Roadway vehicle-road cooperative unmanned vehicle control system

Abstract

The present application relates to unmanned technology field, especially to a kind of roadway in-car road coordination unmanned vehicle control system. Including roadside data acquisition unit, roadside computing unit, car end control unit, cloud management unit;The present application applies car road coordination system to the environment in roadway, and most of data acquisition and calculation are completed through roadside data acquisition unit and roadside computing unit, so as to reduce the load of car end control unit, improve the communication efficiency between unmanned vehicle and roadside equipment;The present application processes roadway environment data acquisition and unmanned vehicle data acquisition separately, assesses the obstacles in roadway and possible environmental hazards through roadside data acquisition unit and roadside computing unit, and supplements data through car end control unit, to ensure the safe operation of unmanned vehicle;Cloud management unit assesses the roadway interval Bi where the unmanned vehicle is located, and then determines whether to re-plan the route.

Description

Technical Field

[0001] This invention relates to the field of autonomous driving technology, and in particular to a vehicle-road cooperative unmanned vehicle control system in an alleyway. Background Technology

[0002] As autonomous driving technology matures, it will be adopted in more and more scenarios, especially in situations that are inherently dangerous but do not require human control. Therefore, the use of autonomous vehicles in mine tunnels has become a trend. However, applying autonomous vehicle technology in mine tunnels still faces many challenges. Compared to open roads, signal transmission in mine tunnels is poor, and the low light and high dust levels also affect the control of autonomous vehicles. Therefore, current autonomous vehicle control systems often work in conjunction with vehicle-road cooperative systems to more precisely control the vehicles, enabling them to avoid obstacles and follow reasonable routes.

[0003] However, in existing technologies, vehicle-road cooperation mostly deals with relationships between vehicles on the ground, between vehicles and traffic signal equipment, and between vehicles and pedestrians. In contrast, the space inside mine tunnels is relatively small, and various pipes and other necessary equipment are installed on the tunnel walls. Therefore, the environment inside the tunnels is relatively simple, and there are fewer interaction scenarios in the vehicle-road cooperation system compared to the above-ground scenarios. However, the environment inside the tunnels is related to the driving safety of unmanned vehicles. When controlling unmanned vehicles, it is necessary not only to avoid collisions between unmanned vehicles and obstacles, but also to consider the impact of changes in the tunnel environment, such as falling rocks, damaged or detached equipment, water accumulation, and debris accumulation, on the control of unmanned vehicles.

[0004] To address the aforementioned technical shortcomings, a solution is proposed. Summary of the Invention

[0005] The purpose of this invention is to provide a vehicle-road cooperative unmanned vehicle control system in alleyways to address the aforementioned technical deficiencies. This invention applies a vehicle-road cooperative system to alleyway environments, completing most data acquisition and computation through roadside data acquisition and computation units, reducing the load on the vehicle-side control unit and improving communication efficiency between the unmanned vehicle and roadside equipment. This invention processes alleyway environmental data acquisition and unmanned vehicle data acquisition separately. The roadside data acquisition and computation units assess obstacles and potential environmental hazards within the alleyway, while the vehicle-side control unit supplements the data to ensure the safe operation of the unmanned vehicle. Using the roadside early warning assessment value GL and risk assessment parameter Rs, the cloud management unit evaluates the alleyway section Bi where the unmanned vehicle is located, thereby determining whether route planning needs to be re-planned.

[0006] The objective of this invention can be achieved through the following technical solution: a vehicle-road cooperative unmanned vehicle control system in a roadway. It includes a roadside unit, a vehicle-mounted unit, and a cloud-based unit; the roadside unit includes a roadside data acquisition unit and a roadside computing unit; the vehicle-mounted unit includes a vehicle-mounted control unit; and the cloud-based unit includes a cloud-based management unit. The roadside unit interacts with the cloud-based unit via a wireless communication network, and the roadside unit and the vehicle-mounted unit interact via V2X.

[0007] The roadside data acquisition unit is used to collect environmental perception data inside the mine roadways.

[0008] The roadside data acquisition unit includes a wireless communication network and a roadside environmental perception system; the wireless communication network of the roadside data acquisition unit includes a 4G / 5G network and a Wi-Fi network.

[0009] The roadside environmental perception system includes lidar, mining cameras, temperature and humidity sensors, infrared sensors, and UWB base stations; when the infrared sensors collect heat source data, the lidar does not collect data.

[0010] It also includes a roadside computing unit, which is used to receive and store environmental perception data inside the mine roadway and calculate roadside point cloud data. Within a set time frequency, it calculates the roadside warning assessment value GL. If the warning assessment value GL exceeds the threshold GB, a roadside warning signal is generated and sent to the vehicle-side control unit through the roadside V2X unit. The roadside warning assessment value GL is also sent to the cloud management unit through the wireless communication network.

[0011] The vehicle-mounted control unit is used to collect road perception data along the route of the mine roadway; the vehicle-mounted control unit includes a vehicle-mounted perception system, an on-board V2X unit, an on-board computing host, a wireless communication network, a UWB positioning unit, an inertial measurement unit, and a speed sensor unit; the vehicle-mounted perception system includes an on-board LiDAR and an on-board camera, used to collect road perception data along the route of the mine roadway; the wireless communication network of the vehicle-mounted control unit includes 4G / 5G network and Wi-Fi network.

[0012] The vehicle-mounted control unit assesses whether abnormal data Zi exists. If abnormal data Zi is found, a vehicle-mounted warning signal is generated and sent to the roadside computing unit via the vehicle-mounted V2X unit. The roadside computing unit controls the roadside environmental perception system to collect environmental perception data inside the mine roadway through the wireless communication network, and performs perception data fusion by combining the abnormal data Zi. The roadside computing unit calculates the risk assessment parameter Rs and sends the risk assessment parameter Rs to the cloud management unit through the wireless communication network. The cloud management unit determines whether the unmanned vehicle should replan its route based on the risk assessment parameter Rs.

[0013] The cloud management unit is used for data storage and analysis, and for controlling unmanned vehicles; the roadside computing unit periodically sends the data it stores to the cloud management unit via a wireless communication network. The cloud management unit plans the route of the unmanned vehicle based on the stored historical data, generates the route instructions, and sends the route instructions to the roadside computing unit through the wireless communication network. The roadside computing unit sends the instructions to the vehicle control unit through the roadside V2X unit, and the vehicle control unit controls the operation of the unmanned vehicle. The historical data includes, but is not limited to, environmental perception data inside the mine roadway, roadside point cloud data, abnormal data Zi, roadside early warning assessment value GL, and risk assessment parameter Rs.

[0014] Furthermore, to better manage environmental data within the mine roadways, a segmented approach is adopted, with each roadway designated and numbered. Data from each segment is collected and stored separately. This allows for targeted alerts and data updates only when a problem occurs in a particular segment, avoiding the need to process the entire roadway. Additionally, the varying environments across different road sections necessitate segmented data collection to accurately reflect the actual conditions within the mine roadways. The cloud-based management unit divides the roadway map data into several roadway segments Bi, i=1~n, where n is a positive integer. Each roadway segment Bi is equipped with a roadside data collection unit and a roadside calculation unit. The coordinates of each roadway segment Bi are determined using multiple sets of UWB base stations installed within the mine roadways. The roadside data acquisition unit communicates with the roadside computing unit via a wireless communication network; the roadside computing unit communicates with the cloud management unit via a wireless communication network.

[0015] To assess the condition of the tunnel walls and their associated equipment, roadside point cloud data is calculated. This calculation reflects the environmental conditions within the tunnel. When some point cloud data exceeds a preset range, it may indicate issues such as falling rocks, debris accumulation, or equipment damage, which could affect the passage of unmanned vehicles. The roadside data acquisition unit and roadside calculation unit can mitigate these risks in advance, while minimizing redundant data collection and calculations between the roadside data acquisition unit and the vehicle-mounted control unit. The specific steps include: S11. Raw point cloud data acquisition: Install several lidars in each tunnel section Bi so that the lidar scanning range can cover the road, inner wall and inner wall auxiliary equipment of the tunnel section Bi in which it is located; after the lidar scan, the raw point cloud data is obtained, which includes the three-dimensional coordinates and timestamp of each point. S12. Perform data preprocessing: Calculate the mean neighborhood distance of each point, remove points that exceed the preset threshold, and then divide the remaining point cloud data into a three-dimensional grid. Each three-dimensional grid is represented by a representative point, which is the centroid of the three-dimensional grid. Then, transform the coordinates of the representative point to obtain the preprocessed point cloud data. S13. Registration of preprocessed point cloud data: Align the preprocessed point cloud data of different frames using the RANSAC algorithm to obtain coarse point cloud data; S14. Registration of coarse point cloud data: The coarse point cloud data is precisely aligned using the ICP algorithm to obtain the point cloud data. S15. Feature Extraction: The normal vector of the point cloud surface is calculated by the normal estimation method to obtain the simulated geometric structure of the road in the lane section Bi, and the simulated geometric structure of the inner wall and the inner wall auxiliary equipment of the lane section Bi. S16. Data storage: The generated point cloud data is stored in the roadside computing unit. The roadside computing unit packages the point cloud data together with the identification data of the lane section Bi where the roadside computing unit is located, and sends the packaged point cloud data to the cloud management unit through the wireless communication network. The identification data of the lane section Bi is the coordinate range of the lane section Bi.

[0016] To determine whether there are obstacles or obstacle risks in the alleyway environment, the roadside early warning assessment value GL is calculated using point cloud data. The specific steps include: S21. Calculate abnormal point cloud data: The roadside computing unit acquires point cloud data, extracts point cloud data of the road area, records it as road point cloud data, sets filter condition K, counts the road point cloud data that meet the condition, and obtains the first abnormal data M; The roadside computing unit extracts point cloud data of the inner wall and inner wall auxiliary equipment, records it as inner wall point cloud data, sets filter condition L, counts the inner wall point cloud data that meet the condition, and obtains the second abnormal data N; S22, Calculation parameters Calculate the ratio of the first abnormal data point M to the set threshold m, calculate the ratio of the second abnormal data point N to the set threshold n, and calculate the parameters. ; S23, Calculation Parameters The roadside calculation unit records the number P of roadside early warning signals generated in its respective lane section, and calculates parameters. ,in These are historical data weighting coefficients; S24, Calculation Parameters The roadside computing unit acquires temperature data T and humidity data H collected by the roadside data acquisition unit, and calculates... ,in This represents the historical minimum temperature within the Bi section of the roadway. This indicates the historical maximum temperature within section Bi of the tunnel. This represents the historical minimum humidity value within the Bi section of the tunnel. This indicates the historical maximum humidity value within the Bi section of the tunnel. These are the weighting coefficients for the temperature data. These are the weighting coefficients for humidity data; S25, Calculate the roadside early warning assessment value GL: If the warning assessment value GL is greater than or equal to the threshold GB, a roadside warning signal will be generated; otherwise, no roadside warning signal will be generated.

[0017] Furthermore, when a roadside warning signal is generated, the roadside computing unit sends the roadside warning signal to the vehicle-side control unit through the roadside V2X unit. If the vehicle-side control unit receives the roadside warning signal, it controls the unmanned vehicle to no longer enter the corresponding lane section Bi. The cloud management unit records the roadside warning signal of the lane section Bi and removes the lane section Bi to replan the route. When the warning evaluation value GL is less than the threshold GB, the cloud management unit re-includes the lane section Bi in the planned route. The vehicle-side control unit determines the coordinates of the unmanned vehicle through the UWB positioning unit and UWB base station.

[0018] Under normal circumstances, the roadside terminal can determine in advance whether the alleyway environment is passable. However, since the data collected by the roadside terminal has a periodicity, if environmental changes occur within the data collection period, a roadside early warning signal cannot be generated in time. Furthermore, if an unmanned vehicle has already entered or is about to enter the alleyway section Bi where the environment has changed, it is necessary to conduct a situation analysis in conjunction with the vehicle-side control unit. The specific details are as follows: When the vehicle-mounted LiDAR detects an obstacle on the unmanned vehicle's path, and the vehicle-mounted control unit does not receive a roadside warning signal, the vehicle-mounted computing host calls the vehicle-mounted perception system to collect lane image data. Then, the vehicle-mounted computing host segments the lane image data according to preset conditions to obtain abnormal data Zi, which includes a timestamp. The vehicle-mounted computing host generates a vehicle-mounted warning signal and calls the vehicle-mounted V2X unit to send the warning signal and abnormal data Zi to the roadside computing unit in the lane section Bi where the unmanned vehicle is located. The roadside computing unit then sends the warning signal and abnormal data Zi to the cloud management unit via a wireless communication network.

[0019] Furthermore, the steps for fusing sensing data of the internal environment of the mine roadway with abnormal data Zi include: S31, Recalculate parameters The roadside computing unit controls the roadside data acquisition unit to reacquire roadside point cloud data and calculate parameters. If the parameter If the value is greater than or equal to the threshold E, the roadside point cloud data and the abnormal data Zi will be fused for perception data; otherwise, the roadside point cloud data will not be considered. The vehicle computing host will analyze the images collected by the vehicle camera to determine whether driving can continue. S32. Perform data acquisition and synchronization processing: when parameters When the threshold E is greater than or equal to the threshold, the roadside computing unit uses the timestamp to align the roadside point cloud data and the abnormal data Zi; S33. Feature Extraction: The roadside computing unit uses the DBSCAN clustering algorithm to identify obstacle features from the roadside point cloud data and extracts the boundary points of the obstacles; the roadside computing unit uses the YOLO algorithm to identify obstacle features from the abnormal data Zi and obtains the bounding boxes of the obstacles. S34. Data Fusion: The point data in the roadside point cloud data after feature extraction is matched with the detection results of abnormal data Zi. The matching is verified by whether the point falls within the bounding box of the detected abnormal data Zi. Then, the roadside point cloud data and abnormal data Zi are fused using the Kalman filter method. S35. Calculate obstacle size and relative position: Calculate the minimum bounding box of the obstacle by using the boundary points of the obstacle, then calculate the coordinates of the center point of the obstacle, and obtain the coordinates of the unmanned vehicle in the global coordinate system through the UWB positioning unit. Convert the coordinates of the center point of the obstacle into coordinates in the global coordinate system, and calculate the Euclidean distance D from the center point of the obstacle to the unmanned vehicle.

[0020] To determine whether the autonomous vehicle can continue driving, the risk assessment parameter Rs is calculated. The specific steps are: calculate the product of the bounding box's length, width, and height to obtain the obstacle volume S; obtain the Euclidean distance D and the vehicle speed v; and obtain the parameters. The obstacle volume S, Euclidean distance D, and vehicle speed v are normalized, and the risk assessment parameter Rs is calculated. Where u1, u2, u3, u4 are weight parameters, and Rs is greater than 0 and less than 1.

[0021] The cloud management unit compares the risk assessment parameter Rs with the risk threshold, where risk threshold T1 is the low-risk threshold and risk threshold T2 is the high-risk threshold. If Rs is less than T1, the vehicle can pass safely. If Rs is greater than or equal to T1 and less than T2, the cloud management unit generates a low-speed travel command and sends it to the roadside computing unit via the wireless communication network. The roadside computing unit then sends the low-speed travel command to the onboard computing host via the roadside V2X unit, thereby controlling the speed of the unmanned vehicle. If Rs is greater than T2, the cloud management unit replans the route.

[0022] The beneficial effects of this invention are as follows: (1) The present invention divides the mine roadway into several sections. By setting up roadside data acquisition units and roadside calculation units in each section, the roadway environment data in that section is collected and processed. This allows for more accurate determination and prediction of whether the roadway environment is suitable for unmanned vehicles to pass through, providing a basis for route planning. (2) The present invention makes a judgment and prediction of the roadway environment in advance, which reduces the amount of data calculation of the vehicle-side control unit and is more conducive to the control of unmanned vehicles. The vehicle-side control unit can assist the roadside calculation unit to judge the driving environment in the roadway more timely. The judgment of the cloud management unit ensures the timeliness of the unmanned vehicle operation, reduces unnecessary route changes, and thus avoids the impact on production activities. Attached Figure Description

[0023] The invention will now be further described with reference to the accompanying drawings; Figure 1 This is a flowchart of the system of the present invention. Detailed Implementation

[0024] 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, and 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.

[0025] Example 1: This invention is a vehicle-road cooperative unmanned vehicle control system in roadways: It includes a roadside data acquisition unit, used to collect environmental perception data inside the mine roadways.

[0026] The roadside data acquisition unit includes a wireless communication network and a roadside environmental perception system; the wireless communication network of the roadside data acquisition unit includes a 4G / 5G network and a Wi-Fi network.

[0027] The roadside environmental perception system includes lidar, mining cameras, temperature and humidity sensors, infrared sensors, and UWB base stations. When the infrared sensor collects heat source data, the lidar will not collect data unless it receives other instructions, to avoid collecting personnel data and causing it to be mistakenly identified as an obstacle.

[0028] It also includes a roadside computing unit, which is used to receive and store environmental perception data inside the mine roadway, and calculate and form roadside point cloud data. The roadside early warning assessment value GL is calculated within a set time frequency. In this embodiment, the set time frequency is to collect data once every 6 hours. If the warning assessment value GL exceeds the threshold GB, a roadside warning signal is generated and sent to the vehicle control unit via the roadside V2X unit. The roadside warning assessment value GL is also sent to the cloud management unit via the wireless communication network.

[0029] The vehicle-mounted control unit is used to collect road perception data along the route of the mine roadway; the vehicle-mounted control unit includes a vehicle-mounted perception system, an on-board V2X unit, an on-board computing host, a wireless communication network, a UWB positioning unit, an inertial measurement unit, and a speed sensor unit; the vehicle-mounted perception system includes an on-board LiDAR and an on-board camera, used to collect road perception data along the route of the mine roadway; the wireless communication network of the vehicle-mounted control unit includes 4G / 5G network and Wi-Fi network.

[0030] The vehicle-mounted control unit assesses whether abnormal data Zi exists. If abnormal data Zi is found, a vehicle-mounted warning signal is generated and sent to the roadside computing unit via the vehicle-mounted V2X unit. The roadside computing unit controls the roadside environmental perception system to collect environmental perception data inside the mine roadway through the wireless communication network, and performs perception data fusion by combining the abnormal data Zi. The roadside computing unit calculates the risk assessment parameter Rs and sends the risk assessment parameter Rs to the cloud management unit through the wireless communication network. The cloud management unit determines whether the unmanned vehicle should replan its route based on the risk assessment parameter Rs.

[0031] The cloud management unit is used for data storage and analysis, and for controlling unmanned vehicles; the roadside computing unit periodically sends the data it stores to the cloud management unit via a wireless communication network. The cloud management unit plans the route of the unmanned vehicle based on the stored historical data, generates the route instructions, and sends the route instructions to the roadside computing unit through the wireless communication network. The roadside computing unit sends the instructions to the vehicle control unit through the roadside V2X unit, and the vehicle control unit controls the operation of the unmanned vehicle. The historical data includes, but is not limited to, environmental perception data inside the mine roadway, roadside point cloud data, abnormal data Zi, roadside early warning assessment value GL, and risk assessment parameter Rs.

[0032] The cloud management unit divides the tunnel map data into several tunnel intervals Bi, i=1~n, where n is a positive integer. Each tunnel interval Bi is equipped with a roadside data acquisition unit and a roadside calculation unit. Each roadside calculation unit in tunnel interval Bi is assigned a number corresponding to the coordinates of that tunnel interval Bi. For example, the roadside calculation unit installed in tunnel interval B1 is numbered SE001, the roadside calculation unit installed in tunnel interval B2 is numbered SE002, and so on. This number is manually set, and each roadside calculation unit has a unique number. The number of the roadside calculation unit serves as the basis for the cloud management unit to identify and store the tunnel interval Bi data. When the roadside calculation unit and the cloud management unit exchange data, the data includes the number of the roadside calculation unit. The cloud management unit first identifies the number data and then maps it to the specific coordinate interval of tunnel interval Bi. The coordinate interval of tunnel interval Bi is located by multiple sets of UWB base stations installed in the mine tunnel. Generally, the UWB base stations calculate the precise coordinates using triangulation and obtain the global coordinate system through the UWB base stations.

[0033] The roadside data acquisition unit communicates with the roadside computing unit via a wireless communication network; the roadside computing unit communicates with the cloud management unit via a wireless communication network.

[0034] To assess the condition of the tunnel walls and their associated equipment, roadside point cloud data was calculated. The specific steps include: S11. Raw point cloud data acquisition: Install several lidars in each tunnel section Bi so that the lidar scanning range can cover the road, inner wall and inner wall auxiliary equipment of the tunnel section Bi in which it is located; after the lidar scan, the raw point cloud data is obtained, which includes the three-dimensional coordinates and timestamp of each point. S12. Perform data preprocessing: Calculate the mean neighborhood distance of each point, remove points exceeding a preset threshold, and then divide the remaining point cloud data into a 3D grid. Each 3D grid is represented by a representative point, which is the centroid of the 3D grid. The centroid calculation formula is: C is the centroid, n is the number of points in the 3D network, and Xi is the coordinates of the point. Then, the coordinates of the representative points are transformed to obtain preprocessed point cloud data. This transformation uses a 3D rotation matrix R and a translation vector t for coordinate system transformation, where Xa = RX + t, X is the coordinates of the representative point, Xa is the coordinates of the transformed point cloud data, R is the rotation matrix, and t is the translation vector. For example, the rotation matrix for rotating around the z-axis by an angle θ is... A rotation matrix can represent rotations about the x, y, and z axes.

[0035] S13. Registration of preprocessed point cloud data: Align the preprocessed point cloud data of different frames using the RANSAC algorithm to obtain coarse point cloud data; S14. Registration of coarse point cloud data: The coarse point cloud data is precisely aligned using the ICP algorithm to obtain the point cloud data. Through iterative optimization, the rotation matrix can be accurately calculated. S15. Feature Extraction: The normal vector of the point cloud surface is calculated by the normal estimation method to obtain the simulated geometric structure of the road in the lane section Bi, and the simulated geometric structure of the inner wall and the inner wall auxiliary equipment of the lane section Bi. S16. Data storage: The generated point cloud data is stored in the roadside computing unit, that is, the point cloud data generated in lane section B1 is stored in the roadside computing unit of lane section B1, and so on; the roadside computing unit packages the point cloud data together with the identification data of lane section Bi in which the roadside computing unit is located, and sends the packaged point cloud data to the cloud management unit through the wireless communication network; the identification data of lane section Bi is the coordinate range of lane section Bi.

[0036] To determine whether there are obstacles or obstacle risks in the alleyway environment, the roadside early warning assessment value GL is calculated using point cloud data. The specific steps include: S21. Calculate abnormal point cloud data: The roadside computing unit acquires point cloud data, extracts point cloud data of the road area, and records it as road point cloud data. Set filter condition K, where filter condition K is a coordinate value, to filter extreme values ​​of point cloud coordinate data that are higher or lower than the road surface. This extreme value is set by technicians according to the actual situation. Count the road point cloud data that meet the condition to obtain the first abnormal data M. The roadside computing unit extracts point cloud data of the inner wall and its auxiliary equipment, and records it as inner wall point cloud data. Set filter condition L, where filter condition L is a coordinate value, to filter extreme values ​​of point cloud coordinate data that protrude from the inner wall of the alley. This extreme value is set by technicians according to the actual situation. Count the inner wall point cloud data that meet the condition to obtain the second abnormal data N. S22, Calculation parameters Calculate the ratio of the first abnormal data point M to the set threshold m, calculate the ratio of the second abnormal data point N to the set threshold n, and calculate the parameters. ; S23, Calculation Parameters The roadside calculation unit records the number P of roadside early warning signals generated in its respective lane section Bi, and calculates the parameters. ,in It is a historical data weighting coefficient, which is taken in this embodiment. It is 1.22; S24, Calculation Parameters The roadside computing unit acquires temperature data T and humidity data H collected by the roadside data acquisition unit, and calculates... ,in This represents the historical minimum temperature within the Bi section of the roadway. This indicates the historical maximum temperature within section Bi of the tunnel. This represents the historical minimum humidity value within the Bi section of the tunnel. This indicates the historical maximum humidity value within the Bi section of the tunnel. These are the weighting coefficients for the temperature data. In this embodiment, the humidity data weighting coefficient is used. Take 1.02, Take 1.36; S25, Calculate the roadside early warning assessment value GL: If the warning assessment value GL is greater than or equal to the threshold GB, a roadside warning signal is generated; otherwise, no roadside warning signal is generated. In this embodiment, the value of GB is 11.56.

[0037] If the warning assessment value GL is less than the threshold GB, no roadside warning signal will be generated, and the unmanned vehicle will continue to travel along the original route. If the warning assessment value GL is greater than or equal to the threshold GB, a roadside warning signal is generated. After the roadside warning signal is generated, the roadside calculation unit sends the roadside warning signal to the vehicle-mounted control unit through the roadside V2X unit. If the vehicle-mounted control unit receives the roadside warning signal, it controls the unmanned vehicle to no longer enter the corresponding lane section Bi. The cloud management unit records the roadside warning signal of the lane section Bi and removes the lane section Bi to replan the route until the warning assessment value GL is less than the threshold GB. Then, the cloud management unit re-includes the lane section Bi in the planned route. The vehicle-mounted control unit determines the coordinates of the unmanned vehicle through the UWB positioning unit and the UWB base station.

[0038] This embodiment divides the mine roadway into several sections. By setting up roadside data acquisition units and roadside computing units in each section, the roadway environment data in that section is collected and processed. This allows for a more accurate determination and prediction of whether the roadway environment is suitable for unmanned vehicles, providing a basis for route planning. By determining and predicting the roadway environment in advance, the amount of data processing required by the vehicle-side control unit is reduced, making it easier to control unmanned vehicles.

[0039] Example 2: The difference between this embodiment and embodiment one is that: if the warning evaluation value GL is less than the threshold GB, no roadside warning signal is generated, and the unmanned vehicle continues to travel along the original route. When the unmanned vehicle approaches or enters a certain lane section Bi, the vehicle-side control unit of the unmanned vehicle detects abnormal data.

[0040] Under normal circumstances, the roadside terminal can determine in advance whether the alleyway environment is passable. However, since the data collected by the roadside terminal has a periodicity, if environmental changes occur within the data collection period, a roadside early warning signal cannot be generated in time. Furthermore, if an unmanned vehicle has already entered or is about to enter the alleyway section Bi where the environment has changed, it is necessary to conduct a situation analysis in conjunction with the vehicle-side control unit. The specific details are as follows: When the vehicle-mounted LiDAR detects an obstacle on the unmanned vehicle's route and the vehicle-mounted control unit does not receive a roadside warning signal, the vehicle-mounted computing host calls the vehicle-mounted perception system to collect alleyway image data. Then, the vehicle-mounted computing host segments the alleyway image data according to preset conditions to obtain abnormal data Zi, which includes a timestamp. The vehicle-mounted computing host then segments the alleyway inner wall and its auxiliary equipment in the alleyway image data by training the YOLO algorithm, and the remaining alleyway image data is used as abnormal data Zi.

[0041] The onboard computing host generates a vehicle-side warning signal and calls the onboard V2X unit to send the vehicle-side warning signal and abnormal data Zi to the roadside computing unit in the lane section Bi where the unmanned vehicle is located. The roadside computing unit then sends the vehicle-side warning signal and abnormal data Zi to the cloud management unit via a wireless communication network.

[0042] The steps for fusing sensing data of the inner side of the mine roadway environment with abnormal data Zi include: S31, Recalculate parameters When the unmanned vehicle follows the set route, the coordinates of the unmanned vehicle are determined by the onboard UWB positioning unit and UWB base station, which can then determine which lane section Bi the unmanned vehicle is located in. When the cloud management unit receives the vehicle-side warning signal and abnormal data Zi, it sends the roadside point cloud data collection command to the roadside computing unit in the lane section Bi where the unmanned vehicle is located and the roadside computing unit in the next lane section Bi in the direction of the unmanned vehicle's travel route via the wireless communication network.

[0043] For example, if the preset route for an unmanned vehicle is to pass through lane sections B1 to B5 sequentially, when the unmanned vehicle enters lane section B2, the onboard LiDAR detects an obstacle in the vehicle's path. Since the vehicle's control unit does not receive a roadside warning signal, the onboard computing unit generates a warning signal and calls the onboard V2X unit to send the warning signal and abnormal data Zi to the roadside computing unit within lane section B2. The roadside computing unit in lane section B2 then transmits the warning signal and abnormal data Zi to the cloud management unit via a wireless communication network. Upon receiving the warning signal and abnormal data Zi, the cloud management unit sends a roadside point cloud data acquisition command to the roadside computing units in lane sections B2 and B3 via a wireless communication network, and collects roadside point cloud data respectively, then calculates the parameters accordingly. .

[0044] The roadside computing unit controls the roadside data acquisition unit to reacquire roadside point cloud data and calculate parameters. If the parameter If the value is greater than or equal to the threshold E, and in this embodiment the threshold E is 1.43, then the roadside point cloud data and the abnormal data Zi are fused for perception data; otherwise, the roadside point cloud data is not considered. The vehicle computing host determines whether driving can continue by analyzing the images collected by the vehicle camera. S32. Perform data acquisition and synchronization processing: when parameters When the threshold E is greater than or equal to the threshold, the roadside computing unit uses the timestamp to align the roadside point cloud data and the abnormal data Zi; S33. Feature Extraction: The roadside computing unit uses the DBSCAN clustering algorithm to identify obstacle features from the roadside point cloud data and extracts the boundary points of the obstacles; the roadside computing unit uses the YOLO algorithm to identify obstacle features from the abnormal data Zi and obtains the bounding boxes of the obstacles. S34. Data Fusion: The point data in the roadside point cloud data after feature extraction is matched with the detection results of abnormal data Zi. The matching is verified by whether the point falls within the bounding box of the detected abnormal data Zi. Then, the roadside point cloud data and abnormal data Zi are fused using the Kalman filter method. S35. Calculate the size and relative position of obstacles: Calculate the minimum bounding box of the obstacle through the boundary points of the obstacle, then calculate the coordinates of the center point of the obstacle, and obtain the coordinates of the unmanned vehicle in the global coordinate system through the UWB positioning unit. Convert the coordinates of the center point of the obstacle into coordinates in the global coordinate system, and calculate the Euclidean distance D from the center point of the obstacle to the unmanned vehicle. The global coordinate system is defined by arranging multiple UWB base stations (anchor points) at known fixed locations, and the origin and axis of the global coordinate system are defined by the locations of the UWB base stations. The coordinates (Xr, Yr, Zr) of the unmanned vehicle in the global coordinate system are obtained through UWB positioning. The point cloud coordinates (Xa, Ya, Za) of the obstacle in the lidar coordinate system are obtained. The position and deflection angle within the tunnel section Bi are known. We can obtain the coordinates (Xb, Yb, Zb) of the lidar in the global coordinate system, transform the point cloud coordinates (Xa, Ya, Za) into global coordinates (Xg, Yg, Zg), and calculate the rotation transformation (assuming only rotation around the Z-axis): Then perform a translation transformation: ; Calculate Euclidean distance .

[0045] To determine whether the unmanned vehicle can continue driving, the risk assessment parameter Rs is calculated. The specific steps are as follows: calculate the product of the length, width and height of the bounding box to obtain the obstacle volume S, obtain the Euclidean distance D and the vehicle speed v, and obtain the vehicle speed through the speed sensor unit.

[0046] Get parameters The obstacle volume S, Euclidean distance D, and vehicle speed v are normalized, and the risk assessment parameter Rs is calculated. Where u1, u2, u3, u4 are weight parameters, and Rs is greater than 0 and less than 1.

[0047] The cloud management unit compares the risk assessment parameter Rs with the risk threshold, where risk threshold T1 is the low-risk threshold and risk threshold T2 is the high-risk threshold. If Rs is less than T1, the vehicle can pass safely. If Rs is greater than or equal to T1 and less than T2, the cloud management unit generates a low-speed travel command and sends it to the roadside computing unit via the wireless communication network. The roadside computing unit then sends the low-speed travel command to the onboard computing host via the roadside V2X unit, thereby controlling the unmanned vehicle's travel speed and position. If Rs is greater than T2, the cloud management unit replans the route.

[0048] In this embodiment, the vehicle-mounted control unit can assist the roadside computing unit in determining the driving environment in the alley more promptly. The determination by the cloud management unit ensures the timeliness of unmanned vehicle operation, reduces unnecessary route changes, and thus avoids the impact on production activities.

[0049] The threshold is set to facilitate comparison. The size of the threshold depends on the amount of sample data and the number of bases set by those skilled in the art for each set of sample data; as long as it does not affect the ratio between the parameter and the quantized value, it is acceptable.

[0050] The above formulas are all derived from software simulation using a large amount of data and are selected to be close to the actual values. The coefficients in the formulas are set by those skilled in the art according to the actual situation. The above description is only a preferred embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any equivalent substitutions or changes made by those skilled in the art within the technical scope disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the protection scope of the present invention.

Claims

1. A vehicle-road cooperative unmanned vehicle control system in a lane, characterized in that, It includes a roadside data acquisition unit, used to collect environmental perception data inside the mine roadway; It also includes a roadside computing unit, which is used to receive environmental perception data inside the mine roadway and calculate roadside point cloud data. Within a set time frequency, it calculates the roadside early warning assessment value GL. If the early warning assessment value GL exceeds the threshold GB, a roadside early warning signal is generated and sent to the vehicle-side control unit through the roadside V2X unit. The roadside early warning assessment value GL is sent to the cloud management unit through the wireless communication network. The vehicle-mounted control unit is used to collect road perception data along the route of the mine roadway and assess whether there is abnormal data Zi. If abnormal data Zi is found, a vehicle-mounted warning signal is generated and sent to the roadside computing unit through the vehicle-mounted V2X unit. The roadside computing unit controls the roadside environmental perception system to collect environmental perception data inside the mine roadway through the wireless communication network, and performs perception data fusion by combining the abnormal data Zi. The roadside computing unit calculates the risk assessment parameter Rs and sends the risk assessment parameter Rs to the cloud management unit through the wireless communication network. The cloud management unit determines whether the unmanned vehicle should replan its route based on the risk assessment parameter Rs. The cloud management unit is used for data storage and analysis, as well as for controlling the unmanned vehicles; The cloud management unit plans the route of the unmanned vehicle based on the stored historical data, generates the route instructions, and sends the route instructions to the roadside computing unit through the wireless communication network. The roadside computing unit sends the instructions to the vehicle control unit through the roadside V2X unit, and the vehicle control unit controls the operation of the unmanned vehicle. The calculation steps for the roadside warning assessment value GL include: S21. Calculate abnormal point cloud data: The roadside computing unit acquires point cloud data, extracts point cloud data of the road area, records it as road point cloud data, sets filter condition K, counts the road point cloud data that meet the condition, and obtains the first abnormal data M; The roadside computing unit extracts point cloud data of the inner wall and inner wall auxiliary equipment, records it as inner wall point cloud data, sets filter condition L, counts the inner wall point cloud data that meet the condition, and obtains the second abnormal data N; S22, Calculation parameters Calculate the ratio of the first abnormal data point M to the set threshold m, calculate the ratio of the second abnormal data point N to the set threshold n, and calculate the parameters. ; S23, Calculation Parameters The roadside calculation unit records the number P of roadside early warning signals generated in its respective lane section, and calculates parameters. ,in These are historical data weighting coefficients; S24, Calculation Parameters The roadside computing unit acquires temperature data T and humidity data H collected by the roadside data acquisition unit, and calculates... ,in This represents the historical minimum temperature within the Bi section of the roadway. This indicates the historical maximum temperature within section Bi of the tunnel. This represents the historical minimum humidity value within the Bi section of the tunnel. This indicates the historical maximum humidity value within the Bi section of the tunnel. These are the weighting coefficients for the temperature data. These are the weighting coefficients for humidity data; S25, Calculate the roadside early warning assessment value GL: If the warning assessment value GL is greater than or equal to the threshold GB, a roadside warning signal will be generated; otherwise, no roadside warning signal will be generated. The roadside computing unit sends the roadside warning signal to the vehicle-side control unit through the roadside V2X unit. If the vehicle-side control unit receives the roadside warning signal, it controls the unmanned vehicle to no longer enter the corresponding lane section Bi. The cloud management unit records the roadside warning signal of the lane section Bi and removes the lane section Bi to replan the route. When the warning evaluation value GL is less than the threshold GB, the cloud management unit re-includes the lane section Bi in the planned route. The roadside unit determines in advance whether the alleyway environment is passable. If environmental changes occur during the data collection period and a roadside warning signal cannot be generated in time, and the unmanned vehicle has already entered or is about to enter the alleyway section Bi where the environment has changed, the following steps are taken: When the vehicle-mounted LiDAR detects an obstacle on the unmanned vehicle's travel route, and the vehicle-mounted control unit does not receive a roadside warning signal, the vehicle-mounted computing host calls the vehicle-mounted perception system to collect alleyway image data. Then, the vehicle-mounted computing host segments the alleyway image data according to preset conditions to obtain abnormal data Zi, which includes a timestamp. The vehicle-mounted computing host generates a vehicle-mounted warning signal and calls the vehicle-mounted V2X unit to send the vehicle-mounted warning signal and abnormal data Zi to the roadside computing unit in the alleyway section Bi where the unmanned vehicle is located. The roadside computing unit sends the vehicle-mounted warning signal and abnormal data Zi to the cloud management unit through a wireless communication network.

2. The vehicle-road cooperative unmanned vehicle control system in a lane according to claim 1, characterized in that, The cloud management unit divides the lane map data into several lane intervals Bi, and each lane interval Bi is equipped with a roadside data acquisition unit and a roadside calculation unit. The roadside data acquisition unit includes a wireless communication network and a roadside environmental perception system. The roadside environmental perception system includes a lidar, a mining camera, a temperature and humidity sensor, an infrared sensor, and a UWB base station. When the infrared sensor acquires heat source data, the lidar does not acquire data. The roadside data acquisition unit communicates with the roadside computing unit via a wireless communication network; the roadside computing unit communicates with the cloud management unit via a wireless communication network.

3. The vehicle-road cooperative unmanned vehicle control system in a lane according to claim 2, characterized in that, The calculation steps for the roadside point cloud data include: S11. Raw point cloud data acquisition: Install several lidars in each of the tunnel sections Bi, so that the lidar scanning range can cover the road, inner wall and inner wall auxiliary equipment of the tunnel section Bi in which it is located; after the lidar scan, raw point cloud data is obtained, which includes the three-dimensional coordinates and timestamp of each point. S12. Perform data preprocessing: Calculate the average neighborhood distance of each point, remove points that exceed a preset threshold, and then divide the remaining point cloud data into a three-dimensional grid. Each three-dimensional grid is represented by a representative point, which is the centroid of the three-dimensional grid. Then, transform the coordinates of the representative point to obtain the preprocessed point cloud data. S13. Registration of preprocessed point cloud data: Align the preprocessed point cloud data of different frames using the RANSAC algorithm to obtain coarse point cloud data; S14. Registration of coarse point cloud data: The coarse point cloud data is precisely aligned using the ICP algorithm to obtain the point cloud data. S15. Feature Extraction: The normal vector of the point cloud surface is calculated by the normal estimation method to obtain the simulated geometric structure of the road, inner wall and inner wall auxiliary equipment in the lane section Bi. S16. Data storage: The generated point cloud data is stored in the roadside computing unit. The roadside computing unit packages the point cloud data together with the identification data of the lane section Bi where the roadside computing unit is located, and sends the packaged point cloud data to the cloud management unit through the wireless communication network.

4. The vehicle-road cooperative unmanned vehicle control system in a lane according to claim 2, characterized in that, The vehicle-mounted control unit includes a vehicle-mounted perception system, an on-board V2X unit, an on-board computing host, a wireless communication network, a UWB positioning unit, an inertial measurement unit, and a speed sensor unit; the vehicle-mounted perception system includes an on-board LiDAR and an on-board camera, used to collect road perception data along the travel route in the mine tunnels.

5. The vehicle-road cooperative unmanned vehicle control system in a lane according to claim 4, characterized in that, When the vehicle-mounted lidar detects an obstacle on the unmanned vehicle's path and the vehicle-mounted control unit does not receive a roadside warning signal, the vehicle-mounted computing host calls the vehicle-mounted perception system to collect lane image data. Then, the vehicle-mounted computing host segments the lane image data according to preset conditions to obtain abnormal data Zi, which includes a timestamp. The vehicle-mounted computing host generates a vehicle-mounted warning signal and calls the vehicle-mounted V2X unit to send the vehicle-mounted warning signal and abnormal data Zi to the roadside computing unit in the lane section Bi where the unmanned vehicle is located.

6. The vehicle-road cooperative unmanned vehicle control system in a lane according to claim 5, characterized in that, The steps for fusing the sensing data of the inner side of the mine roadway environment with the abnormal data Zi include: S31, Recalculate parameters The roadside computing unit controls the roadside data acquisition unit to reacquire roadside point cloud data and calculate parameters. If the parameter If the value is greater than or equal to the threshold E, the roadside point cloud data and the abnormal data Zi will be fused for perception data; otherwise, the roadside point cloud data will not be considered. The vehicle computing host will analyze the images collected by the vehicle camera to determine whether driving can continue. S32. Perform data acquisition and synchronization processing: when parameters When the threshold E is greater than or equal to the threshold, the roadside computing unit uses the timestamp to align the roadside point cloud data and the abnormal data Zi; S33. Feature Extraction: The roadside computing unit uses the DBSCAN clustering algorithm to identify obstacle features from the roadside point cloud data and extracts the boundary points of the obstacles; the roadside computing unit uses the YOLO algorithm to identify obstacle features from the abnormal data Zi and obtains the bounding boxes of the obstacles. S34. Data Fusion: The point data in the roadside point cloud data after feature extraction is matched with the detection results of abnormal data Zi. The matching is verified by whether the point falls within the bounding box of the detected abnormal data Zi. Then, the roadside point cloud data and abnormal data Zi are fused using the Kalman filter method. S35. Calculate obstacle size and relative position: Calculate the minimum bounding box of the obstacle through the boundary points of the obstacle, then calculate the coordinates of the center point of the obstacle, and obtain the coordinates of the unmanned vehicle in the global coordinate system through the UWB positioning unit. Convert the coordinates of the center point of the obstacle into coordinates in the global coordinate system, and calculate the Euclidean distance D from the center point of the obstacle to the unmanned vehicle.

7. The vehicle-road cooperative unmanned vehicle control system in a lane according to claim 6, characterized in that, The calculation steps for the risk assessment parameter Rs are as follows: The obstacle volume S is obtained by calculating the product of the bounding box's length, width, and height. The Euclidean distance D and vehicle speed v are then obtained, along with other parameters. The obstacle volume S, Euclidean distance D, and vehicle speed v are normalized, and the risk assessment parameter Rs is calculated. Where u1, u2, u3, u4 are weight parameters, and Rs is greater than 0 and less than 1.

8. The vehicle-road cooperative unmanned vehicle control system in a lane according to claim 7, characterized in that, The cloud management unit compares the risk assessment parameter Rs with the risk threshold, where risk threshold T1 is the low-risk threshold and risk threshold T2 is the high-risk threshold. If Rs is less than T1, the vehicle can pass safely. If Rs is greater than or equal to T1 and less than T2, the cloud management unit generates a low-speed travel command and sends it to the roadside computing unit via the wireless communication network. The roadside computing unit then sends the low-speed travel command to the on-board computing host via the roadside V2X unit. If Rs is greater than T2, the cloud management unit replans the route.

Images