Li Deyi, an academician of the Chinese Academy of Engineering: How does automatic driving go from L2 to L3?

Factory direct sales - solid capacitor - patch aluminum electrolysis, to ensure that the factory first-class goods!

At the China Artificial Intelligence Conference held on July 21st, Li Deyi, an academician of the Chinese Academy of Engineering and chairman of the China Artificial Intelligence Society, said, “If you can't even do automatic parking, what about the automatic driving L3?” Let's take a look at the related content.

As early as 2005, Citroen pioneered the automatic parking assist system, originally called the city park system, which only released the steering wheel. So far, where novices and female drivers can't park, the automatic parking system is also in a three-minute period. It is called an automatic parking hoe.

According to the classification of the Society of Automotive Engineers (SAE), autonomous driving can be divided into five levels, namely L1 driving assistance, L2 partial automation, L3 conditional automation, L4 high automation, and L5 full automation.

Li Deyi introduced that in recent years, the car-assisted driving has never been more prosperous, from L1 to L2 to L3. From the early warning, reminding the driver to cross the machine to drive the car, the role of the fixed driver no longer exists, the momentum of L3 mass production is not available. Block, but the breakthrough from L2 to L3 is not easy.

Li Deyi pointed out that almost all of the self-driving cars on the road, including Tesla, are still at the L2 level and must be controlled by people. In the 2016 Autopilot Report released by the California Vehicle Authority, the highest level of intervention frequency of 0.2/thousand miles, which level of autopilot?

The jump from L2 to L3 is the transfer of control of the car from person to machine. Where is it difficult? Li Deyi raised three questions: How to measure the automatic driving level conversion point? How to measure the control point handover? Accident in the process of handover How to measure?

At present, Audi's new L3 grade A8 will be launched at the end of this year, which will become the world's first auto-driving model car brand. The question is, will the Traffic Management Bureau issue a driver's license for such vehicles? Do ordinary people without a driver's license dare to take advantage of it?

Li Deyi pointed out that L3 faces the challenges of Germany and the mass production dilemma. In fact, it is the bottom line requirement for the driver's car to get a driver's license. To break through the ceiling of the L2, the key is to use artificial intelligence.

“Once the window conditions set by the automatic driving, such as the geofence, the climate fence, and the human customs fence, are broken, the handover of the driving control is required immediately. The accident in such handover may be more dangerous than pure manual driving. Li Deyi explained.

Then the problem is changed. The basic question of L3 is to solve the problem of the car, or solve the problem of people? Li Deyi explained that the so-called "car" refers to the better the car, the software-defined machine to achieve automatic driving. The so-called "human" refers to the replacement of the driver's cognition with a robot. It is the subject of people with memory, decision-making and behavioral ability. It achieves self-driving, skill and personality.

“Where does the interaction between the driver and the surrounding vehicle go during the automatic driving process? Who is the driver's experience and the ability to handle the situation?” Li Deyi analyzed, so the automatic driving is good at focus, never fatigue; automatic driving Difficult to be anthropomorphic, not only in the car.

During the driving process, the driver's prediction and control in the loop is irreplaceable by the car itself. Wheeled robot driving brain came into being. It is different from the sensor of radar and other sensors. It is necessary to complete driving cognition including memory cognition, computational cognition and interactive cognition, and become an important zero in the industrial chain of smart cars. Parts, in-vehicle computers and robotic operating systems are also not a substitute.

Li Deyi analysis, using microelectronics technology, using CPU + GPU + FPGA + ASIC architecture to produce dedicated chips and boards, research and development of the driving brain, autonomous response to the various uncertainties often encountered in the driving process.

"Different driving brains, cognitive levels can be different, skills and experience can be different, but they all have the basic driving cognitive ability, they are driven by the driver's license." Li Deyi explained.

Li Deyi revealed to the 21st Century Business Herald that his team has cooperated with companies such as Yutong, Chery and SAIC, and is developing and industrializing the driving brain based on the automatic driving road test.

The above is about automotive electronics - Li Deyi, an academician of the Chinese Academy of Engineering: How does automatic driving from L2 to L3? For related introductions, if you want to know more information, please pay more attention, electronic engineering will provide you with more complete, more detailed and updated information.