A sensor-computer integrated control system and method for merging zones of highways

Abstract

This invention discloses an integrated sensing, sensing, and computing management system and method for merging traffic on highways. In the system, vehicle terminals and roadside terminals are connected to an edge computing unit, sending perceived data to the edge computing unit. The vehicle terminals and roadside terminals are equipped with SNN models, performing VAI and IAI tasks respectively, while the edge computing unit is equipped with a DNN model, performing EAI tasks, together forming a dynamic distributed computing network. The edge computing unit coordinates and collaborates with the vehicle terminals and roadside terminals to perform distributed iterative training of the edge federated learning model until convergence. Based on the data, the edge computing unit determines traffic operation status, vehicle control strategies, and safety warning information, and feeds this information back to the vehicle terminals and roadside terminals. This invention significantly improves data processing efficiency and decision-making accuracy, enabling intelligent prediction and management of traffic congestion, and providing drivers with a smoother and safer travel experience.

Description

Technical Field

[0001] This invention relates to the field of road traffic control system technology, and in particular to an integrated control system and method for the integrated control of communication, sensing, and computing in highway merging zones. Background Technology

[0002] As critical areas where vehicles converge, highway merging zones exhibit complex and variable traffic flows, posing a high risk of traffic accidents and easily causing traffic congestion. In traditional highway merging zone management systems, sensing, communication, and computing functions are often executed independently and in a decentralized manner. Sensing functions primarily rely on roadside radar, video, and other equipment to collect information such as traffic flow, speed, and distance. These sensing devices often provide only basic data, and their accuracy and resolution are limited by the hardware itself. Communication functions utilize communication networks (wired or wireless) to transmit sensing data to the management center or edge computing unit. This communication process can introduce latency or network congestion. Computing functions involve analyzing and processing the received data at the management center or edge computing unit to assess the traffic operation status of the merging zone and formulate corresponding control strategies.

[0003] With the development of mobile communication and artificial intelligence technologies, communication networks and AI are increasingly converging. Communication networks are gradually evolving from centralized intelligence to edge intelligence, giving rise to a new network paradigm with "edge intelligence" capabilities, aiming to provide ubiquitous real-time intelligent services. Edge intelligent networks expand application scenarios such as smart highways through the collaboration and interaction of communication, perception, and computing. Data-driven highway merging zone management scenarios not only place more demanding performance requirements on communication and computing but also demand high-precision, high-resolution perception. In terms of perception, higher-precision data on vehicle position, speed, and distance are needed, while high-resolution image and video data provide the foundation for subsequent analysis and decision-making. In terms of communication, an edge intelligent network capable of supporting random access from roadside nodes needs to be built, possessing low-latency information perception, transmission, and processing capabilities. In terms of computing, technologies such as artificial intelligence and machine learning are needed to deeply mine and analyze traffic data and formulate corresponding control strategies.

[0004] To meet the demand for high-precision, high-resolution sensing data in highway merging zones, and to achieve low-latency data transmission and efficient computational processing, an innovative control system and device are urgently needed. Simultaneously, the system also requires the construction of a stable, reliable, and low-latency edge intelligent communication network to support random access by roadside nodes and high-speed data transmission. Summary of the Invention

[0005] To address the aforementioned issues, this invention provides an integrated sensing, perception, computing, and control system and method for highway merging zones. By integrating advanced technologies in communication, perception, computing, and control, it constructs a real-time interactive and collaborative management network for vehicle, road, and traffic operation systems. Through the introduction of a dynamic distributed computing network architecture and a refined multi-level AI task classification strategy, it significantly improves data processing efficiency and decision-making accuracy. Simultaneously, by utilizing a variable speed limit model and federated learning technology, it achieves intelligent prediction and management of traffic congestion, optimizing driving safety. This transforms the highway merging zone control system from a "perception-first, communication-last-computation" model to a "distributed iteration of model parameters based on perception data" mode, providing drivers with a smoother and safer travel experience and propelling intelligent transportation systems towards greater intelligence and efficiency.

[0006] To achieve the above objectives, the present invention provides an integrated control system for the merging zone of a highway, comprising: a vehicle terminal, a roadside terminal, and an edge computing unit;

[0007] The vehicle terminal is communicatively connected to the edge computing unit via a roadside unit (RSU), and the roadside terminal is communicatively connected to the edge computing unit.

[0008] The vehicle terminal sends the vehicle perception data acquired by the vehicle perception device to the edge computing unit through the RSU, and the roadside terminal sends the roadside perception data acquired by the roadside perception device to the edge computing unit.

[0009] The vehicle terminal and the roadside terminal are equipped with SNN models, which execute VAI and IAI tasks respectively, and send the task calculation results to the edge computing unit. The edge computing unit is equipped with DNN models to execute EAI tasks and calculate the information uploaded by the vehicle terminal and the roadside terminal. The vehicle terminal, the roadside terminal and the edge computing unit jointly construct a dynamic distributed computing network.

[0010] The edge computing unit, the vehicle terminal, and the roadside terminal perform distributed machine learning based on the edge federated learning model. The edge computing unit coordinates and cooperates with the vehicle terminal and the roadside terminal to perform iterative training until the edge federated learning model converges.

[0011] The edge computing unit performs fusion calculations based on the acquired perception data to determine the traffic operation status, vehicle control strategy, and safety warning information based on the preset merging zone section of the highway, and feeds it back to the vehicle terminal and the roadside terminal.

[0012] In the above technical solution, preferably, the vehicle sensing device includes an on-board sensor, an on-board radar, and an on-board camera, and the roadside sensing device includes a roadside radar, a roadside camera, and a weather detector;

[0013] The roadside terminal also includes a variable information sign, which receives and publishes safety warning information pushed by the edge computing unit.

[0014] In the above technical solution, preferably, the roadside terminals are distributed in the upstream and downstream sections and ramp sections of the main line merging point of the preset merging area of ​​the expressway.

[0015] In the above technical solution, preferably, the vehicle terminal performs VAI tasks based on the SNN model with the support of local computing resources on the vehicle end, and processes the vehicle perception data obtained by the vehicle perception device.

[0016] With the support of local computing resources at the roadside, the roadside terminal executes IAI tasks based on the SNN model to process the roadside sensing data acquired by the roadside sensing device.

[0017] With the support of edge computing resources, the edge computing unit performs EAI tasks based on the DNN model to fuse the perception data sent from the vehicle terminal and the roadside terminal.

[0018] The SNN model and the DNN model are pre-trained offline using massive amounts of historical data.

[0019] In the above technical solution, preferably, the vehicle terminal, the roadside terminal, and the edge computing unit use OTA technology for remote model upgrades;

[0020] The vehicle terminal, the roadside terminal, and the edge computing unit perform distributed machine learning based on an edge federated learning model. The learning process includes:

[0021] The edge computing unit broadcasts the model parameters of the edge federated learning model to the vehicle terminal and the roadside terminal via the downlink, so that the vehicle terminal and the roadside terminal synchronize and initialize the corresponding local model;

[0022] The vehicle terminal and the roadside terminal train and update local model parameters based on local datasets.

[0023] The vehicle terminal and the roadside terminal send the updated local model parameters to the edge computing unit via the uplink for model parameter aggregation.

[0024] Iterate through the training process until the edge federated learning model converges.

[0025] This invention also proposes an integrated sensor-computer interface (SCCI) management method for highway merging zones, used in the integrated SCCI management system for highway merging zones disclosed in any of the above technical solutions, comprising:

[0026] Vehicle perception data is acquired through the vehicle terminal, and VAI tasks are executed to process the vehicle perception data.

[0027] Roadside sensing data is acquired through roadside terminals, and IAI tasks are executed to process the roadside sensing data.

[0028] The processed vehicle perception data and roadside perception data are sent to the edge computing unit, where the perception data are fused by executing an EAI task.

[0029] Variable speed limit rules are determined based on traffic congestion levels, and a KNN-LSTM-based model for predicting congestion on the main line of the merging zone is trained based on historical traffic data of the main line segment of the highway in the pre-set merging zone.

[0030] Based on the fusion processing results of the sensing data by the edge computing unit, the congestion level of the corresponding merging zone main line segment is predicted by the merging zone main line congestion prediction model, and the vehicle control strategy and safety warning information of the merging zone segment are obtained according to the variable speed limit rules.

[0031] The vehicle control strategy is fed back to the vehicle terminal via the RSU for vehicle management, and the safety warning information is fed back to the roadside terminal for warning information dissemination.

[0032] In the above technical solution, preferably, the process of determining the variable speed limit rule based on traffic congestion and training a KNN-LSTM-based mainline congestion prediction model for the merging area based on historical traffic data of the mainline section of the highway's preset merging area includes:

[0033] Traffic congestion levels on highway sections are classified into different levels, and warning speed limits are determined for each level of congestion.

[0034] Based on the historical average speed data of the main road segment in the preset merging zone, the k-means clustering algorithm is used to cluster the traffic to obtain the speed range under each level of congestion in the current road segment. The number of cluster centers is equal to the number of traffic congestion levels.

[0035] The historical average speed dataset is divided into training and testing sets, and trained using a recurrent neural network LSTM to obtain a KNN-LSTM-based mainline congestion prediction model for merging areas. This model can predict the congestion level based on real-time sensing data.

[0036] In the above technical solution, preferably, the process of predicting the congestion level of the corresponding merging zone mainline segment based on the fusion processing result of the sensing data by the edge computing unit and the merging zone mainline congestion prediction model, and obtaining the vehicle control strategy and safety warning information for the merging zone segment according to the variable speed limit rules, specifically includes:

[0037] The edge computing unit calculates the average vehicle speed of the current road segment based on the sensing data, and inputs the average vehicle speed into the main line congestion prediction model of the merging area to obtain the congestion level of the current main line segment of the merging area.

[0038] Based on the variable speed limit rules, the warning speed limit value under the current congestion level of the merging zone section is obtained;

[0039] Based on the traffic conditions and surrounding vehicle information of each vehicle, combined with the aforementioned warning speed limit, a vehicle control strategy and safety warning information are obtained for each vehicle.

[0040] The vehicle control strategy includes lane changing and acceleration / deceleration strategies, and the safety warning information includes road obstacle warning, close-range vehicle danger warning, rear vehicle overtaking warning, side vehicle collision warning, and speed limit warning.

[0041] In the above technical solution, preferably, the integrated control method of sensing and computing in the merging zone of a highway also includes:

[0042] The SNN model and the DNN model are pre-trained offline using massive amounts of historical data;

[0043] The trained SNN model and DNN model are deployed to the vehicle terminal, the roadside terminal, and the edge computing unit, respectively.

[0044] The specific process of acquiring vehicle perception data through a vehicle terminal and processing the vehicle perception data by executing a VAI task, acquiring roadside perception data through a roadside terminal and processing the roadside perception data by executing an IAI task, and sending the processed vehicle perception data and roadside perception data to an edge computing unit, and performing fusion processing of the perception data by executing an EAI task includes:

[0045] With the support of local computing resources on the vehicle, the vehicle terminal performs VAI tasks based on the SNN model to process the vehicle perception data;

[0046] With the support of local computing resources at the roadside, the roadside terminal executes IAI tasks based on the SNN model to process the roadside sensing data;

[0047] With the support of edge computing resources, the edge computing unit performs EAI tasks based on the DNN model to fuse the perception data sent from the vehicle terminal and the roadside terminal.

[0048] In the above technical solution, preferably, OTA technology is used to remotely upgrade the models of the vehicle terminal, the roadside terminal, and the edge computing unit;

[0049] Distributed machine learning is performed on the vehicle terminal, the roadside terminal, and the edge computing unit based on an edge federated learning model. The learning process includes:

[0050] The edge computing unit broadcasts the model parameters of the edge federated learning model to the vehicle terminal and the roadside terminal via the downlink, so that the vehicle terminal and the roadside terminal synchronize and initialize the corresponding local model;

[0051] The vehicle terminal and the roadside terminal train and update local model parameters based on local datasets.

[0052] The vehicle terminal and the roadside terminal send the updated local model parameters to the edge computing unit via the uplink for model parameter aggregation.

[0053] Iterate through the training process until the edge federated learning model converges.

[0054] Compared with existing technologies, the beneficial effects of this invention are as follows: By integrating advanced technologies of communication, perception, computing, and control, a real-time interactive and collaborative management network for vehicle, road, and traffic operation systems is constructed. Through the introduction of a dynamic distributed computing network architecture and a refined multi-level AI task classification strategy, data processing efficiency and decision-making accuracy are significantly improved. Simultaneously, by utilizing a variable speed limit model and federated learning technology, intelligent prediction and management of traffic congestion are achieved, optimizing driving safety. The highway merging zone control system is transformed from a "perception-first, communication-last-computation" model to a "distributed iteration of model parameters based on perception data" mode, providing drivers with a smoother and safer travel experience and propelling intelligent transportation systems towards greater intelligence and efficiency. Attached Figure Description

[0055] Figure 1 This is a schematic diagram of the architecture of an integrated sensor-computer interface control system for highway merging zones, as disclosed in one embodiment of the present invention.

[0056] Figure 2 This is a schematic diagram of the integrated sensing and computing process of a sensing and computing integrated management and control system for merging areas of highways, as disclosed in one embodiment of the present invention.

[0057] Figure 3This is a schematic diagram of a hierarchical communication control model of an integrated control system for merging zones on highways, as disclosed in one embodiment of the present invention.

[0058] Figure 4 This is a schematic diagram of the learning process of an edge federated machine learning model disclosed in an embodiment of the present invention;

[0059] Figure 5 This is a schematic diagram illustrating the division of the weaving zone between the mainline and ramps of a highway, as disclosed in one embodiment of the present invention.

[0060] Figure 6 This is a schematic diagram of the facility layout of an integrated sensor-computer interface control system for highway merging zones, as disclosed in one embodiment of the present invention.

[0061] Figure 7 This is a schematic diagram of the prediction process of a KNN-LSTM-based mainline congestion prediction model for merging areas, as disclosed in one embodiment of the present invention. Detailed Implementation

[0062] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0063] The present invention will now be described in further detail with reference to the accompanying drawings:

[0064] like Figure 1 and Figure 2 As shown, a sensor-computer integrated management and control system for merging zones of highways provided by the present invention includes: a vehicle terminal, a roadside terminal, and an edge computing unit;

[0065] The vehicle terminal communicates with the edge computing unit through the roadside unit (RSU), and the roadside terminal communicates with the edge computing unit.

[0066] The vehicle terminal sends the vehicle perception data acquired by the vehicle perception device to the edge computing unit via the RSU, and the roadside terminal sends the roadside perception data acquired by the roadside perception device to the edge computing unit.

[0067] The vehicle terminal and roadside terminal are equipped with SNN models, which execute VAI and IAI tasks respectively, and send the task calculation results to the edge computing unit. The edge computing unit is equipped with DNN models to execute EAI tasks and calculate the information uploaded by the vehicle terminal and roadside terminal. The vehicle terminal, roadside terminal and edge computing unit together form a dynamic distributed computing network.

[0068] The edge computing unit, vehicle terminal, and roadside terminal perform distributed machine learning based on the edge federated learning model. The edge computing unit coordinates and collaborates with the vehicle terminal and roadside terminal to perform iterative training until the edge federated learning model converges.

[0069] The edge computing unit performs fusion calculations based on the acquired perception data to determine the traffic operation status, vehicle control strategies, and safety warning information based on the pre-set merging zone section of the highway, and feeds it back to the vehicle terminal and roadside terminal.

[0070] In this implementation, an advanced network for real-time interaction and collaborative management of vehicle, road, and traffic operation systems is constructed by integrating communication, sensing, computing, and control technologies. By introducing a dynamic distributed computing network architecture and a refined multi-level AI task classification strategy, data processing efficiency and decision-making accuracy are significantly improved. Simultaneously, by utilizing a variable speed limit model and federated learning technology, intelligent prediction and management of traffic congestion are achieved, optimizing driving safety. The highway merging zone control system is transformed from a "sensing first, then communicating, and finally computing" model to a "distributed iteration of model parameters based on sensing data" mode, providing drivers with a smoother and safer travel experience and propelling intelligent transportation systems towards greater intelligence and efficiency.

[0071] Specifically, the aforementioned integrated control system combining communication, sensing, and computing has the following characteristics:

[0072] (1) The system integrates perception and computing, realizing an integrated process from data collection to processing. The vehicle end and roadside end preprocess the raw perception information and then upload it, while the edge computing unit is responsible for centralized processing and decision-making, which significantly improves the efficiency and accuracy of data processing.

[0073] (2) Distributed computing network: A dynamic distributed computing network for the merging zone of highways was constructed, realizing the on-demand allocation and efficient utilization of computing resources, and ensuring the real-time performance and efficiency of computing tasks.

[0074] (3) Multi-level AI task classification: Based on the different task processing locations, AI tasks are divided into three categories: VAI, IAI and EAI, which correspond to the vehicle end, roadside end and edge computing unit, respectively. This effectively utilizes the computing power of each level of equipment and improves the overall system performance.

[0075] (4) Variable speed limit model: Based on traffic congestion, variable speed limit rules are designed and KNN-LSTM model is used to predict road congestion. This realizes intelligent management of the main line speed limit in the merging zone, effectively alleviates traffic congestion and improves driving safety.

[0076] Furthermore, this invention, through integrated design, is the first to deeply integrate sensing, computing, and communication technologies to construct an integrated sensing, computing, and communication control system for highway merging zones, achieving comprehensive perception and intelligent processing of traffic information. Secondly, by constructing a dynamic distributed computing network, it innovatively proposes a dynamic distributed computing network for highway merging zones, enabling flexible scheduling and efficient utilization of computing resources, providing strong support for large-scale data processing and real-time decision-making. Finally, the application of multi-level AI task classification and federated learning models not only optimizes system resource allocation but also improves model iteration speed and machine learning capabilities, laying the foundation for the sustainable development of intelligent transportation systems.

[0077] The integrated sensing, sensing, and computing management system for highway merging areas consists of one control center, L roadside units (RSUs), an edge computing unit, M multi-functional terminals, and N targets to be sensed. Vehicle terminals and roadside terminals utilize radar, cameras, and other equipment to collect real-time data on vehicle operation status, road conditions, and weather conditions. The RSUs receive the sensing data from the vehicle terminals and transmit it to the edge computing units. The edge computing units or control center aggregate the vehicle sensing data from the vehicle terminals and the roadside sensing data from the roadside terminals, and use built-in algorithm models to formulate merging area management strategies. Within this integrated sensing, sensing, and computing management system, the servers of the vehicle terminals and roadside terminals, along with the servers of the edge computing units, construct a dynamic distributed computing network for highway merging areas. This network provides on-demand computing power support for tasks such as multi-dimensional sensing information fusion, feature extraction, and autonomous decision-making by the vehicle terminals, roadside terminals, and edge computing units.

[0078] Specifically, in the integrated sensing, perception, and computing management system for highway merging zones, the Roadside Unit (RSU) can provide vehicle-road cooperative services such as cooperative driving, beyond-line-of-sight perception, and collision warning to vehicles within its coverage area, and provide relay communication between vehicle terminals and edge computing units. The edge computing unit needs to use artificial intelligence (AI) algorithms to fuse and process various types of perception information sensed by vehicle terminals and roadside terminals. Vehicle terminals process the raw perception information acquired by onboard sensors such as radar and video, and upload the processed results to the RSU, which then uploads them to the edge computing unit. Similarly, roadside terminals process the raw perception information collected by roadside perception devices such as radar and video, and upload the processed results to the edge computing unit. The edge computing unit server can acquire vehicle perception information and roadside perception information in real time, enabling centralized control and decision-making for perception information processing. AI tasks are tasks processed using pre-trained AI models, and outputs are obtained through inference. In the integrated sensing, perception, and computing management system for highway merging zones, AI tasks are represented as the fusion, dimensionality reduction, and recognition of various types of perception information.

[0079] Specifically, based on the location where the AI ​​task is processed (vehicle terminal, roadside terminal, or edge computing unit), AI tasks can be further classified into VAI (Vehicle Artificial Intelligence) tasks, IAI (Infrastructure Artificial Intelligence) tasks, and EAI (Edge Artificial Intelligence) tasks.

[0080] (1) VAI task:

[0081] At time k, vehicle i has computational tasks that need to be processed by AI algorithms. The computational power consumed by these algorithms depends on the amount of video and radar perception information of vehicle i. Vehicle i's VAI tasks are processed locally, and the vehicle uploads the results to the RSU.

[0082] (2) IAI task

[0083] At time k, roadside terminal j has computational tasks that need to be processed by AI algorithms. The amount of computing power consumed by these algorithms depends on the amount of information collected by the video and radar equipment deployed on roadside terminal j. The IAI tasks of roadside terminal j are processed locally, and the results are uploaded to the edge computing unit.

[0084] (3) EAI Task

[0085] At time k, edge computing unit j has a computational task that needs to be processed by an AI algorithm. The amount of computing power consumed by this algorithm depends on the amount of perception information (raw perception information or locally processed information) uploaded by the vehicle terminal and the roadside terminal.

[0086] In the above embodiments, preferably, the vehicle sensing device includes an on-board sensor, an on-board radar, and an on-board camera, and the roadside sensing device includes a roadside radar, a roadside camera, and a weather detector.

[0087] The roadside terminal also includes variable message signs, which receive and publish safety warning information pushed by the edge computing unit.

[0088] In order for vehicles to pass safely and smoothly through the merging section of the main line and ramps, it is necessary to obtain real-time information on the surrounding traffic conditions, including vehicle speed, traffic incidents, and vehicle queues.

[0089] In order to adjust vehicle control strategies such as lane changing, acceleration and deceleration, vehicles need to obtain information such as the position and speed of surrounding vehicles in real time.

[0090] To avoid traffic accidents, vehicles need to receive real-time safety warnings such as road obstacle warnings, close-range danger warnings, overtaking warnings from vehicles behind, and side collision warnings.

[0091] Specifically, at the merging sections of the main line and ramps, the road environment changes and traffic safety risks increase. Therefore, roadside smart facilities such as sensing, communication, and control systems are deployed, and vehicle-road cooperative systems are built. Visibility and road surface condition detection equipment is also deployed to perceive the traffic and meteorological environment in real time and issue meteorological warning information to ensure driving safety at the merging sections of the main line and ramps.

[0092] The sensing facilities include:

[0093] (1) Traffic operation status perception: traffic flow, average vehicle speed, occupancy rate, etc.

[0094] (2) Vehicle operation status perception: vehicle speed, vehicle trajectory, abnormal acceleration and deceleration of vehicles, etc.

[0095] (3) Traffic incident perception: queue length, traffic congestion, traffic accidents, etc.;

[0096] (4) AI computing power (SNN).

[0097] The communication facilities include:

[0098] (1) Wired communication facilities support high-bandwidth data transmission of images and videos;

[0099] (2) Wireless communication facilities (RSU).

[0100] The edge computing units and control facilities include:

[0101] (1) Edge computing unit: Accesses information pushed by vehicle terminals and roadside terminal sensing facilities, analyzes and processes the information, generates safety warning information such as dynamic speed limits and traffic incidents (AI computing power, DNN), and pushes it to roadside RSUs and variable message signs;

[0102] (2) Roadside RSU: Receives the perception information processed by the vehicle terminal or the original perception information, receives the safety warning information pushed by the edge computing terminal, and pushes it to the vehicle OBU to provide the driver with voice / image information;

[0103] (3) Variable Information Flag: Receives safety warning information pushed by the edge computing unit and publishes it to the driver.

[0104] In the above embodiments, preferably, the roadside terminals are distributed on the upstream and downstream sections and ramp sections of the mainline merging point in the pre-set merging area of ​​the expressway.

[0105] Specifically, such as Figure 5 As shown, due to the differences in the design of each ramp, the number and location of roadside facilities and meteorological monitoring facilities need to be determined based on the actual conditions of each mainline and ramp merging section in actual engineering. To ensure driving safety in the mainline and ramp merging area, the basic deployment principle is: no blind spot coverage.

[0106] like Figure 6 As shown, the basic deployment scheme is as follows:

[0107] Main line: One integrated sensor-computer interface facility is deployed on each of the upstream and downstream sections of the merging point, including equipment such as video, radar, edge computing, and RSU;

[0108] Ramp: Deploy a set of integrated sensing and computing facilities, including video, radar, edge computing, RSU and other equipment; if the ramp has a large curvature, or the ramp is long and wide, and a set of equipment cannot cover it, then additional equipment needs to be installed as needed.

[0109] In the above embodiments, preferably, the vehicle terminal performs VAI tasks based on the SNN model with the support of local computing resources on the vehicle end, and processes the vehicle perception data acquired by the vehicle perception device.

[0110] With the support of local computing resources at the roadside, the roadside terminal executes IAI tasks based on the SNN model to process the roadside sensing data acquired by the roadside sensing devices;

[0111] With the support of edge computing resources, the edge computing unit performs EAI tasks based on the DNN model to fuse and process the perception data sent from vehicle terminals and roadside terminals.

[0112] Among them, the SNN model and the DNN model are pre-trained offline using massive amounts of historical data.

[0113] In the above embodiments, preferably, the vehicle terminal, roadside terminal, and edge computing unit use OTA technology to remotely upgrade the model;

[0114] Vehicle terminals, roadside terminals, and edge computing units perform distributed machine learning based on an edge federated learning model. The learning process includes:

[0115] The edge computing unit broadcasts the model parameters of the edge federated learning model to the vehicle terminal and the roadside terminal via the downlink, so that the vehicle terminal and the roadside terminal can synchronize and initialize the corresponding local model.

[0116] The vehicle terminal and roadside terminal train and update local model parameters based on the local dataset;

[0117] The vehicle terminal and roadside terminal will send the updated local model parameters to the edge computing unit via the uplink for model parameter aggregation;

[0118] Iterate through training until the edge federated learning model converges.

[0119] Specifically, the aforementioned integrated control system for sensing and computing also features an efficient model algorithm upgrade strategy: it adopts OTA (Over-the-Air Technology) to achieve remote upgrades of global, edge, and local models, and the edge computing unit and multi-functional terminal achieve distributed machine learning based on federated learning models, which not only ensures data privacy and security, but also improves model iteration speed and machine learning capabilities.

[0120] like Figure 3 As shown, in this technical system, the algorithm model is divided into three levels: global model, edge model, and local model. Both edge devices (vehicle terminals, roadside terminals, and edge computing units) and central servers utilize OTA (Over-The-Air) technology to remotely upgrade the model algorithm. Both central and edge model algorithms are upgraded by downloading and installing new program packages. However, significant differences exist between the edge and central systems in the core aspects of algorithm upgrades—data iteration and machine learning.

[0121] Central servers possess significantly greater computing power than edge devices. While handling computational tasks, they can allocate a substantial portion of their computing power (e.g., 50%) to machine learning. This results in rapid data iteration and strong machine learning capabilities at the central level, enabling rapid upgrades of the global model. Edge devices, on the other hand, primarily utilize their computing power for edge computing and perception tasks, with only a limited portion (e.g., 10%–20%) allocated to machine learning. This limited power restricts their ability to simultaneously perform perception, computation, and training. Furthermore, machine learning on edge devices is typically time-consuming and slow. The central server supports the evolution and updating of model algorithms, with upgraded models pushed to edge computing units via OTA (Over-The-Air) technology.

[0122] like Figure 4 As shown, edge computing units, vehicle terminals, and roadside terminals implement privacy-preserving distributed machine learning based on an edge federated learning model. The edge federated learning model utilizes the computing power of multiple intelligent terminals, coordinated by the edge computing unit, to collaboratively train the edge model in an iterative manner. The edge federated learning process includes the following three steps.

[0123] Step 1: The edge computing unit broadcasts the edge model parameters to the terminal via the downlink so that all terminals participating in federated learning can synchronize and initialize their local models.

[0124] Step 2: Each terminal participating in federated learning trains and updates its local model parameters based on its own dataset.

[0125] Step 3: The terminal sends the updated local model parameters to the edge computing unit of the base station via the uplink for edge model parameter aggregation. The iteration stops when the edge model converges.

[0126] According to the above-described implementation method, the integrated sensing and computing control system for highway merging zones integrates the sensing and computing processes in its system architecture. Based on the AI ​​task processing at the terminal, data is transmitted to the edge computing unit via a communication unit for fusion computing, forming a multi-level AI task classification and processing method. This constructs a dynamic distributed computing network, enabling on-demand allocation and efficient utilization of computing resources. Furthermore, based on the calculated data, control strategies can be calculated and analyzed according to a built-in algorithm model, and the corresponding control strategy information can be fed back to vehicle terminals and roadside terminals, achieving vehicle control in highway merging zones.

[0127] This invention also proposes an integrated sensor-computer interface (SCCI) management method for highway merging zones, used in the integrated SCCI management system for highway merging zones disclosed in any of the above embodiments, comprising:

[0128] Vehicle perception data is acquired through the vehicle terminal, and VAI tasks are executed to process the vehicle perception data.

[0129] Roadside sensing data is acquired through roadside terminals, and IAI tasks are executed to process the roadside sensing data.

[0130] The processed vehicle perception data and roadside perception data are sent to the edge computing unit, where the perception data is fused by executing an EAI task.

[0131] Variable speed limit rules are determined based on traffic congestion levels, and a KNN-LSTM-based model for predicting congestion on the main line of the merging zone is trained based on historical traffic data of the main line segment of the highway in the pre-set merging zone.

[0132] Based on the fusion processing results of the sensing data by the edge computing unit, the congestion level of the corresponding main line segment in the merging area is predicted by the main line congestion prediction model, and the vehicle control strategy and safety warning information of the merging area segment are obtained according to the variable speed limit rules.

[0133] Vehicle control strategies are fed back to vehicle terminals via RSUs for vehicle management, and safety warning information is fed back to roadside terminals for warning dissemination.

[0134] In this implementation, an advanced network for real-time interaction and collaborative management of vehicle, road, and traffic operation systems is constructed by integrating communication, sensing, computing, and control technologies. By introducing a dynamic distributed computing network architecture and a refined multi-level AI task classification strategy, data processing efficiency and decision-making accuracy are significantly improved. Simultaneously, by utilizing a variable speed limit model and federated learning technology, intelligent prediction and management of traffic congestion are achieved, optimizing driving safety. The highway merging zone control system is transformed from a "sensing first, then communicating, and finally computing" model to a "distributed iteration of model parameters based on sensing data" mode, providing drivers with a smoother and safer travel experience and propelling intelligent transportation systems towards greater intelligence and efficiency.

[0135] During implementation, based on the real-time sensing data collected by vehicle terminals and roadside terminals, and with the computing power support of a dynamic distributed computing network composed of vehicle terminals, roadside terminals, and edge computing units, the congestion level of the corresponding road segment is analyzed and predicted by a KNN-LSTM-based mainline congestion prediction model. The vehicle speed limit warning value for the current lane is then calculated comprehensively based on predetermined variable speed limit rules.

[0136] Based on the calculated data, vehicle control can be implemented for vehicles in the current lane, enabling the control of driverless vehicles. Warning information can also be issued to vehicles in the current lane to remind drivers to drive safely. Warning information can also be issued through roadside terminals on the current road segment to remind drivers to drive safely.

[0137] In the above implementation, preferably, a variable speed limit rule is determined based on traffic congestion, and a KNN-LSTM-based mainline congestion prediction model for the merging zone is trained based on historical traffic data of the mainline section of the highway's preset merging zone. The specific process includes:

[0138] Traffic congestion levels on highway sections are classified into different levels, and warning speed limits are determined for each level of congestion.

[0139] Based on the historical average speed data of the main road segment in the preset merging zone, the k-means clustering algorithm is used to cluster the traffic to obtain the speed range under each level of congestion in the current road segment. The number of cluster centers is equal to the number of traffic congestion levels.

[0140] The historical average speed dataset is divided into training and testing sets, and trained using a recurrent neural network LSTM to obtain a KNN-LSTM-based mainline congestion prediction model for merging areas. This model can predict the congestion level based on real-time sensing data.

[0141] Specifically, the control objective of variable speed limit rules is to reduce the dispersion of vehicle speeds, control vehicle speed in advance, ensure safety while alleviating traffic congestion, and reducing the probability of secondary accidents. The speed limit triggered by the system is determined based on the level of traffic congestion, with different optimal speed limits corresponding to different levels of traffic congestion.

[0142] Rule: Variable speed limit rule based on traffic congestion:

[0143] If C = N, then S limit =V congestion

[0144] Where C represents traffic congestion level, N represents congestion level value, and S represents traffic congestion level. limit (km / h) is the warning speed limit value, V congestion

[0145] (km / h) represents the warning speed limit value under different levels of congestion.

[0146] Traffic congestion levels are classified based on micro-parameters and traffic data, with vehicle speed being a commonly used parameter for evaluating traffic conditions. Table 1 shows how vehicle speed is used as the standard for classifying congestion levels on highways, and how variable speed limits are determined based on these levels.

[0147] Table 1 Variable speed limit rules based on traffic congestion

[0148]

[0149] During implementation, the k-means clustering algorithm uses 5 cluster centers because road congestion levels are divided into five levels: smooth, mostly smooth, moderate, congested, and blocked.

[0150] like Figure 7 As shown, the model is trained by inputting historical average speeds. After training, the historical average speeds are divided into 5 categories, and the upper and lower limits of each category correspond to the standard interval for classifying congestion levels. An LSTM recurrent neural network is used, dividing the historical average speed dataset into training and testing sets. During training, the number of layers and neurons in each layer are adjusted to maximize the network's prediction accuracy.

[0151] In the above implementation, preferably, based on the fusion processing results of the sensing data by the edge computing unit, the congestion level of the corresponding merging zone mainline segment is predicted by the merging zone mainline congestion prediction model, and the vehicle control strategy and safety warning information for the merging zone segment are obtained according to the variable speed limit rules. The specific process includes:

[0152] The edge computing unit calculates the average vehicle speed of the current road segment based on the perception data, and inputs the average vehicle speed into the main line congestion prediction model of the merging area to obtain the congestion level of the current main line segment of the merging area.

[0153] Based on the variable speed limit rules, the warning speed limit value under the current congestion level of the merging zone section is obtained;

[0154] Based on the traffic conditions and surrounding vehicle information of each vehicle, combined with the warning speed limit, a vehicle control strategy and safety warning information are obtained for each vehicle.

[0155] The vehicle control strategies include lane changing and acceleration / deceleration strategies, and the safety warning information includes road obstacle warnings, close-range vehicle danger warnings, overtaking warnings from behind, side collision warnings, and speed limit warnings.

[0156] In the above embodiments, preferably, the integrated control method for the merging zone of a highway further includes:

[0157] Offline pre-training of SNN and DNN models using massive amounts of historical data;

[0158] The trained SNN and DNN models are deployed to vehicle terminals, roadside terminals, and edge computing units, respectively.

[0159] Based on this, the above-mentioned process involves acquiring vehicle perception data through vehicle terminals and processing it using VAI (Vehicle Awareness) tasks, acquiring roadside perception data through roadside terminals and processing it using IAI (Infrastructure Awareness) tasks, and sending the processed vehicle perception data and roadside perception data to the edge computing unit for fusion processing using EAI (Emergency Assistive Technology) tasks. The specific process includes:

[0160] With the support of local computing resources on the vehicle, the vehicle terminal performs VAI tasks based on the SNN model to process vehicle perception data;

[0161] With the support of local computing resources at the roadside, the roadside terminal performs IAI tasks based on the SNN model to process roadside sensing data;

[0162] With the support of edge computing resources, the edge computing unit performs EAI tasks based on the DNN model to fuse and process the perception data sent from vehicle terminals and roadside terminals.

[0163] In the above embodiments, preferably, OTA technology is used to remotely upgrade the models of vehicle terminals, roadside terminals, and edge computing units;

[0164] Distributed machine learning is performed on vehicle terminals, roadside terminals, and edge computing units based on an edge federated learning model. The learning process includes:

[0165] The edge computing unit broadcasts the model parameters of the edge federated learning model to the vehicle terminal and the roadside terminal via the downlink, so that the vehicle terminal and the roadside terminal can synchronize and initialize the corresponding local model.

[0166] The vehicle terminal and roadside terminal train and update local model parameters based on the local dataset;

[0167] The vehicle terminal and roadside terminal will send the updated local model parameters to the edge computing unit via the uplink for model parameter aggregation;

[0168] Iterate through training until the edge federated learning model converges.

[0169] According to the integrated management and control method for merging zones of highways disclosed in the above embodiments, each step is implemented based on the integrated management and control system for merging zones of highways disclosed in the above embodiments. During implementation, the deployment and implementation methods of each module of the system disclosed in the above embodiments can be referred to for operation, and will not be repeated here.

[0170] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A sensor-computer integrated control system for merging zones of highways, characterized in that, include: Vehicle terminals, roadside terminals, and edge computing units; The vehicle terminal is communicatively connected to the edge computing unit via a roadside unit (RSU), and the roadside terminal is communicatively connected to the edge computing unit. The vehicle terminal sends the vehicle perception data acquired by the vehicle perception device to the edge computing unit through the RSU, and the roadside terminal sends the roadside perception data acquired by the roadside perception device to the edge computing unit. The vehicle terminal and the roadside terminal are equipped with SNN models, which execute VAI and IAI tasks respectively, and send the task calculation results to the edge computing unit. The edge computing unit is equipped with DNN models to execute EAI tasks and calculate the information uploaded by the vehicle terminal and the roadside terminal. The vehicle terminal, the roadside terminal and the edge computing unit jointly construct a dynamic distributed computing network. The edge computing unit, the vehicle terminal, and the roadside terminal perform distributed machine learning based on the edge federated learning model. The edge computing unit coordinates and cooperates with the vehicle terminal and the roadside terminal to perform iterative training until the edge federated learning model converges. The edge computing unit performs fusion calculations based on the acquired perception data to determine the traffic operation status, vehicle control strategy, and safety warning information based on the preset merging zone section of the highway, and feeds it back to the vehicle terminal and the roadside terminal. The vehicle terminal, the roadside terminal, and the edge computing unit use OTA technology for remote model upgrades; The vehicle terminal, the roadside terminal, and the edge computing unit perform distributed machine learning based on an edge federated learning model. The learning process includes: The edge computing unit broadcasts the model parameters of the edge federated learning model to the vehicle terminal and the roadside terminal via the downlink, so that the vehicle terminal and the roadside terminal synchronize and initialize the corresponding local model; The vehicle terminal and the roadside terminal train and update local model parameters based on local datasets. The vehicle terminal and the roadside terminal send the updated local model parameters to the edge computing unit via the uplink for model parameter aggregation. Iterate through the training process until the edge federated learning model converges.

2. The integrated sensor-computer interface control system for highway merging zones according to claim 1, characterized in that, The vehicle sensing device includes onboard sensors, onboard radar, and onboard cameras; the roadside sensing device includes roadside radar, roadside cameras, and weather detectors. The roadside terminal also includes a variable information sign, which receives and publishes safety warning information pushed by the edge computing unit.

3. The integrated sensor-computer interface control system for highway merging zones according to claim 1, characterized in that, The roadside terminals are distributed on the upstream and downstream sections and ramp sections of the main line merging point in the pre-designated merging area of ​​the expressway.

4. The integrated sensor-computer interface control system for highway merging zones according to claim 1, characterized in that, The vehicle terminal, supported by local computing resources on the vehicle, executes VAI tasks based on an SNN model to process vehicle perception data acquired by the vehicle perception device. With the support of local computing resources at the roadside, the roadside terminal executes IAI tasks based on the SNN model to process the roadside sensing data acquired by the roadside sensing device. With the support of edge computing resources, the edge computing unit performs EAI tasks based on the DNN model to fuse the perception data sent from the vehicle terminal and the roadside terminal. The SNN model and the DNN model are pre-trained offline using massive amounts of historical data.

5. A method for integrated control of sensing and calculation in merging zones of highways, characterized in that, The integrated sensor-computer interface control system for the merging zone of a highway as described in any one of claims 1 to 4 includes: Vehicle perception data is acquired through the vehicle terminal, and VAI tasks are executed to process the vehicle perception data. Roadside sensing data is acquired through roadside terminals, and IAI tasks are executed to process the roadside sensing data. The processed vehicle perception data and roadside perception data are sent to the edge computing unit, where the perception data are fused by executing an EAI task. Variable speed limit rules are determined based on traffic congestion levels, and a KNN-LSTM-based model for predicting congestion on the main line of the merging zone is trained based on historical traffic data of the main line segment of the highway in the pre-set merging zone. Based on the fusion processing results of the sensing data by the edge computing unit, the congestion level of the corresponding merging zone main line segment is predicted by the merging zone main line congestion prediction model, and the vehicle control strategy and safety warning information of the merging zone segment are obtained according to the variable speed limit rules. The vehicle control strategy is fed back to the vehicle terminal via the RSU for vehicle management, and the safety warning information is fed back to the roadside terminal for warning information dissemination.

6. The integrated sensor-computer interface control method for merging zones of highways according to claim 5, characterized in that, The process of determining variable speed limit rules based on traffic congestion and training a KNN-LSTM-based mainline congestion prediction model for merging zones using historical traffic data from pre-defined merging zone mainline sections of highways includes: Traffic congestion levels on highway sections are classified into different levels, and warning speed limits are determined for each level of congestion. Based on the historical average speed data of the main road segment in the preset merging zone, the k-means clustering algorithm is used to cluster the traffic to obtain the speed range under each level of congestion in the current road segment. The number of cluster centers is equal to the number of traffic congestion levels. The historical average speed dataset is divided into training and testing sets, and trained using a recurrent neural network LSTM to obtain a KNN-LSTM-based mainline congestion prediction model for merging areas. This model can predict the congestion level based on real-time sensing data.

7. The integrated sensor-computer interface control method for merging zones of highways according to claim 6, characterized in that, The process of predicting the congestion level of the corresponding merging zone mainline segment based on the fusion processing results of the sensing data by the edge computing unit and the merging zone mainline congestion prediction model, and obtaining the vehicle control strategy and safety warning information for the merging zone segment according to the variable speed limit rules, specifically includes: The edge computing unit calculates the average vehicle speed of the current road segment based on the sensing data, and inputs the average vehicle speed into the main line congestion prediction model of the merging area to obtain the congestion level of the current main line segment of the merging area. Based on the variable speed limit rules, the warning speed limit value under the current congestion level of the merging zone section is obtained; Based on the traffic conditions and surrounding vehicle information around each vehicle, combined with the aforementioned warning speed limit, a vehicle control strategy and safety warning information are obtained for each vehicle. The vehicle control strategy includes lane changing and acceleration / deceleration strategies, and the safety warning information includes road obstacle warning, close-range vehicle danger warning, rear vehicle overtaking warning, side vehicle collision warning, and speed limit warning.

8. The integrated sensor-computer interface control method for merging zones of highways according to claim 5, characterized in that, Also includes: The SNN model and the DNN model are pre-trained offline using massive amounts of historical data; The trained SNN model and DNN model are deployed to the vehicle terminal, the roadside terminal, and the edge computing unit, respectively. The specific process of acquiring vehicle perception data through a vehicle terminal and processing the vehicle perception data by executing a VAI task, acquiring roadside perception data through a roadside terminal and processing the roadside perception data by executing an IAI task, and sending the processed vehicle perception data and roadside perception data to an edge computing unit, and performing fusion processing of the perception data by executing an EAI task includes: With the support of local computing resources on the vehicle, the vehicle terminal performs VAI tasks based on the SNN model to process the vehicle perception data; With the support of local computing resources at the roadside, the roadside terminal executes IAI tasks based on the SNN model to process the roadside sensing data; With the support of edge computing resources, the edge computing unit performs EAI tasks based on the DNN model to fuse the perception data sent from the vehicle terminal and the roadside terminal.

9. The integrated sensor-computer interface control method for merging zones of highways according to claim 8, characterized in that, The vehicle terminal, the roadside terminal, and the edge computing unit are remotely upgraded using OTA technology. Distributed machine learning is performed on the vehicle terminal, the roadside terminal, and the edge computing unit based on an edge federated learning model. The learning process includes: The edge computing unit broadcasts the model parameters of the edge federated learning model to the vehicle terminal and the roadside terminal via the downlink, so that the vehicle terminal and the roadside terminal synchronize and initialize the corresponding local model; The vehicle terminal and the roadside terminal train and update local model parameters based on local datasets. The vehicle terminal and the roadside terminal send the updated local model parameters to the edge computing unit via the uplink for model parameter aggregation. Iterate through the training process until the edge federated learning model converges.

Images