A real vehicle-in-the-loop simulation test platform based on vehicle-road-cloud collaboration

Abstract

The present invention discloses a real vehicle in-loop simulation test platform based on vehicle-road-cloud collaboration, which belongs to the field of autonomous driving technology. The present invention includes a smart road terminal, an intelligent networked vehicle terminal, a simulation cloud terminal and a communication service terminal. The smart road terminal uploads traffic flow data to the simulation cloud terminal through the communication service terminal, and the intelligent networked vehicle terminal uploads its positioning information to the simulation cloud terminal through the communication service terminal. The simulation cloud terminal quickly simulates interference factors such as different virtual traffic flow elements based on real scenes. The simulation cloud terminal uses the intelligent networked vehicle twin algorithm to make correct decisions based on real-time traffic flow data and transmits it to the intelligent networked vehicle terminal in real time to drive it to make corresponding movements. The intelligent networked vehicle terminal updates its own posture and positioning information, and then uploads it to the simulation cloud terminal in real time. The present invention constructs a closed-loop test system combining intelligent networked vehicles and hybrid environments. The present invention can be used as a safer and more effective testing technology before the road side test of an autonomous driving vehicle.

Description

Technical Field

[0001] The present invention belongs to the technical field of autonomous driving, and specifically relates to a real vehicle-in-the-loop simulation test platform based on vehicle-road-cloud collaboration. Background Art

[0002] The testing and verification methods of autonomous driving are crucial in the future. At present, the testing of autonomous driving systems is mainly based on real road tests and simulation tests. The advantage of traditional road tests is that the traffic environment is real, but the diversity of test scenarios is limited, and it is impossible to customize edge test scenarios safely and efficiently; simulation tests are flexible and efficient, but lack a real traffic flow test environment, which breaks the strong coupling relationship between the vehicle and the environment in actual application scenarios. In order to solve these problems, real vehicle-in-the-Loop (ViL) testing is used as an effective verification method before road testing. However, the current ViL system is complex and relies heavily on expensive real vehicle bench systems or vehicles with complete autonomous driving functions, and it is impossible to apply real-time traffic flow data on the road to the simulation test environment. Therefore, it is of practical significance to develop a low-cost, high-efficiency, and high-reliability test method for autonomous driving systems. Summary of the invention

[0003] The present invention provides a real vehicle-in-the-loop simulation test platform based on vehicle-road-cloud collaboration to improve the safety and effectiveness of road testing of autonomous vehicles.

[0004] The technical solution adopted by the present invention is:

[0005] A real vehicle-in-the-loop simulation test platform based on vehicle-road-cloud collaboration, the platform includes: a smart road end, an intelligent networked vehicle end, a simulation cloud end and a communication service end;

[0006] The intelligent networked vehicle end includes a combined inertial navigation module, a control module and a communication module, wherein the combined inertial navigation module is used for positioning the intelligent networked vehicle end and obtaining the vehicle end positioning message; the control module is used to control the wire-controlled chassis of the intelligent networked vehicle end to drive the vehicle movement; the communication module is used to interact with the communication service end for data, including: the intelligent networked vehicle end sends the vehicle end positioning message to the smart road end through the communication service end, and receives the decision information from the simulation cloud through the communication service end;

[0007] The smart road terminal includes a perception unit, a positioning module and a communication module, wherein the perception unit includes a roadside camera, an identification module, a screening module and a tracking module, the roadside camera is used to collect road image data and send it to the identification module, the identification module is used to identify the specified traffic element object, and send the identification result to the screening module, the screening module selects the traffic element object within the lane range and sends the screening result to the tracking module, the tracking module performs multi-target tracking on the screened traffic element object, and sends the pixel point information of the tracked target to the positioning module in real time, the positioning module converts the representative pixel point of each target into physical coordinate information based on the configured perspective transformation matrix, obtains the perception information of each target based on the category information and physical coordinate information of the target, that is, the first traffic flow information; and deletes the perception information corresponding to the vehicle end in the first traffic information to obtain the second traffic information; sends the first and second traffic information to the communication service end through the communication module of the smart road terminal, and the communication service end forwards the first and second traffic information to the simulation cloud;

[0008] The simulation cloud includes an online traffic flow simulation module, a real vehicle twin decision module and a communication module. The online traffic flow simulation module is used to visualize the first traffic flow information from the smart road end; the real vehicle twin decision module obtains the decision information under the current traffic scenario based on the configured decision algorithm to be tested and the second traffic flow information from the smart road end, and sends it to the communication server through the communication module of the simulation cloud; the communication server forwards the decision information to the intelligent connected vehicle end, and the communication module of the intelligent connected vehicle end parses the decision information and sends it to the control module of the intelligent connected vehicle end to drive the intelligent connected vehicle end to make corresponding movements.

[0009] Furthermore, the perspective transformation matrix of the positioning module of the intelligent road terminal is specifically set as follows:

[0010] Determine the traffic flow identification and positioning area;

[0011] The pixel information and coordinate information of the four corner points of the traffic flow identification and positioning area are collected, and a transformation matrix from pixel information to coordinate information is obtained through perspective transformation to obtain a perspective transformation matrix.

[0012] Furthermore, the traffic flow online simulation module in the simulation cloud is also used to: add virtual interference elements based on real traffic information and visualize them, and send the added virtual interference elements to the real vehicle twin decision module.

[0013] Furthermore, the traffic flow online simulation module of the simulation cloud visualizes the first traffic flow information from the smart road end including: simulation and visualization of static roads, static buildings and dynamic traffic flow scenes;

[0014] The modeling of static roads includes: collecting GPS geographical location information of roads, converting the geographical location information of roads into plane coordinates through GPS data conversion tools, and obtaining roads modeled based on real GPS data;

[0015] The modeling of static buildings includes: collecting real image textures of buildings within the driving visual range, performing 1:1 modeling based on CAD drawings, and mapping the constructed building models at corresponding locations;

[0016] Dynamic traffic flow scene modeling includes: generating a traffic flow model based on first traffic flow information and controlling its movement.

[0017] The technical solution provided by the present invention brings at least the following beneficial effects:

[0018] Compared with the existing real vehicle bench-in-the-loop and complete autonomous driving real vehicle-in-the-loop simulation test solutions, the advantages of the present invention are:

[0019] 1) Introduce real-time traffic flow data to make up for the difference between the simulation test environment and the actual environment.

[0020] 2) Add a real car to the test loop to ensure a strong coupling relationship between the real car and the environment.

[0021] 3) By building a closed-loop test environment that combines real vehicles and virtual reality, the testing capabilities for extreme scenarios are improved while ensuring the validity of the test results. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0023] Figure 1 It is a schematic diagram of the principle of a real vehicle-in-the-loop simulation test platform based on vehicle-road-cloud collaboration provided by an embodiment of the present invention;

[0024] Figure 2 1 is a schematic diagram of a processing process of a perception unit of a road edge strong real-time perception server in an embodiment of the present invention;

[0025] Figure 3 is a schematic diagram of a simulation processing process in a simulation cloud in an embodiment of the present invention;

[0026] Figure 4 is a schematic diagram of static scene modeling in an embodiment of the present invention;

[0027] Figure 5It is an online simulation process of simulating dynamic traffic flow in the cloud in an embodiment of the present invention.

[0028] Figure 6 This is an online simulation rendering effect diagram in an embodiment of the present invention.

[0029] Figure 7 It is a structural diagram of a real vehicle-in-the-loop simulation test platform based on vehicle-road-cloud collaboration provided by an embodiment of the present invention.

[0030] Figure 8 It is a schematic diagram of vehicle positioning correction in an embodiment of the present invention. DETAILED DESCRIPTION

[0031] In order to make the objectives, technical solutions and advantages of the present invention more clear, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0032] The embodiments of the present invention address the difference between the autonomous driving simulation test environment and the real traffic environment. By introducing real traffic flow information to make up for the lack of authenticity of the simulation test scene, a real-car in-the-loop simulation test platform based on vehicle-road-cloud collaboration is proposed. On the roadside, the embodiments of the present invention extract real-time traffic flow elements from the road and apply real traffic flow behavior information to the autonomous driving simulation test. On the cloud platform, the embodiments of the present invention set up a software simulation platform based on OpenSceneGraph, which can quickly simulate different weather, lighting, virtual traffic flow elements and other interference factors based on real scenes. On the intelligent connected real car, the embodiments of the present invention construct a closed-loop test system that combines intelligent connected vehicles and hybrid environments. The real-car in-the-loop simulation test platform based on vehicle-road-cloud collaboration provided by the embodiments of the present invention can be used as a safer and more effective testing technology before road testing of autonomous driving vehicles.

[0033] The real-vehicle-in-the-loop simulation test platform based on vehicle-road-cloud collaboration provided by the embodiment of the present invention is a real-vehicle-in-the-loop twin simulation system based on real-time traffic flow. The platform includes a strong real-time perception server at the road edge (referred to as the smart road end), an intelligent connected vehicle, a simulation cloud end, and a communication service end. Among them, the smart road end mainly performs beyond-line-of-sight perception through the road end equipment, and uploads the perceived traffic flow data to the simulation cloud end in real time. On the one hand, the smart road end can liberate the perception system of the intelligent connected vehicle and reduce the internal computing requirements of the vehicle end. On the other hand, it can increase the road perception range of the intelligent connected vehicle, making the decision result more reasonable. The intelligent connected vehicle end is the main terminal equipment of the road. It interacts with the smart road end and the simulation cloud end in real time through the high-speed network according to the corresponding data interface specifications. The simulation cloud end performs fusion processing based on the data of the intelligent connected vehicle end and the smart road end. On the one hand, it completes the visual simulation, and on the other hand, it performs decision planning for the intelligent connected vehicle end. Figure 1 shown.

[0034] (1) Road-side edge strong real-time perception server.

[0035] Since the perception distance of vehicles is limited, smart road terminals can decouple the perception system of autonomous vehicles on the one hand, and provide beyond-visual-range perception effects for autonomous vehicles on the other. Specifically, the embodiments of the present invention are based on the real-time traffic flow perception method of the roadside monocular camera: using the fixed posture advantage of the roadside camera, based on the traffic flow pictures collected by the roadside camera, multi-target recognition and tracking are completed through deep learning-based target recognition and tracking, such as YOLOv5 and DeepSort. At the same time, the perspective transformation of pixel coordinates into physical coordinates is used to complete the positioning of traffic flow elements, such as Figure 2 As shown, that is, firstly, a roadside image is acquired through a roadside camera, and the image is preprocessed (such as resizing to match the input of a target recognition network (such as YOLOv5 target recognition)) and then input into the target recognition network, and the recognition result is screened, and traffic element objects within the lane range are selected. The tracking module tracks the screened traffic element objects, and then multi-target tracking is realized. The perspective transformation of pixel coordinates into physical coordinates is used to complete the positioning of traffic flow elements, so as to obtain multi-target categories and trajectories.

[0036] (2)Simulation cloud.

[0037] In the embodiment of the present invention, the simulation cloud provided adopts a real-time simulation system based on traffic flow twins to realize online simulation of traffic flow and real vehicle twin decision-making. The consistency between the autonomous driving simulation test environment and the real environment largely determines the reliability of the simulation test results. The mapping of virtual traffic elements and the behavior of actual road traffic participants in the simulation test is the key.

[0038] Specifically, the specific processing process of the simulation cloud provided by the embodiment of the present invention is as follows: Figure 3As shown, in the process of constructing the virtual scene, it is mainly divided into static roads, buildings and dynamic traffic flow scene simulation. The construction of static roads includes: collecting actual road data (geographic location information of the road), processing the road data, and obtaining the road modeled by real GPS data. In this embodiment, the static road is driven once on a real road by a vehicle equipped with high-precision positioning, recording the route geographic location information (latitude and longitude altitude points), and converting the recorded route geographic location information into plane coordinates through GPS data conversion software, and finally saving it in a specified file. In this embodiment, it is saved in a CSV file, and finally a road modeled based on real GPS data is obtained. Static building modeling is based on a complete 1:1 modeling based on CAD drawings, and by collecting real image textures of the main areas of the building (within the driving visual range), and completing the building model mapping, the overall modeling diagram is shown as follows. Figure 4 shown.

[0039] The dynamic traffic flow scene modeling data comes from the real-time full-element perception of road traffic flow information by intelligent sensing equipment, including vehicles, pedestrians, etc. The dynamic traffic flow online simulation processing process in the simulation cloud is as follows: Figure 5 As shown, the final mixed traffic flow simulation rendering effect is as follows Figure 6 shown.

[0040] Among them, the dynamic traffic flow online simulation processing is specifically as follows:

[0041] (1) Initialize the model pool of the traffic flow model;

[0042] (2) receiving traffic flow data;

[0043] (3) Determine whether the traffic flow data currently received is the first frame. If so, execute step (4); otherwise, execute step (5);

[0044] (4) After calling the traffic flow model from the model pool and assigning a model ID, continue to execute step (12); each time the traffic flow model is called, one traffic flow model in the model pool is reduced;

[0045] (5) Is the model ID set the same as that of the previous frame? If so, execute step (12); otherwise, execute step (6);

[0046] (6) Whether there is a newly added model ID, if so, execute step (7), otherwise, directly execute step (11);

[0047] (7) Whether there are still available traffic flow models in the model pool, if so, execute step (8), otherwise, execute step (9);

[0048] (8) Assign a model ID to the current model, and then proceed to step (10);

[0049] (9) Initialize a traffic flow model and assign a model ID, then proceed to step (10);

[0050] (10) Check whether any model ID has disappeared. If so, execute step (11); otherwise, directly execute step (12);

[0051] (11) The traffic flow model of the disappeared model ID is recycled into the model pool (the number of traffic flow models in the model pool is increased by 1), and then step (12) is continued;

[0052] (12) Parsing traffic flow data (including category information and physical coordinate information) according to the model ID;

[0053] (13) Control the movement of the current traffic flow model based on the analyzed traffic flow data.

[0054] like Figure 7 As shown, as a possible implementation method, the real vehicle in-the-loop simulation test platform based on vehicle-road-cloud collaboration provided by an embodiment of the present invention includes: a strong real-time perception server at the road edge, an intelligent networked vehicle, a simulation cloud and a communication service end, wherein the intelligent networked vehicle end includes a combined inertial navigation module, a control module and a communication module, the combined inertial navigation module is used for positioning (high-precision positioning) of the intelligent networked vehicle end, and obtaining a vehicle-end positioning message, the control module is used to control the wire-controlled chassis of the intelligent networked vehicle end to drive the vehicle to move, and the communication module is used for data interaction with the communication service end, and the interactive data includes: the intelligent networked vehicle end sends a vehicle-end positioning message to the intelligent road end through the communication service end, and receives decision information from the simulation cloud through the communication service end;

[0055] The strong real-time perception server at the road edge includes a perception unit, a positioning module and a communication module, wherein the perception unit includes a roadside camera, an identification module, a screening module and a tracking module, wherein the roadside camera is used to collect road image data and send it to the identification module, the identification module is used to specify the identification of traffic element objects (for example, vehicles and pedestrians), and send the identification results to the screening module, the screening module selects traffic element objects within the road range and sends the screening results to the tracking module, the tracking module performs multi-target tracking on the received traffic element objects, and sends the pixel point information of the tracked targets to the positioning module in real time; the positioning module converts the representative pixel of each target (which can be the center point or the specified point such as the midpoint at the bottom of the target frame) into physical coordinate information based on the configured perspective transformation matrix, obtains the perception information of each target based on the category information and physical coordinate information of the target, i.e., the first traffic flow information; and deletes the perception information corresponding to the vehicle end in the first traffic information to obtain the second traffic information, i.e., the obstacle information; the first and second traffic information are sent to the communication server through the communication module of the smart road end, and the communication server then forwards the first and second traffic information to the simulation cloud;

[0056] The simulation cloud includes a traffic flow online simulation module, a real vehicle twin decision module and a communication module. The traffic flow online simulation module is used to visualize the first traffic flow information from the smart road end. In addition, the traffic flow online simulation module is also used to add virtual interference elements based on real traffic information and visualize them, and send the added virtual interference elements to the real vehicle twin decision planning module; the real vehicle twin decision module obtains the decision information under the current traffic scenario based on the configured decision algorithm to be tested and the second traffic flow information from the smart road end, and sends it to the communication server through the communication module of the simulation cloud; the communication server forwards the decision information to the intelligent connected vehicle end, and the communication module of the intelligent connected vehicle end parses the decision information and sends it to the control module of the intelligent connected vehicle end to drive the intelligent connected vehicle end to make corresponding movements.

[0057] That is, for the real vehicle in-the-loop simulation test platform based on vehicle-road-cloud collaboration provided by the embodiment of the present invention, the communication modules deployed on the intelligent connected vehicle side, the smart road side and the simulation cloud side constitute the communication system of the platform. In this embodiment, the communication system selects the MQTT-Mosquitto message platform solution that is widely used and open source in the field of the Internet of Things as the communication framework, and combines high-speed communication technology (such as mobile 5G / Wifi) to realize the real-time data interaction function between the intelligent connected vehicle side-the smart road side-the simulation cloud side. In this embodiment, the traffic flow online simulation module on the simulation cloud side can subscribe to the first traffic flow information of the smart road side through the topic publishing / subscription method, and the real vehicle twin decision module on the simulation cloud side can subscribe to the second traffic flow information of the smart road side. The intelligent connected vehicle side also obtains the decision information published by the simulation cloud side by subscription, and the smart road side also obtains the vehicle-side positioning information published by the intelligent connected vehicle side by subscription.

[0058] In this example, a high-performance PC is running on the intelligent connected vehicle. The software environment is the Robot Operating System (ROS), in which all functional modules, such as the positioning module (combined inertial navigation module), control module and communication module, use the ROS topic mechanism to achieve internal communication. The communication module mainly completes the task of data interaction between the intelligent connected vehicle and the simulation cloud / smart road. The intelligent connected vehicle can report information in Json format. The specific data content is shown in Table 1. The format and description of the reported data information of the strong real-time perception server at the road edge are shown in Table 2.

[0059] Table 1 Data format reported by the intelligent connected vehicle

[0060]

[0061]

[0062] Table 2 Data format reported by the edge server

[0063]

[0064] The simulation cloud (also called simulation cloud server) integrates the self-data of the intelligent vehicle network and the road-side traffic flow data of the strong real-time perception server at the road-side edge to perform scene twinning. The simulation cloud applies the intelligent network vehicle twin algorithm (i.e., the algorithm to be tested) to obtain decision information based on real-time traffic flow data, and transmits the decision information to the communication module of the intelligent network vehicle in real time through the communication server, and publishes the corresponding control message to the control module (control module) of the intelligent network vehicle. The control module controls the chassis of the real vehicle to make corresponding movements through the controller area network (CAN), and updates its own posture and positioning information at the same time, and then uploads it to the simulation cloud in real time through the communication server, completing the closed loop of the entire system and realizing the two-way mapping from the physical world to the digital world. Among them, the data format sent by the simulation cloud is shown in Table 3.

[0065] Table 3 Data format sent by the cloud simulation platform

[0066]

[0067] As a possible implementation method, the construction process of the real vehicle-in-the-loop simulation test platform based on vehicle-road-cloud collaboration provided by the embodiment of the present invention is as follows:

[0068] Step S1: The smart road terminal acquires real-time traffic flow data based on vehicle-road collaboration:

[0069] This step includes: acquiring roadside video image data, determining the traffic flow identification and positioning area, multi-target identification and tracking, setting the perspective transformation matrix, vehicle positioning correction, and calculating the traffic flow element position information;

[0070] Wherein, step S1 specifically includes:

[0071] Set the perspective transformation matrix: determine the traffic flow identification and positioning area, collect the pixel information and coordinate information of the four corner points of the area, and obtain the transformation matrix from pixel information to coordinate information through perspective transformation, that is, the perspective transformation matrix.

[0072] The roadside traffic flow data is acquired, and the roadside camera image data of the perception unit is acquired. The traffic flow elements are identified through YOLOv5, and the traffic lane is selected as the positioning and tracking area. The identification results are screened and put into the Deepsort multi-target tracker to realize the identification and tracking of traffic elements in the lane. Based on the set perspective transformation matrix, the pixels of each target can be converted into physical coordinate information.

[0073] Among them, the vehicle positioning correction, that is, using the geometric position relationship between the target vehicle and the road to realize the target vehicle positioning estimation correction, is specifically:

[0074] 1. Find the cross center point of the target vehicle tracking frame from the vehicle tracking frame pixel information extracted from the road end, and use the local physical coordinates of the cross center point pixel of the target vehicle tracking frame after perspective transformation to estimate the actual position of the target vehicle;

[0075] 2. Complete the positioning estimation correction of the vehicle based on the geometric relationship between the local physical coordinates corresponding to the cross center point of the target vehicle tracking frame and the actual center point coordinates of the target vehicle.

[0076] In actual use, if the physical coordinates of the perspective-transformed cross center pixel of the vehicle tracking frame are directly used to estimate the actual position of the vehicle, based on the geometric relationship between the physical coordinates corresponding to the center of the identification tracking frame and the actual center coordinates of the vehicle, the error between the two will increase as the distance between the vehicle and the camera increases when the vehicle height is constant. To reduce the error, this embodiment adds a positioning correction process, during which the target vehicle is still regarded as a cuboid to calculate the correction amount. Figure 8 As shown, the center of the vehicle identification frame corresponds to the geometric center of the cuboid of the target vehicle. By calculating the physical coordinates of the center point pixel of the tracking frame, it is determined that the center point is located in the left front or right front of the actual position of the vehicle; based on the judgment result combined with the camera height and the target vehicle height, the parameters Δx and Δy that need to be corrected for the X (horizontal coordinate) and Y (vertical coordinate) values ​​of the center point are calculated, and the correction is performed based on the previously obtained target vehicle center point (X, Y). The specific calculation formula is as follows:

[0077] Δy / y=h / 2H,Δx=Δy * tanβ;

[0078] Update X = X + / - Δx, Update Y = Y - Δy;

[0079] Where h represents the vehicle height, H represents the camera height, and β represents the angle between the vehicle-camera line and the camera optical axis.

[0080] Figure 8 In the figure, points A and B represent the two vertices of a rectangle with sides Δx and Δy.

[0081] Step S2, online simulation based on real-time traffic flow in the simulation cloud:

[0082] Static 3D modeling is based on real road and building data; dynamic information mainly comes from the real-time traffic flow data obtained in step S1, and the dynamic traffic flow information data is parsed and then rendered in real time; the combination of dynamic traffic flow and static roads restores the real traffic scene.

[0083] Wherein, step S2 specifically includes:

[0084] Step S21: Virtual static scene construction. It is mainly divided into static road and building construction. Static roads collect latitude and longitude coordinates on real roads by using vehicles equipped with high-precision positioning to obtain road curves based on real GPS data. A complete 1:1 modeling is performed based on CAD drawings, and real image textures of the main areas of the building are collected and mapped.

[0085] Step S22: Dynamic traffic flow scene construction. The dynamic traffic flow scene modeling data comes from the real-time full-element perception of road traffic flow information, including vehicles, pedestrians, etc., by intelligent perception equipment.

[0086] Step S3, autonomous driving simulation based on real vehicle in the loop:

[0087] The intelligent connected vehicle sends the positioning status information (real vehicle data) to the simulation cloud through the combined inertial navigation module through the communication server; the simulation cloud screens and makes decisions based on the real vehicle data and roadside traffic flow data; the simulation cloud feeds back the decision-making steering wheel and throttle results to the intelligent connected vehicle through the communication server; the communication module of the intelligent connected vehicle sends the corresponding control message to the control module to drive the real vehicle to make corresponding posture changes.

[0088] Wherein, step S3 specifically includes:

[0089] S31: Communication system. Optimize the deployment of Mosquitto MQTT server; deploy the communication module MQTTClient1 on the intelligent connected vehicle side; deploy the communication module MQTT Client2 on the strong real-time perception server on the road side; deploy the communication module MQTT Client3 on the simulation cloud side; achieve cross-network transmission through router port mapping; and achieve real-time data interaction between the intelligent connected vehicle side, the smart road side, and the simulation cloud side.

[0090] S32: Intelligent connected vehicle system deployment, intelligent connected vehicle ROS system deployment; combined inertial navigation positioning module deployment; control module deployment; communication module deployment; the communication module mainly completes the task of data interaction between the intelligent connected vehicle and the simulation cloud.

[0091] S32: Deployment of the real vehicle in-the-loop system, the road-side perception results (traffic flow data) of the smart road-side are uploaded to the simulation cloud; the data of the intelligent connected vehicle-side is uploaded to the simulation cloud in real time; the traffic flow online simulation module of the simulation cloud integrates the intelligent vehicle-connected vehicle-side data and the road-side traffic flow data to perform scene twinning, and realize dynamic traffic flow online simulation processing; the real vehicle twin decision module of the simulation cloud applies the intelligent connected vehicle twin algorithm to obtain decision information under the current traffic scenario according to the real-time traffic flow data; the simulation cloud transmits the decision information to the communication module of the intelligent connected vehicle-side in real time; the communication module of the intelligent connected vehicle-side publishes the corresponding control message to the control module; the control module controls the real vehicle chassis to make corresponding movements through CAN (Controller Area Network); the intelligent connected vehicle-side updates its own posture and positioning information, and then uploads it to the simulation cloud in real time.

[0092] The present invention constructs a closed-loop system consisting of an intelligent connected vehicle and a hybrid test environment. By introducing real-time traffic flow information, it can quickly generate test scenarios that fit the actual traffic conditions. It realizes the customizability of complex scenarios by combining virtual and real methods, and uses the random injection of scene elements to test the stability and robustness of the autonomous driving algorithm, providing strong support for the transition from simulation testing to road testing.

[0093] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

[0094] The above are only some embodiments of the present invention. For those skilled in the art, several modifications and improvements can be made without departing from the creative concept of the present invention, which all belong to the protection scope of the present invention.

Claims

1. A real vehicle-in-the-loop simulation test platform based on vehicle-road-cloud collaboration, characterized in that: include: Smart road side, intelligent connected vehicle side, simulation cloud side and communication service side; The intelligent networked vehicle end includes a combined inertial navigation module, a control module and a communication module, wherein the combined inertial navigation module is used for positioning the intelligent networked vehicle end and obtaining the vehicle end positioning message; the control module is used to control the wire-controlled chassis of the intelligent networked vehicle end to drive the vehicle movement; the communication module is used to interact with the communication service end for data, including: the intelligent networked vehicle end sends the vehicle end positioning message to the smart road end through the communication service end, and receives the decision information from the simulation cloud through the communication service end; The smart road terminal includes a perception unit, a positioning module and a communication module, wherein the perception unit includes a roadside camera, an identification module, a screening module and a tracking module, the roadside camera is used to collect road image data and send it to the identification module, the identification module is used to identify the specified traffic element object, and send the identification result to the screening module, the screening module selects the traffic element object within the lane range and sends the screening result to the tracking module, the tracking module performs multi-target tracking on the screened traffic element object, and sends the pixel point information of the tracked target to the positioning module in real time, the positioning module converts the representative pixel point of each target into physical coordinate information based on the configured perspective transformation matrix, obtains the perception information of each target based on the category information and physical coordinate information of the target, that is, the first traffic flow information; and deletes the perception information corresponding to the vehicle end in the first traffic information to obtain the second traffic information; sends the first and second traffic information to the communication service end through the communication module of the smart road terminal, and the communication service end forwards the first and second traffic information to the simulation cloud; The simulation cloud includes an online traffic flow simulation module, a real vehicle twin decision module and a communication module. The online traffic flow simulation module is used to visualize the first traffic flow information from the smart road end; the real vehicle twin decision module obtains the decision information under the current traffic scenario based on the configured decision algorithm to be tested and the second traffic flow information from the smart road end, and sends it to the communication server through the communication module of the simulation cloud; the communication server forwards the decision information to the intelligent connected vehicle end, and the communication module of the intelligent connected vehicle end parses the decision information and sends it to the control module of the intelligent connected vehicle end to drive the intelligent connected vehicle end to make corresponding movements.

2. The real vehicle-in-the-loop simulation test platform based on vehicle-road-cloud collaboration as claimed in claim 1, characterized in that: The specific setting method of the perspective transformation matrix of the positioning module of the intelligent road terminal is: Determine the traffic flow identification and positioning area; The pixel information and coordinate information of the four corner points of the traffic flow identification and positioning area are collected, and a transformation matrix from pixel information to coordinate information is obtained through perspective transformation to obtain a perspective transformation matrix.

3. The real vehicle-in-the-loop simulation test platform based on vehicle-road-cloud collaboration as claimed in claim 1 is characterized in that: The traffic flow online simulation module in the simulation cloud is also used to: add and visualize virtual interference elements based on real traffic information, and send the added virtual interference elements to the real vehicle twin decision module.

4. The real vehicle-in-the-loop simulation test platform based on vehicle-road-cloud collaboration as claimed in claim 1, characterized in that: The traffic flow online simulation module of the simulation cloud visualizes the first traffic flow information from the smart road end, including: simulation and visualization of static roads, static buildings and dynamic traffic flow scenarios; The modeling of static roads includes: collecting GPS geographical location information of roads, converting the GPS geographical location information of roads into plane coordinates through GPS data conversion tools, and obtaining roads modeled based on real GPS data; The modeling of static buildings includes: collecting real image textures of buildings within the driving visual range, performing 1:1 modeling based on CAD drawings, and mapping the constructed building models at corresponding locations; Dynamic traffic flow scene modeling includes: generating a traffic flow model based on first traffic flow information and controlling its movement.

5. The real vehicle-in-the-loop simulation test platform based on vehicle-road-cloud collaboration as claimed in claim 4 is characterized in that: The processing of dynamic traffic flow scenario modeling includes: (1) Initialize the model pool of the traffic flow model; (2) determining whether the received first traffic flow information is the first frame, if so, executing step (3); otherwise, executing step (4); (3) After calling the traffic flow model from the model pool and assigning a model ID, continue to execute step (11); each time the traffic flow model is called, one traffic flow model in the model pool is reduced; (4) Is the model ID set the same as that of the previous frame? If so, execute step (11); otherwise, execute step (5); (5) Whether there is a newly added model ID, if so, execute step (6), otherwise, directly execute step (10); (6) Whether there are still available traffic flow models in the model pool, if so, execute step (7), otherwise, execute step (8); (7) Assign a model ID to the current model and continue with step (9); (8) Initialize a traffic flow model and assign a model ID, and continue with step (9); (9) Check whether any model ID has disappeared. If so, execute step (10); otherwise, directly execute step (11); (10) The traffic flow model of the disappeared model ID is recycled into the model pool, and the number of traffic flow models in the model pool is increased by 1. And continue to execute step (11); (11) parsing the first traffic flow information according to the model ID to obtain category information and physical coordinate information of the traffic flow model; (12) Control the movement of the current traffic flow model based on the category information and physical coordinate information of the traffic flow model.

6. The real vehicle-in-the-loop simulation test platform based on vehicle-road-cloud collaboration according to any one of claims 1 to 5, characterized in that: The smart road end and the communication server end, the intelligent connected vehicle end and the communication server end, and the simulation cloud end and the communication server end interact with each other through topic publishing / subscription.

Images