Autonomous driving is a product that deeply integrates the car industry with new generation information technologies such as artificial intelligence, the Internet of Things, and high-performance computing. It includes Google, Baidu, Alibaba, Toyota, and Ford. All of the world’s leading internet and giant manufacturers are actively driving in layout development. It was the primary direction of intelligent and networked development in the global car and transportation travel field at that time, and has become a strategic commanding point for countries to compete for, The central skill of active control has also developed at a rapid pace in recent years.
5G Skills
Cars are an important scenario for 5G usage. The editor of the circuit board of the car communication module has learned that in the 4G era, our understanding of the Internet of Vehicles was vehicle entertainment systems, while in the 5G era, the Internet of Vehicles meant V2X (Vehicle to X, Vehicle to Everything). When vehicles are interconnected with the entire transportation system (including but not limited to other traffic participants, traffic lights, and road information), each vehicle can obtain real-time speed and steering information of surrounding vehicles, and then avoid incidents; Each vehicle can also obtain real-time road information, and the traffic management system can adjust traffic signals based on real-time road conditions, thereby significantly reducing congestion.
5G skills will greatly assist in autonomous driving. Currently, the mainstream active driving skills rely entirely on the perception ability of the vehicle itself. It is necessary to carry a series of sensors such as laser radar worth hundreds of thousands of yuan on the vehicle, 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.
After obtaining the assistance of 5G skills, in many cases, vehicles no longer need to actively perceive other vehicles, because the information of the other party has already been transmitted to your vehicle through the network, and you and your vehicle do not need to see it to know its existence. In this “divine perspective”, the importance of sensors has greatly decreased, but active control is therefore simpler, cheaper, more reliable, and safer.
Identification skills
Recognition skills primarily include three aspects: road surfaces, static objects, and dynamic objects. Regarding a dynamic object, it is not only necessary to detect but also to track its orbit, and based on the tracking results, predict the next orbit (orientation) of the object. This is a must in urban areas, especially in China. The most typical scene is Beijing’s Wudaokou: If you stop when you see pedestrians, you will never be able to pass through Wudaokou, and pedestrians almost never stop walking through the car. Human drivers will roughly evaluate the direction of the pedestrian’s next step based on their moving trajectory, and then calculate a safe space (route planning) based on their vehicle speed. Bus drivers are best at this route. Unmanned cars must be able to do the same. Note that this is the tracking and guessing of the orbits 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 skill for autonomous cars.
V2X Skills
Vehicle to Everything (V2X) is a new generation of information communication skills that connect vehicles with everything. V represents vehicles, and X represents any object that interacts with vehicles. At that time, X primarily included vehicles, people, roadside infrastructure, and networks. Circuit board manufacturers have learned that the information mode of V2X interaction includes: vehicle to vehicle (V2V), vehicle to infrastructure (V2I), vehicle to people (V2P), and vehicle to network (V2N) interactions. V2V technology allows vehicles to prevent incidents by forwarding real-time information about themselves and ahead, and then reduce driving time, ultimately achieving the intention of improving the traffic environment and reducing traffic congestion.
V2I skills assist vehicles and roadside traffic facilities in completing data exchange through wireless means. The primary uses include intersection safety management, vehicle speed limit control, electronic toll collection, transportation safety management, as well as road construction and height limit warning. This skill will promote the intellectualization of transportation facilities, including stop traffic lights, weather information systems, and other transportation facilities. These facilities can be evolved into intelligent transportation facilities that can identify high-risk situations through multiple algorithms and actively adopt warning methods.
The V2X field is now divided into two normative and industrial camps, Dsrc and C-V2X. In the domestic market, due to the world’s largest 4G LTE network and mature industrial chain, and the lack of much accumulation of Dsrc skills, some analysts believe that the development of domestic V2X will tilt towards C-V2X.
Human computer interaction skills
Human computer interaction skills, especially voice manipulation, gesture recognition, and touch screen skills, will be widely adopted in the global future car market. The ultimate intention of planning for actively driving a large human-computer interaction screen on a car is to provide a good user experience, enhance the user’s joy in driving or operating experience during driving. It pays more attention to the safety of driving, making it necessary to balance good user experience and safety in human-computer interface planning, and to a large extent, safety always comes first. The human-computer interface for actively driving a car should integrate multiple functions such as vehicle control, function setting, infotainment, navigation system, car phone, etc., to facilitate the driver to quickly query, set, and switch various information about the vehicle system, and then enable the vehicle to achieve its desired operation and operation. In the future, vehicle information display systems and smartphones will seamlessly connect, and the input methods provided by the human-computer interface will have multiple choices. After using different skills, customers can freely switch based on different operations and functions.
High-precision map
High precision maps possess accurate vehicle location information and rich route element data information, which can help cars predict complex road surface information, such as slope, curvature, heading, and so on. Compared with traditional methods, it has higher real-time performance. Due to frequent changes in the road surface, such as road renovations, worn or repainted identification lines, and changes in traffic signs, these changes must be reflected in a timely manner on a high-quality map. High precision maps will place more emphasis on the three-dimensional model and accuracy of space, reducing the accuracy from the meter level to the centimeter level, and it is necessary to accurately display every feature and situation on the road surface.
Resolution Planning Skills
Resolution planning is a central skill for autonomous driving, which is equivalent to actively driving the brain of a car. It comprehensively analyzes the information provided by the environmental awareness system, and routes and addresses the results from a high-precision map to plan the vehicle at that time (speed planning, orientation planning, acceleration planning, etc.), and generates corresponding resolution plans (car following, lane changing, parking, etc.). PCB factories believe that planning skills also need to consider the mechanical, dynamic, and kinematic characteristics of vehicles. Common decision planning skills include expert manipulation, hidden Markov models, Bayesian networks, and fuzzy logic.
Positioning skills
In addition to GPS and inertial sensors, we often use LiDAR point clouds to match high-precision maps, as well as visual mileage calculation methods, and other positioning methods to correct each other to achieve more accurate results. With the development of active driving, positioning skills will definitely continue to be optimized.
Nowadays, the skills of active driving are basically derived from robots. Active driving a car can be seen as a wheeled robot plus a comfortable sofa. Location and path planning are a problem in robotic systems. Without location, it is impossible to plan paths. Centimeter level real-time positioning is now one of the biggest challenges in active control. For robot systems, positioning primarily relies on the interpolation and comparison of SLAM and Prior Map.
Communication security skills
While actively driving cars through the vehicle network to access the grid, there are also issues of information security. In use, the information of each car and its owner will be transmitted to the network anytime and anywhere to be perceived. This information exposed in the network can easily be stolen, disturbed, or even modified, and then directly affect the security of the intelligent connected car system. Therefore, in intelligent connected cars, Attention must be paid to the discussion of information security and privacy protection skills.