5G takes the lead: An inventory of core technologies for autonomous driving

Autonomous driving is the product of the deep integration of the automotive industry with new-generation information technologies such as artificial intelligence, the Internet of Things, and high-performance computing. All of the world's top Internet and giant manufacturers, including Google, Baidu, Alibaba, Toyota, and Ford, are planning to develop autonomous driving. It is the main direction of the current development of intelligence and networking in the global automotive and transportation fields, and has become a strategic commanding height that countries are vying for. The core technology of autonomous driving has also been developing at a rapid pace in recent years.

5G Technology

Cars are an important scenario for 5G applications. In the 4G era, we understand the Internet of Vehicles as an in-car entertainment system, while in the 5G era, the Internet of Vehicles means V2X (Vehicle to X). When vehicles are connected to the entire traffic system (including but not limited to other traffic participants, traffic lights, and road information), each vehicle can obtain the speed and steering information of surrounding vehicles in real time, thereby avoiding accidents; each vehicle can also obtain real-time road condition information, and the traffic management system can adjust traffic signals according to real-time road conditions, thereby greatly reducing congestion.

5G technology will greatly help autonomous driving. The current mainstream autonomous driving technology route completely relies on the vehicle's own perception ability. The vehicle must be equipped with a series of sensors such as laser radar worth hundreds of thousands of yuan, but the detection distance and accuracy still need to be improved. At the same time, blind spots and the unpredictability of other vehicles mean the existence of risks.

With the help of 5G technology, in many cases, vehicles no longer need to actively sense other vehicles, because the other party's information has already been transmitted to your car through the network. You and your car don't need to see it to know its existence. Under such a "God's perspective", the importance of sensors is greatly reduced, but autonomous driving is simpler, cheaper, more reliable and safer.

Identification Technology

Recognition technology mainly includes three aspects: road surface, static objects and dynamic objects. For dynamic objects, it is necessary not only to detect but also to track their trajectories, and based on the tracking results, predict the next trajectory (position) of the object. This is essential in urban areas, especially in China. The most typical scene is Wudaokou in Beijing: if you stop when you see pedestrians, you will never be able to pass Wudaokou, as pedestrians almost never stop walking in front of the car. Human drivers will roughly assess the next position of pedestrians based on their moving trajectories, and then calculate the safe space (path planning) based on the speed of the vehicle. Bus drivers are best at this. Autonomous vehicles must also be able to do this. It should be noted that this is the tracking and prediction of the trajectories of multiple moving objects, which is much more difficult than a single object. This is MODAT (Moving Object Detection and Tracking). It is also the most difficult technology for autonomous vehicles.

V2X Technology

Vehicle to Everything (V2X) is a new generation of information and communication technology that connects vehicles to everything, where V stands for vehicle and X stands for any object that interacts with the vehicle. Currently, X mainly includes vehicles, people, roadside infrastructure and networks. The information modes of V2X interaction include: interaction between vehicles (Vehicle to Vehicle, V2V), between vehicles and roads (Vehicle to Infrastructure, V2I), between vehicles and pedestrians (Vehicle to Pedestrian, V2P), and between vehicles and networks (Vehicle to Network, V2N). V2V technology allows vehicles to prevent accidents by forwarding real-time information about themselves and the vehicles ahead, thereby reducing driving time and ultimately achieving the goal of improving the traffic environment and reducing traffic congestion.

V2I technology helps vehicles and roadside traffic facilities exchange data wirelessly. Its main applications include intersection safety management, vehicle speed limit control, electronic toll collection, transportation safety management, and road construction and height limit warnings. This technology will promote the intelligence of traffic facilities, including no-entry lights, weather information systems and other traffic facilities, which can evolve into intelligent traffic facilities that can identify high-risk situations through multiple algorithms and automatically take warning measures.

At present, the V2X field is divided into two standards and industry camps: Dsrc and C-V2X. In the domestic market, since it has the world's largest 4G LTE network and a mature industry chain, and there is not much accumulation in Dsrc technology, some analysts believe that the development of domestic V2X will lean towards C-V2X.

Human-computer interaction technology

Human-computer interaction technology, especially voice control, gesture recognition and touch screen technology, will be widely adopted in the global future automobile market. The ultimate goal of the design of the large human-computer interaction screen of the self-driving car is to provide a good user experience, enhance the user's driving pleasure or the operating experience during driving. It pays more attention to driving safety, so the design of the human-computer interface must strike a balance between good user experience and safety. Safety is always the first priority. The human-computer interface of the self-driving car should integrate multiple functions such as vehicle control, function setting, infotainment, navigation system, car phone, etc., so that the driver can quickly query, set and switch various information of the vehicle system, so as to achieve the ideal operation and operation state of the vehicle. In the future, the in-vehicle information display system and the smart phone will be seamlessly connected, and the input method provided by the human-computer interface will have multiple options. By using different technologies, consumers can switch freely according to different operations and different functions.

High-precision maps

High-precision maps have accurate vehicle location information and rich road element data information, which can help cars predict complex road information, such as slope, curvature, heading, etc. Compared with traditional maps, they have higher real-time performance. Since road surfaces often change, such as road repairs, wear or repainting of marking lines, changes in traffic signs, etc., these changes must be reflected in high-precision maps in a timely manner. High-precision maps will place more emphasis on the three-dimensional model and accuracy of space, reducing the accuracy from meters to centimeters, and must display every feature and condition on the road very accurately.

Decision-making technology

Decision-making is the core technology of autonomous driving, which is equivalent to the brain of autonomous vehicles. It plans the current vehicle (speed planning, direction planning, acceleration planning, etc.) by comprehensively analyzing the information provided by the environmental perception system and the results of high-precision map routing addressing, and makes corresponding decisions (following the vehicle, changing lanes, parking, etc.). Planning technology also needs to consider the mechanical characteristics, dynamic characteristics, and kinematic characteristics of the vehicle. Common decision-making technologies include expert control, hidden Markov model, Bayesian network, fuzzy logic, etc.

Positioning Technology

In addition to GPS and inertial sensors, we usually use LiDAR point cloud and high-precision map matching, as well as visual mileage calculation and other positioning methods, so that various positioning methods can correct each other to achieve more accurate results. With the development of autonomous driving, positioning technology will continue to be optimized.

At present, the technology of autonomous driving is basically derived from robots. Autonomous driving cars can be regarded as wheeled robots plus a comfortable sofa. Positioning and path planning are a problem in robot systems. Without positioning, it is impossible to plan the path. Centimeter-level real-time positioning is one of the biggest challenges of autonomous driving. For robot systems, positioning mainly depends on the cross-comparison between SLAM and prior maps.

Communication security technology

While self-driving cars are connected to the grid through the vehicle network, they also bring about information security issues. In the application, the information of each car and its owner will be transmitted to the network and sensed anytime and anywhere. This information exposed on the network can be easily stolen, interfered with or even modified, which directly affects the security of the intelligent connected vehicle system. Therefore, in intelligent connected vehicles, it is necessary to pay attention to the research of information security and privacy protection technology.