Baidu patent decryption: How driverless vehicles communicate with each other

Recently, the 3rd World Intelligence Conference 2019 has just ended. At the conference, the MUSE concept car produced by the company became the focus of everyone's attention.

This car is equipped with an advanced vehicle intelligent interconnection user experience platform. The car has no steering wheel at all. The camera captures and measures the surrounding data and transmits it to the main control chip for calculation and issuance of instructions. The driver only needs to sit in the cab and input the desired address. The vehicle can automatically reach the destination under the guidance of the high-precision navigation system. This also means that China has entered a new stage in the technology of unmanned driving.

Although the technology of autonomous driving has entered a new stage, it also faces some challenges, the most prominent of which is: understanding of pre-behavior.

Autonomous vehicles often need to perceive the subjective intentions of other vehicles nearby in order to make timely predictions and correct driving decisions. Therefore, timely communication between autonomous vehicles is extremely important.

The editor found a patent filed by Baidu (application number: CN107454128A), which proposes a system and method for providing vehicle-to-vehicle communication between unmanned vehicles.

First, let us understand the system as a whole, as shown in Figure 1.

Figure 1 Overall diagram of the communication system between unmanned vehicles

As can be seen from the figure, the intermediary for communication between different groups of unmanned vehicles is the cloud server. The communication between vehicles in each group can be carried out through the internal LAN.

In order to enable smooth communication between vehicles, it is generally necessary to define the data structure for storing vehicle information on the cloud server in advance.

It can be seen that the information attributes of each driverless vehicle include: AV ID, which is used to uniquely identify the vehicle, the current vehicle location information (latitude, longitude and sea wave height), and the vehicle's current driving status (driving speed and driving direction).

Figure 2: Storage data structure of unmanned vehicle

Finally, the author will introduce in detail the communication processing process between unmanned vehicles, as shown in Figure 3.

Figure 3 Flowchart of the communication process between unmanned vehicles

The vehicle's internal logic system will obtain the vehicle information of the unmanned vehicle, and the detailed information is the attribute values ​​of each column in Figure 2.

The first unmanned vehicle transmits a vehicle identifier (ID) and vehicle information of the first unmanned vehicle to the cloud server via the network. The vehicle information may include location information, speed, or moving direction of the first unmanned vehicle.

The cloud server is communicatively connected to the plurality of unmanned vehicles via a network. The cloud server may maintain a data structure or database to compile and store vehicle information of the unmanned vehicles. Some of the vehicle information may be received from the unmanned vehicles, while other information may be compiled and generated at the server based on updated information received from the unmanned vehicles.

The first unmanned vehicle receives a list of one or more vehicle IDs identifying one or more unmanned vehicles from a server. The server identifies one or more unmanned vehicles within a predetermined geographic proximity of the first unmanned vehicle. Based on the vehicle IDs, the first unmanned vehicle communicates with a second unmanned vehicle selected from the list via a wireless local area network to exchange an operating status of the first unmanned vehicle with the second unmanned vehicle.

The patent proposed by Baidu for "System and method for providing inter-vehicle communication between driverless vehicles" provides a feasible solution for understanding the predictive behavior of driverless vehicles. The editor is looking forward to it.