has not been landfilled for a long time has not been updated to the map library of "Little Wild Donkey" in time.
In the case of the map route update is not timely, the "little wild donkey" failed to avoid obstacles for two main reasons:
First, during the day, cameras and lidar failed to identify wet cement floors. Generally speaking, for normal unmanned vehicles, this kind of ground should be judged as a mirror reflection and treated according to the surface of the water. However, after the cement is 80% dry, there is no reflection, and it is likely that it will not be judged as water.
Second, the wet cement scene belongs to the blind spot scene tested by the "little wild donkey", similar to the "long tail scene". Long-tail scenes refer to a wide variety of scenes with low probability or sudden occurrence, such as vehicles running red lights, pedestrians crossing the road, intersections with damaged traffic lights, and vehicles parked illegally on the side of the road. These scenes do not play cards according to common sense, there are many styles, and it is difficult to deal with, but they are one of the keys to the landing of automatic driving technology.
Because the autonomous driving function is based on AI algorithms, with current technology, AI algorithms can only complete tasks that have been trained before. Autonomous driving technology cannot simulate all complex road conditions, and can only approach infinitely without completely finding all scenes, so there are many "long-tail scenes" that have not been solved.
From a certain point of view, this time the occurrence of new accidents and new failures is also to help find new "long-tail scenarios" in unmanned vehicles and solve them, so as to promote the pace of scale and commercialization.
Large-scale commercialization: the road is obstructed and long, and the line is coming
As early as the end of 2015, Ali set up a cainiao ET logistics laboratory, focusing on the development of unmanned logistics vehicles.
In June 2018, Zhang Chunhui, then director of Ali Cainiao ET Logistics Laboratory, said: "In the next three years, Ali Cainiao unmanned equipment will reach 100,000 units. ”
In October 2021, on the first anniversary of the launch of "Little Wild Donkey", Ali announced that "Little Wild Donkey" completed the "from 0 to 1" mass production climb, sold more than 350 units, and entered more than 200 universities and communities. But it's clear that Ali's goal three years ago was far from being achieved.
Speaking of which, the technical team behind the little donkey of Ali Dharma Academy has also experienced a lot of turmoil:
Around 2018, Ali's self-driving car research and development tasks were still concentrated in the Cainiao ET Lab, but then Cainiao and the Damo Academy's Autonomous Driving Lab merged after structural adjustments, released the Little Wild Donkey in 2020, and established The Hangzhou Little Wild Donkey Intelligent Technology.
Unfortunately, Wang Gang, who led the "Little Wild Donkey" from birth to landing, a scientist at the autonomous driving laboratory of the Dharma Academy and the general manager of the Little Wild Donkey Intelligent Technology, announced his resignation in January this year and began to start a business in the field of cleaning robots.
Some analysts believe that it is the Internet company that overestimates the value of technology and underestimates the difficulty of technology landing, so that the payment and output of technology bulls are not proportional, resulting in technology bulls leaving the Internet.
Some people also believe that Wang Gang's departure may be related to Ali's multi-route expansion in the field of automatic driving. He has always advocated that when making products, "we should focus" rather than "both and want". Ali obviously has not satisfied the layout of unmanned delivery vehicles, not only to launch the "big wild donkey" into the trunk logistics field, but also to lay out the direction of passenger car automatic driving.
In fact, with reference to the "three laws" of automatic driving development, no matter how Ali expands the autonomous driving route in the future, the low-speed, limited scenario of unmanned distribution will always achieve large-scale commercialization faster.
At this stage, Ali has realized the layout of 350 "little wild donkeys", and the latest plan to expand to 1,000 units in 1 year and deploy tens of thousands of units in 3 years is also gradually approaching. Although there are still many difficulties to be overcome in the operation process of unmanned vehicles, we firmly believe that the industrial chain symbiosis, co-creation and win-win situation will surely usher in a new development situation.