Prediction of the Development trend of Automobile sensors in the era of Intelligent Network connection

During the two sessions in 2022, a number of deputies to the National people's Congress of the automobile industry put forward a number of proposals related to intelligent networking, including the establishment of a legal system, the creation of operating system ecology, the protection of automobile data privacy, and the promotion of domestic specification-level chip industrialization. It has become the common vision of the people in the automobile industry and the long-term goal of the development of the automobile industry to build the advantage of intelligent network-connected automobile system. The strong promotion of the intelligent network connection automobile industry system will also make the vehicle sensor usher in a major opportunity for rapid development.

Autopilot drives a surge in the carrying capacity of bicycle sensors

Intelligent networked vehicle is a new generation of cars that can perceive complex environment, achieve intelligent decision-making, cooperative control and execution by carrying advanced on-board sensors, controllers, actuators and other devices, combining modern communication and network technology. According to the latest standards of the American Society of Automotive Engineers (SAE), the automobile driving automation system can be divided into six levels: L0 (no autopilot), L1 (driving assistance), L2 (partial autopilot), L3 (conditional autopilot), L4 (highly autopilot) and L5 (fully autopilot). At present, 90% of the new energy vehicles can reach the L2 level (partially self-driving), equipped with ACC adaptive cruise, lane maintenance system, automatic braking assistance system and automatic parking system, etc. A small number of new energy vehicles can reach the L3 level and add functions such as automatic emergency braking, which can automatically judge whether the driving environment can drive automatically in a specific environment and road sections, and automatically prompt the driver to take over the operation on complex road sections; from level L4 into the real sense of self-driving

Driving, which is still a restricted area; at the highest level, L5, the vehicle can drive automatically in any area.

▲ data source: Sadie consultant collation

The higher the degree of automation of the car, the stronger the dependence on sensors. At present, according to the different purpose of equipment, vehicle sensors can be divided into two categories: traditional sensors to improve the informatization level of bicycles and intelligent sensors to support self-driving system. Traditional sensors include pressure, position, temperature, acceleration, flow, gas sensors and so on, which are mainly used in powertrain system, chassis system and body control system. When the automobile enters the era of intelligent network connection, in addition to the necessary traditional sensors, more intelligent sensors are needed as the facial features of the car to perceive the external environment and internal human-computer interaction. At present, the mainstream vehicle sensors used for environmental perception include lidar, millimeter wave radar, ultrasonic radar, cameras and so on; the sensors used for vehicle personnel perception and interaction mainly include driver monitoring camera system, voice recognition sensor and so on. The number of environment sensing sensors assembled by high-level self-driving vehicles is generally about 30. With the gradual popularization of intelligent networking vehicles, the demand for on-board sensors will increase exponentially.

▲ data source: Sadie consultant collation

Second, the development trend of automobile sensors in the era of intelligent network connection.

The requirement of low delay urges the vehicle sensor to develop in the direction of near sensor computing. Self-driving vehicles need to quickly analyze data and make real-time decisions, while in the traditional perceptual computing architecture, the separation of sensors and computing units in physical space will lead to serious signal delay. The use of near-sensor computing (such as edge computing, fog computing) can effectively solve the problem of information delay. Among them, edge computing technology is to install an open platform with storage, computing, network and other functions close to the sensor, such as placing the automobile electronic control unit near each sensor to receive and process sensor data nearby. and make corresponding decisions. Fog computing places a large number of servers on both sides of the road, so that the vehicle sensor data can be analyzed and calculated nearby, so as to achieve faster and more reliable navigation and decision-making.

Safety requirements urge the sensor system to develop in the direction of vehicle-road coordination. Due to the limited installation position and sensing range of the sensor, as well as roadside obstacles and the size of the surrounding vehicles, it is impossible to fully and accurately perceive the road conditions only by using on-board sensors. on the other hand, the installation of sensing sensors on both sides of the road can provide a "God perspective" for self-driving vehicles, thus complementing the information of on-board sensors. The development of 5G technology also provides the basic communication technology guarantee for the rapid cooperation of vehicle road sensors. The intensive deployment of sensing sensors on both sides of the road also reduces the cost pressure of on-board sensors to a certain extent, which is more conducive to the later market promotion of self-driving cars.

Data security requirements promote the gradual integration of sensor and block chain technology. Self-driving cars are essentially Internet of things devices that use computers and sensors and are vulnerable to hackers. Through the core technologies such as distributed ledger, asymmetric encryption, consensus mechanism and intelligent contract, blockchain establishes a data storage system with decentralization, openness, independence, security and anonymity, which not only has the advantage of data traceability, but also ensures that node data can not be tampered with, which can effectively help vehicle sensor system to avoid DDoS attacks. However, the use of block chain technology also puts forward higher requirements for the data storage capacity of the sensor module.

The possibility of single point failure urges the intelligent network-connected vehicle to adopt sensor redundancy design. The failure of a single sensor in a specific environment will lead to the lack of perception of a certain environmental information, which will lead to wrong or conservative decisions. In order to prevent the loss of some key information that may exist on the way of driving, the intelligent network connected vehicle will adopt redundant sensor system. The working principle and technical characteristics of different sensors are different, which are suitable for different application scenarios. The redundancy scheme of the combination of multiple types of sensors can be used to ensure driving safety. For example, at the external perception level, the multi-sensor redundancy scheme of ultrasonic radar + millimeter wave radar + stereo camera + lidar can be used to ensure accurate perception of the external environment in any situation. In the positioning module, the redundant design of the combination of satellite positioning intelligent sensor and radar is adopted to compare the absolute positioning information based on satellite signals with the relative positioning information based on road features to confirm the position of the car in the current road, thus the positioning accuracy of centimeter level can be achieved.

The improvement of the technical demand of self-driving promotes the rapid upgrading of sensor technology. With the continuous iteration of new products, the market scale of vehicle sensors will expand rapidly. In the future, vehicle sensors will usher in grea