news

As science and technology develop rapidly, the concept of urban development is evolving from smart city to intelligent city. However, these two concepts are not simply progressive, but have significant differences in connotation and implementation methods.

Smart cities focus on using information technology to digitally transform all aspects of a city, such as presenting and managing the city's infrastructure and public services in a digital form. Smart cities will obtain information in real time through AI big model capabilities and other advanced information technologies, making physical systems "intelligent."

In the field of transportation, current digitalization efforts are mostly just to transform physical transportation forms into digital forms, but have failed to effectively solve the two key issues of traffic efficiency and safety. Traffic congestion still occurs frequently, and the shadow of traffic accidents has never gone away.

Until the large-scale development of vehicle-road-cloud integration and the announcement of the list of pilot application cities, we have a new dawn. The construction of the vehicle-road-cloud network has truly realized the organic connection between cloud (central cloud, regional cloud), edge (edge ​​computing system) and end (intelligent connected vehicles, intelligent roadside infrastructure), providing a new idea and method for solving traffic problems.

In this architecture, advanced sensing technology is used to obtain real-time information on vehicle and road status; powerful prediction capabilities are used to predict changes in traffic flow and possible problems in advance; and based on precise decision-making mechanisms, traffic signal control is optimized and vehicle routes are adjusted. Combined with the AI ​​big model, the entire process of data collection, labeling, modeling, training, testing, generation, and cloud application is intelligent. The central cloud is like an intelligent brain, controlling the operation status of the entire city and traffic, realizing a major change in the city's intelligent productivity, as if opening up Moore's Law for smart cities.

In the past, urban management and infrastructure upgrades were often periodic and large-scale investments, but the results were far from satisfactory, and traffic efficiency improvements have not reached the ideal level. Today, the maturity of vehicle-road-cloud integration technology has completely changed this situation, opening a new chapter for the development of smart cities.

In this process, intelligent road infrastructure is certainly important, but there is another key role that cannot be ignored, that is the roadside edge computing system (MRS) independently developed by Mushroom Car Union. At present, MRS plays an important role in connecting people, cars, and non-motor vehicles, laying a solid foundation for achieving a more efficient and safer transportation environment.

As a leading artificial intelligence company, Mushroom Autolink has always been user-centric and committed to building an artificial intelligence network for transportation and automobiles. A set of mature technical solutions that can be implemented, replicated and promoted has been formed around the integration of vehicles, roads and clouds. In 2023, Mushroom Autolink released the latest MOGO Package, covering a full range of products for vehicles, roads and clouds: on the road side, the AI ​​digital road base station MOGO AI Station can quickly realize real-time digitalization of roads; on the vehicle side, Mushroom Autolink creates standardized autonomous driving vehicles of multiple categories and models, and all models are equipped with vehicle-road collaborative V2X functions as standard; on the cloud side, the AI ​​smart transportation cloud platform can fully empower L0-L4 level intelligent connected vehicles, improve traffic safety and efficiency, optimize traffic management, and quickly realize road digitalization in highway scenarios, urban scenarios, and scenic park scenarios.

Recently, after testing and reporting by China Automotive (Beijing) Intelligent Connected Vehicle Research Institute Co., Ltd. (hereinafter referred to as "China Automotive Intelligent Connected"), Mushroom Autolink's AI digital road base station and its system (MRS) met the highest requirements of the "Test Methods and Standards for Vehicle-Road Collaborative Roadside Perception Systems (Draft for Approval)" of the China Academy of Information and Communications Technology. The test results show that Mushroom Autolink's AI digital road base station can achieve perception coverage, system perception time accuracy, and perception latency that meet and exceed the SL3 standard. At the same time, the positioning accuracy, heading angle, and size detection accuracy have reached the industry-leading level. Through years of technology development and experience accumulation, Mushroom Autolink has formed a complete solution for vehicle-road-cloud integration that can be implemented, replicated, and promoted. It combines massive traffic big data from the three ends of vehicle, road, and cloud to build an AI big model, realizing autonomous driving from perception to cognition, collaborative decision-making, roadside data empowerment simulation, and model training, supporting the large-scale implementation of autonomous driving that is smarter and safer.

Currently, it has been implemented in open roads, highways, parks, scenic spots, airports and other scenarios in many cities including Beijing, Shenzhen, Hunan, Yunnan, Hubei, Sichuan, Liaoning, Shandong, and Tianjin.

Its independently developed MRS system, as the core computing part of the roadside infrastructure in the vehicle-road-cloud integrated system, is a roadside computing system dedicated to serving the vehicle-road-cloud integrated business. Based on the original perception data or structured data of the roadside perception equipment, through high-performance real-time data processing, perception fusion, calculation, and analysis, it can provide real-time dynamic traffic perception data for the cloud control basic platform and networked vehicles.

It has high-precision fusion perception and computing capabilities, and all indicators meet the SL3 standard of the "Vehicle-Road Collaborative Roadside Perception System" of the China Academy of Information and Communications Technology. At the same time, it also has the ability to reuse old equipment and supports system upgrades based on existing roadside computing equipment, greatly reducing construction costs and resource waste. In addition, the rich application scenarios enable it to fully cover different areas such as urban open roads, highways, and parks, and adapt to various complex traffic environments. Moreover, it is fully adapted to the vehicle-road-cloud integrated system, and has the road-cloud communication standard docking capability of the cloud control basic platform, ensuring seamless transmission and efficient processing of data.

As the intelligent core of Mushroom Car-to-Road-Cloud strategy, the MRS system is innovative not only in the integration and optimization of hardware, but also in the in-depth application and innovation of AI algorithms:

High-precision perception model:The perception model that integrates deep learning algorithms can accurately identify all types of traffic participants, including pedestrians, non-motor vehicles and various types of vehicles, and can maintain high accuracy even in complex environments and extreme weather conditions.

Real-time prediction algorithm:The prediction model that integrates spatiotemporal series analysis and deep reinforcement learning can not only accurately predict the changing trend of traffic flow, but also provide real-time early warning for traffic events (such as accidents and congestion), providing a basis for preventive management.

Decision optimization algorithm:Using advanced operations optimization and machine learning algorithms, traffic signal control, road network resource allocation, etc. are dynamically optimized to ensure maximum traffic flow while reducing carbon emissions and energy consumption.

From the perspective of product architecture, the MRS system includes five modules: Zhidao OS, roadside edge fusion perception engine, roadside edge computing engine, supporting data, and data sharing interface. Each module has important functions.

Zhidao OS, as the underlying capability of MRS, includes the kernel layer, hardware abstraction layer, and data access standard interface service. The kernel layer is the most basic functional layer for interacting with the underlying hardware, supporting multi-tasking and multi-process, memory management, hardware driver support, and security functions. The hardware abstraction layer implements a unified hardware interface for roadside perception devices, and can uniformly process structured data output by devices including cameras, millimeter-wave radars, and lidars. The data access standard interface service covers the real-time operation status and perception data collection of vehicles, roadside perception equipment data collection, and traffic facility data collection, providing a rich data source for subsequent processing and analysis.

The roadside edge fusion perception engine can perform algorithm recognition, fusion analysis and processing on the original perception data or structured data of the roadside perception facilities to obtain high-precision perception result information. It can realize the recognition, detection and positioning functions of traffic participants such as motor vehicles, non-motor vehicles, and pedestrians, and output characteristic information including traffic participant type, speed, location, movement direction, etc.

The roadside edge computing engine is one of the cores of the MRS system. It can receive data from roadside sensing devices and edge fusion sensing engines in real time, and perform high-speed processing and analysis to obtain high-precision event, traffic flow and other data, thereby supporting the decision-making and control of the intelligent transportation system.

In terms of traffic event detection, it can realize automatic detection, alarm reminder and process recording of traffic congestion, traffic accidents, abnormal parking, reverse driving, illegal lane change, pedestrians running red lights, spilled objects, etc. For example, when a reverse driving event is detected, it can send an early warning to the main vehicle in time; when a construction road occupation event is found, it can remind the vehicle to avoid traffic jams and accidents.

In terms of traffic flow detection, it can accurately detect information such as traffic flow, average vehicle speed, time occupancy, queue length, etc., and support traffic flow information statistics by lane. These data are of great significance for optimizing traffic signal control and rationally allocating road resources.

In addition, the roadside edge computing engine also has a global video monitoring function, supports multiple video encoding standards, can view video images in real time, and keep track of road conditions in a timely manner. At the same time, it can obtain the status of traffic lights and provide accurate traffic information for vehicles. Moreover, through real-time hardware monitoring, it can monitor the status of hardware in real time, including performance, temperature, voltage and other information, to ensure the stable operation of the system.

The supporting data mainly comes from traffic facilities and other infrastructure, including infrastructure status monitoring data, traffic light data and traffic weather environment monitoring data. These data provide the basis for auxiliary decision-making for the calculation engine, enabling it to output more accurate calculation results. The data sharing interface is used to ensure the standardized interaction of data between MRS and intelligent roadside infrastructure and cloud control basic platform, and provide data standardization conversion and service output in a universal interface manner, supporting applications such as remote driving, assisted driving and safety warning.

The importance of the MRS system is not only reflected in its powerful functions and accurate data analysis, but also in its significant improvement in traffic dispatch efficiency, similar to the dispatch systems of high-speed rail and civil aviation. High-speed rail uses precise dispatching to allow trains to run efficiently on complex routes and arrive at every station on time; civil aviation uses advanced air traffic management systems to ensure the orderly takeoff and landing of aircraft in the air and at airports to avoid flight delays. The MRS system in the vehicle-road-cloud network is like the dispatching core of high-speed rail and civil aviation, which can finely manage traffic flow in the city and maximize the use of road resources.

With the continuous advancement of technology, the development prospects of the MRS system are extremely broad. It will not only be limited to the current functions, but will also connect more intelligent devices, such as smart street lights, smart sanitation vehicles, smart logistics vehicles, drones, robots, etc., to jointly build a more complete smart city ecosystem. For example, smart street lights can automatically adjust the brightness and lighting range according to real-time traffic flow and ambient light conditions, and can also work with the MRS system to monitor road conditions and feedback information; smart sanitation vehicles can efficiently plan cleaning routes under the guidance of the MRS system, adjust the cleaning frequency and intensity according to the conditions of different areas, and transmit garbage collection conditions in real time; smart logistics vehicles combined with the MRS system can achieve more accurate cargo distribution, plan the optimal transportation route, and monitor the status of cargo and vehicles; drones can perform air patrol missions under the control of the MRS system, scout road conditions and send back images, and can also deliver emergency supplies in special circumstances; robots can also provide better service guidance in public places through the MRS system, or cooperate in factories and warehouses to achieve efficient cargo handling and storage.

Around the world, different technology companies are exploring similar technical solutions. In the future, when Musk's Starlink plans to connect Tesla, Optimus Prime robots, drones and other equipment, the "Chinese solution" of vehicle-road-cloud integration has obvious advantages.

The integrated vehicle-road-cloud network is superior to Starlink in terms of roadside latency, perception accuracy and security. The MRS system is deployed at the edge of the road and can transmit and process data at close range, reduce latency, and ensure timely and accurate traffic decisions. Starlink relies on satellite communications and is susceptible to interference and delay in urban environments, which affects the application effect. Through the collaborative work of multiple roadside sensing devices, it can perceive detailed information with high precision and provide accurate data support for traffic management. Starlink is weak in capturing subtle changes and accurate information on the ground due to its long distance. The MRS system uses localized processing and encryption technology to ensure the security of traffic data. Starlink satellite communications face greater security risks in the face of network attacks, which may seriously affect the traffic system.

In short, the "Chinese solution" of vehicle-road-cloud integration is undoubtedly the best choice for realizing smart city transportation. The physical world connected by the vehicle-road-cloud network will also lead the upgrade of smart cities to intelligent cities.