During the 2024 Spring Festival, the topic of the most complicated Spring Festival weather since 2008 in China topped the hot search, with a reading volume of up to 3600 million times. When the "busiest Spring Festival" encounters extremely complicated weather, it affects the hearts of everyone on the way home.
Previously, according to **, self-driving travel in the Spring Festival in 2024 will hit a record high, and the surge in traffic will put great pressure on urban traffic.
Behind the smooth journey is not the blessing of luck, but the comprehensive embodiment of the intelligent cloud based on large model technology and intelligent signal control scheme to comprehensively improve the operation efficiency of urban transportation.
Bloom more and travel is very different
Intelligent transportation solutions supported by large-scale model technology are improving the efficiency of Spring Festival transportation in 2024.
As one of the main places for the Spring Festival population, Changchun City has suffered great traffic pressure. In the past, due to the high-frequency travel during the holiday and the influx of vehicles from other places, the traffic pressure on the eastern expressway in Changchun urban area was huge during the Spring Festival.
In order to ensure the safe and smooth flow of urban road traffic, Changchun City has made every effort to deploy and carry out traffic congestion control.
In the process of congestion control, the Changchun traffic police introduced intelligent information control technology, combined with AI traffic monitoring data at the entrance of the expressway, the core congestion point of the city, combined with AI traffic monitoring data, to achieve the linkage control of expressway and ramp, expressway and ground; Realize the temporary control of lanes at the merging point to relieve the traffic pressure with intelligent management.
At the same time, Changchun has continuously deepened the reform of the public travel structure, vigorously developed public transportation, continuously improved the share rate of public transportation, combined with the optimization of traffic engineering and other means, and now the problem of urban traffic congestion in Changchun has been significantly improved. Map data shows that compared with before the treatment, the overall congestion index of expressways in Changchun fell by 20 during weekday peak hours in January9%, and the average speed is increased by 213%, and the congestion mileage decreased by 573%, during the Spring Festival, a large number of returning vehicles were smoothly undertaken.
For more than 70 ramp crossings of expressways, through the fusion of roadside laser data and Internet floating vehicle data, single-ramp control, high-altitude joint control, high-altitude joint control, and induction linkage multi-scene optimization control are realized, and the system has been put into normal operation and use. Map data shows that compared with before the treatment, the overall congestion index of expressways during peak hours in February decreased by 35%, and the average speed increased by 45%.
Data**: February 16-17, map.
During the peak period of the Spring Festival return this year, the congestion index of Changchun ring road and expressway was 119, which is 4. lower than the return peak of the Spring Festival in 20231%。In addition to the traffic pressure in some core business districts affected by the Spring Festival, the overall road traffic operation in Changchun City is improving.
As the only state-level economic and technological development zone in Beijing, Yizhuang also suffered from greater pressure on the road network during the Spring Festival.
At present, Yizhuang has deployed global signal control and perception equipment in 60 square kilometers of the main urban area, dynamically adjusting the duration of green lights at intersections, effectively eliminating the traffic pressure caused by the tidal traffic flow during the Spring Festival. The data shows that after the deployment of AI global signal control, the average delay rate of vehicles at Yizhuang intersection decreased by 2848%, and the length of the vehicle queue decreased by 303%, green light wasted time dropped 1833%。
Data**: February 16-17, map.
During the peak of the Spring Festival return this year, the overall congestion index in Yizhuang area was 111. The overall road network traffic is running very smoothly, and it has kept "a piece of green" during the return peak of the Spring Festival.
It is understoodIn actual operation, the AI information control system relies on the traffic model to determine the possibility of congestion, and superimpose the best deduction to generate a global signal control scheme, from the optimal single-point trunk line to the regional optimum. At the same time, through navigation maps, traffic guidance screens and other channels, the congestion reminders of major roads, traffic light information, and detour route suggestions are pushed to each driver to improve the driving experience and reduce traffic safety accidents.
The above-mentioned capabilities are also used in the traffic congestion mitigation work in Guiyang City.
Guiyang City is located in the central part of Guizhou Province, located in the Yunnan-Guizhou Plateau, with mountainous areas, hills, depressions dominated by landform characteristics, road traffic conditions are relatively weak, compared with plain cities, traffic congestion is very prone to occur, and congestion recovery time is relatively slow. In recent years, Guiyang's economic development momentum has been rapid, and its GDP growth rate has ranked among the top provincial capitals in the country for six consecutive years. In order to balance the traffic contradictions brought about by the rapid development of the city, since 2021, Guiyang City has been determined to fight a tough battle of traffic congestion with an iron hand.
In the process of congestion control, all departments at all levels in the city work together and make concerted efforts, and make full use of big data means, combined with intelligent cloud, through the monitoring means of big data "Internet traffic", refine the perception of traffic operation at different levels of intersections, roads and regions, analyze the spatial and temporal distribution of traffic travel in the city, and help relevant departments to support the decision-making of traffic congestion control, and have achieved good application results, which are embodied in the following three aspects.
First, the AI signal control system. The intelligent cloud collides with the Internet traffic big data through the data obtained by the IoT sensing device, which can obtain a multi-dimensional and multi-faceted distribution of traffic travel in the city, and uses AI algorithms to summarize travel rules and future traffic trends.
At the same time, the system is linked with the timing data of the intersection signal light to realize the visual command and control of the "light state map". Since the application of the system, in addition to the automatic optimization of the timing of the single-point adaptive signal at the intersection, it can also realize the adaptive coordination of the green wave of the trunk line.
By the end of 2023, the Guiyang traffic police have carried out single-point adaptive optimization of signal timing at more than 30 intersections such as Shubo Avenue, Jinyang Road, and Qianlingshan Road in Guanshanhu District, Guiyang, and realized the timing coordination of multiple trunk lines, which was combined into a regional-level signal control optimization. In the future, it is planned to expand the scope of regional information control to the whole area of Guiyang City, helping to further improve the level of urban traffic control in Guiyang.
Second, share data to facilitate services. The traffic police in Guiyang share traffic light data with the intelligent cloud, and use the map navigation application as the carrier to provide traffic light countdown, light change prompt warning, green wave speed suggestions and other information to car owners in Guiyang City, so as to facilitate the travel of citizens.
Third, multi-dimensional data evaluation is adopted. Big data also provides multi-dimensional evaluation parameters for the Guiyang traffic police, according to the map traffic big data monitoring results, since the start of the congestion control action, in 2022, the average congestion index of Guiyang City on weekdays throughout the day is 1296, a year-on-year decrease in the congestion index in 2021, a decrease of 643%；In 2022, the average congestion index of Guiyang City's commuting peak is 1607, compared with the average congestion index of the commuting peak in 2021 of 1935, a decrease of 1695%。
After continuous traffic optimization and management, in the national 100-city commuting peak traffic congestion ranking, Guiyang's congestion ranking has dropped from 4th in 2021 to 20th in 2023, and the results are very significant. In February 2024, the average vehicle delay in the central area of Guanshanhu District decreased by 13%, and the congestion index decreased by 62%。
Data**: February 16-17, map.
During the peak period of the return trip during the Spring Festival this year, the overall traffic congestion index of Guiyang City was 113. The overall road traffic operation in the city is good to ensure that the citizens return all the way to the road.
The smooth travel of Changchun, Yizhuang, and Guiyang is a microcosm of AI governance and congestion alleviation. It is understood that the global information control scheme has also been implemented in Zhuzhou, Hunan, Baoding, Hebei and other cities and is in continuous operation.
Since 2019, the Baoding traffic police have joined forces with the intelligent cloud to upgrade the intelligent signal control at more than 170 intersections, and the peak congestion index in the urban area has decreased by 46%, the average speed increased by 116%, and the trunk travel time is reduced by 20%.
Data**: February 16-17, map.
During the peak period of the return trip during the Spring Festival this year, the overall traffic congestion index of Baoding City was 118. In addition to the Baoding Zhili Governor's Office Museum, the ancient lotus pond and other key scenic spots, the overall road traffic operation in Baoding urban area is relatively smooth.
As an important transportation hub in Hunan Province, the local traffic management department in Zhuzhou and the intelligent cloud have realized the regional-level signal control optimization of all working conditions at 51 intersections, reducing the regional congestion index by 97%, and the average speed increased by 106%, the trunk congestion index fell by 134%, and the efficiency of police handling has increased by about 30%.
Data**: February 16-17, map.
The successful implementation of many projects has made the practical experience of intelligent cloud in urban transportation more abundant. At the same time, for the completed projects, the intelligent cloud is also continuously optimized and improved to ensure the follow-up operation effect of urban transportation.
Traffic model + global signal control
Help alleviate urban traffic congestion and ensure smooth flow
As one of the few high-tech enterprises in the industry with the full-stack closed-loop capability of "vehicle road cloud map", smart cars, smart roads, smart clouds and vivid maps have always been the four trump cards in hand.
In 2023, we will continue to make efforts in the direction of large models. In March, the official release of "Wenxin Yiyan". Soon after, the "Global Information Control and Congestion Mitigation Solution Based on Traffic Large Model" was released.
With the powerful ability of large models to compress knowledge, in September 2023, the intelligent cloud upgrade released the "ACE Intelligent Transportation Engine 3."0”，The basic capabilities of vehicles, roads, clouds, and maps are aggregated and generalized and upgraded, and the reconstruction of subdivided application scenarios such as traffic management brain, global information control, and smart highway based on large models is further realized, and efforts are made to improve the capabilities of traffic perception, traffic cognition, traffic prediction, traffic knowledge, and traffic interaction.
The ACE intelligent transportation engine has gone through three stages of development, starting from 1Phase 0 clarifies the framework of "vehicle-road-cloud-map" intelligent transportation basic capacity building, to 2Phase 0 focuses on the establishment of connections and iterations in the general direction of vehicle-road-cloud integration, and releases solutions for subdivided scenarios such as intelligent networking, smart traffic management, smart highways, and smart parking, and then to the intelligent transportation engine 3 with the transportation industry model as the core0 architecture,The application of transportation + large model in the field of intelligent transportation segmentation has been further stimulated and strengthened.
Compared with before, the AI signal control under the blessing of the large model has been further improved in the optimization scope, optimization algorithm, and optimization efficiency, based on the iterative algorithm effect and continuous optimization algorithm, it can further solve the scenario-based traffic problem and a series of long-tail problems, and continuously narrow the effect difference between the first environment and the real environment, and can also provide efficient man-machine dialogue answers, and intelligently generate detailed and accurate traffic analysis reports, which has brought more surprises to users in the interaction ability of the previous signal control.
With the deepening of technical research, in the future, more capabilities of the large transportation model will continue to be stimulated to empower more scenarios of traffic management.
General technology will go further
Applications in the transportation industry
Recently, OpenAI's Wensheng **large model SORA can generate up to 60 seconds of ** according to the text description, which not only maintains good visual quality, but also completely and accurately restores the user's prompt needs, and the strong physical world understanding and simulation capabilities hidden behind it have caused a series of shocks and exclamations in the science and technology circle.
Advances in general-purpose technology have made people look forward to the future of verticals. Perhaps in the near future, being able to drive home unhindered during peak periods such as the Spring Festival can become a reality.
Technology is constantly advancing. From the very beginning of Internet technology, to big data, cloud computing, artificial intelligence, and then to today's large models, a group of Internet companies represented by them have been striving to explore and apply advanced general technologies in the transportation industry in the past many years, changing travel and life.
Vice President Shi Qinghua said: In the future, the large model has a broader prospect in the field of transportation, quickly embraces the large model, quickly arms the research and development capabilities, application transformation capabilities, and landing capabilities in the large model, and uses the large model products to help customers solve the problem of congestion on the road pragmatically and landing.
The state of the road flow is dynamically changing, so signal optimization is an ongoing effort. At present, cities such as Tongzhou, Shijiazhuang, Haikou, Luoyang, Tai'an, Mianyang, and Hejian in Beijing are also further exploring and carrying out pilot demonstrations of AI information control and congestion mitigation. In the future, we will continue to explore more technologies, including large traffic models, in the field of transportation to bring more surprises to urban life.
Editor丨Guo Qi.
Audit丨Lian Meng.
Submission email: xww@zgjtbcomBusiness: 010-64266095