Unmanned cluster hardware-in-loop simulation test and evaluation system

Abstract

The invention discloses an unmanned cluster hardware-in-the-loop simulation test and evaluation system, and relates to the field of unmanned cluster simulation test and evaluation. The problems that real machine testing cost of the unmanned cluster is high and virtual testing is unreliable are solved, and accelerated testing and reliable evaluation of the unmanned cluster performance are achieved. The technical key points are as follows: the test and evaluation system comprises a virtual simulation module, a hardware calculation module and a credible evaluation module; wherein the virtual simulation module is mainly used for building a task execution scene of an unmanned cluster and displaying a task execution process of the unmanned cluster; the hardware calculation module is used for executing a deployed unmanned cluster algorithm; the credible evaluation module is used for evaluating the execution result of the unmanned cluster algorithm; in the virtual simulation module, a virtual unmanned cluster senses environment information through a sensor and transmits the environment information to a hardware calculation module, an unmanned cluster algorithm deployed on the hardware calculation module carries out operation based on the sensed environment information, and an action instruction or a planning instruction of the unmanned cluster is obtained and transmitted to the virtual simulation module; a virtual unmanned cluster in the virtual simulation module transmits information such as position coordinates and movement speed of the virtual unmanned cluster and target information to the credible evaluation module through real-time interaction, and the credible evaluation module evaluates the execution effect of an unmanned cluster algorithm. The method is used for comprehensively testing the hardware computing embedded program.

Description

Technical Field

[0001] The present invention relates to the field of hardware-in-the-loop simulation test and evaluation of unmanned clusters, and in particular to an unmanned cluster hardware-in-the-loop simulation test and evaluation system. Background Art

[0002] With the continuous development of unmanned system cluster technology, unmanned clusters have been widely used in agriculture, industry and military fields. At present, there are many intelligent algorithms related to unmanned clusters, but the actual application effects of these algorithms are difficult to effectively test and evaluate. Unmanned clusters need to rely on efficient communication and coordination capabilities to complete tasks. The current testing methods may not be sufficient to simulate various possible environmental interference situations, nor can they accurately test the ability of unmanned clusters to collaboratively perform tasks in these situations. In actual environments, unmanned clusters may face problems such as complex terrain, severe weather conditions and perception interference. Existing testing methods may be difficult to fully simulate and test the performance of unmanned clusters in these complex environments. Unmanned clusters need to learn and make decisions autonomously according to actual conditions during the execution of tasks. However, existing testing methods may not be able to fully and effectively test the learning speed and decision-making quality of unmanned clusters in unknown environments. At present, the essence of unmanned cluster testing is to repeat multiple single-unit tests, which is obviously insufficient for testing that reflects the characteristics of cluster tasks, and lacks method and technical system support. Therefore, it is urgently necessary to drive unmanned swarm tasks, adopt in-field hardware-in-the-loop accelerated simulation methods, focus on areas such as typical mission information sharing, formation collaborative gain, on-the-spot handling and mission planning, build a distributed hardware-in-the-loop simulation test and evaluation platform, design standardized test subjects, and provide experimental specifications and technical basis for actual machine testing of unmanned swarms.

[0003] For example: the existing patent document with document number CN110865627A discloses a test bed architecture for an intelligent unmanned cluster system, and the entire test bed architecture is divided into six layers: test bed operation platform layer, test bed support environment layer, simulation model library layer, cluster operation engine layer, test management and evaluation layer, and experimental effect presentation layer. The operation platform layer, support environment layer, simulation model library layer, cluster operation engine layer, test management and evaluation layer, and experimental effect presentation layer are abstracted layer by layer to provide support for the upper layer. The UAV simulation will use the parallel computing toolbox and distributed computing server provided by Matlab to provide scalable and customizable virtual simulation capabilities. A private cloud computing environment is built using high-performance servers as the hardware environment. Each server runs a Matlab / Simulink distributed computing server. The distributed computing server drives each simulated UAV to run stand-alone, and the entire simulation cloud environment is configured and managed by a parallel computing toolbox. This prior art does not mention hardware-in-the-loop simulation testing, and cannot evaluate the hardware deployment effect of the unmanned cluster algorithm. The existing patent document with the document number CN116088526A discloses a collision avoidance capability test method and system for an unmanned cluster system, which uses the internal collision avoidance rate and external obstacle avoidance degree as the evaluation scale of the unmanned cluster collision avoidance capability, and evaluates the collision avoidance capability by observing the distance between the cluster individuals and the distance between the individuals and obstacles. When testing the obstacle avoidance control capability of the unmanned aerial vehicle cluster formation, the size of the cluster, the individual position, speed, and obstacle position and other information are obtained in real time, the internal collision avoidance rate and external obstacle avoidance degree of the unmanned cluster formation system are calculated, and the evaluation index is obtained by combining the internal collision avoidance rate and the external obstacle avoidance degree, thereby testing the collision avoidance control capability of the unmanned cluster system. This prior art provides a feasible method for evaluating the intelligence of the unmanned cluster system, which can quickly and effectively evaluate the collision avoidance control capability of the unmanned cluster formation, and helps to improve the effectiveness and stability of the obstacle avoidance control of the unmanned cluster system. This prior art does not mention hardware-in-the-loop simulation testing, and cannot evaluate the hardware deployment effect of the unmanned cluster algorithm, and does not describe the simulation test scenario.

[0004] Problems such as high cost, high difficulty, and single test tasks in field testing have become increasingly prominent. Sufficient and comprehensive hardware-in-the-loop simulation testing can provide solid support for actual machine experiments. However, few people have studied how to effectively test the learning speed and decision-making quality of unmanned swarms in unknown environments. Summary of the invention

[0005] The technical problems to be solved by the present invention are:

[0006] The present invention aims to solve the problems of difficulty and high cost in real-machine testing of unmanned clusters, unreliable virtual testing, simple and single simulation test scenarios and difficulty in evaluating the hardware deployment effect of unmanned cluster algorithms, and further proposes an unmanned cluster hardware-in-the-loop simulation test and evaluation system.

[0007] The technical solution adopted by the present invention to solve the above technical problems is:

[0008] A hardware-in-the-loop simulation test and evaluation system for unmanned clusters, the test and evaluation system specifically includes a virtual simulation module, a hardware computing module and a trusted evaluation module. The virtual simulation module is mainly used to build a task execution scenario for an unmanned cluster and display the task execution process of the unmanned cluster; the hardware computing module is used to execute the deployed unmanned cluster algorithm; and the trusted evaluation module is used to evaluate the execution result of the unmanned cluster algorithm. In the hardware-in-the-loop simulation test and evaluation system for unmanned clusters, the virtual simulation module, the hardware computing module and the trusted evaluation module can communicate with each other and complete real-time data transmission. In the virtual simulation module, the virtual unmanned cluster senses environmental information through sensors and transmits it to the hardware computing platform. The unmanned cluster algorithm deployed on the hardware computing platform performs calculations based on the sensed environmental information, obtains the action instructions or planning instructions of the unmanned cluster and transmits them to the virtual simulation module. The virtual unmanned cluster equipment in the virtual simulation module executes the action instructions or planning instructions in the simulation environment. In the hardware-in-the-loop simulation test and evaluation system for unmanned clusters, the virtual unmanned cluster in the virtual simulation module transmits its own position coordinates, movement speed and other information as well as target information to the trusted evaluation module through real-time interaction, and the trusted evaluation module evaluates the execution effect of the unmanned cluster algorithm.

[0009] The unmanned cluster hardware-in-the-loop simulation test and evaluation system is suitable for virtual simulation and test evaluation of various unmanned systems such as unmanned aerial vehicles, unmanned vehicles, and unmanned ships.

[0010] The unmanned cluster hardware-in-the-loop simulation test and evaluation system, the simulation environment in the virtual simulation module is built by simulation platforms such as Gazebo and Unity, and specifically includes virtual scenes, virtual dynamic and static obstacles, virtual target positions or subjects, and virtual unmanned cluster equipment.

[0011] The unmanned cluster hardware-in-the-loop simulation test and evaluation system, wherein the hardware computing module is composed of one or more sets of hardware boards and is used to execute the unmanned cluster algorithm.

[0012] The unmanned cluster hardware-in-the-loop simulation test and evaluation system can evaluate the cluster tasks such as formation control, path planning, collaborative perception and task allocation of the unmanned cluster. The evaluation indicators of formation control include: formation control success rate, formation efficiency, formation stability, formation collaborative accuracy and control effectiveness; the evaluation indicators of path planning include: planning success rate, execution time, execution distance and planning efficiency; the evaluation indicators of collaborative perception include: perception accuracy, tracking accuracy and perception timeliness; the evaluation indicators of task allocation include: allocation real-time, allocation effectiveness and cluster execution efficiency.

[0013] The test process of the evaluation indicators related to formation control, path planning, collaborative perception and task allocation is as follows:

[0014] (1) Formation control evaluation indicators

[0015] 1) Control success rate

[0016] Evaluate the control of the unmanned swarm formation in various scenarios, using the formation control success rate as the evaluation indicator. The higher the formation control success rate, the stronger the formation's obstacle avoidance capability. The test steps are as follows:

[0017] (a) Deploy a group of unmanned swarms, set up multiple randomly distributed static obstacles and multiple dynamic obstacles, and let the swarms reach the target position from the initial position;

[0018] (b) Dynamic obstacles move randomly to test the unmanned swarm formation’s ability to avoid obstacles and count the number of unmanned equipment that successfully reach the target point;

[0019] (c) Multiple experiments were conducted to calculate the success rate of formation control of the cluster under the formation obstacle avoidance algorithm.

[0020] 2) Formation efficiency

[0021] Test the ability of the cluster to form a team efficiently, using the number of algorithm iterations and path distance as evaluation indicators. The shorter the algorithm calculation time and the smaller the average path distance, the higher the efficiency of the algorithm. The test steps are as follows:

[0022] (a) Deploy a group of unmanned swarms, set the starting position and target position of each unmanned equipment in the swarm, and determine the swarm mission;

[0023] (b) Use the formation control algorithm to calculate the path trajectory of each unmanned equipment under the swarm task, record the time required for the algorithm to generate the path trajectory and the total path length of the drone group;

[0024] (c) Multiple experiments, the calculation algorithm calculates the distance between calculations and the average path distance under multiple experiments.

[0025] 3) Formation stability

[0026] The formation execution status is used as an evaluation indicator to test the formation stability. The test steps are as follows:

[0027] (a) Deploy a group of unmanned swarms, set the starting position and target position of each unmanned equipment in the swarm, and determine the swarm mission;

[0028] (b) Use the formation algorithm to form a formation, and record the speed, altitude, heading angle and other information between unmanned clusters every second during the execution process;

[0029] (c) Observe the formation control situation by the change of unmanned cluster information over time. If the change of unmanned cluster information over time tends to be stable, it is considered that the formation stability is high;

[0030] (d) After the formation is completed, the size of the unmanned cluster is increased or decreased, and the formation of the cluster is observed through a line graph of the unmanned cluster information changing over time. If the final line graph still tends to be stable, it is considered that the re-formation is successful and the formation is relatively stable.

[0031] 4) Formation coordination accuracy

[0032] This indicator measures the maximum number of movement commands output per minute when the unmanned swarm is in formation coordination control, reflecting the stability and sensitivity of the algorithm. The test steps are as follows:

[0033] (a) In the formation control task, for each unmanned equipment, record the number of commands output by the formation control algorithm per minute for maintaining formation coordination;

[0034] (b) The maximum number of movement commands output by all unmanned equipment is the formation coordination accuracy.

[0035] 5) Control effectiveness

[0036] Test whether the formation can be carried out effectively in certain scenarios. The test steps are as follows:

[0037] (a) Deploy a group of unmanned swarms, start and execute the formation control algorithm at different locations, and the unmanned swarms execute the trajectory planned by the algorithm;

[0038] (b) Record the trajectory of the unmanned swarm from the beginning to the completion of the formation, and check whether the formation is completed and the ability to maintain the formation based on the cluster trajectory.

[0039] (2) Path planning evaluation indicators

[0040] 1) Planning success rate

[0041] Test the rate at which the unmanned swarm successfully completes its navigation mission. The test steps are as follows:

[0042] (a) Deploy a group of unmanned swarms, set up multiple randomly distributed static obstacles and multiple dynamic obstacles, and let the swarms reach the target position from the initial position;

[0043] (b) Count the number of tasks performed in a complex scenario;

[0044] (c) Record the number of successful completions of cluster tasks, that is, the number of times the unmanned cluster successfully reaches the target point in complex scenarios;

[0045] (d) Calculate the success rate percentage.

[0046] 2) Execution time

[0047] The average time, maximum time, and minimum time it takes for an unmanned cluster to complete a task. The test steps are as follows:

[0048] (a) Count the total time it takes for the unmanned swarm to execute all tasks;

[0049] (b) Divide the total time by the total number of tasks to get the average execution time; average execution time = (total execution time of all tasks) / (total number of tasks);

[0050] (c) Find the longest execution time, which is the longest execution time;

[0051] (d) Find the shortest execution time among them, which is the shortest execution time.

[0052] 3) Execution distance

[0053] The average distance, maximum distance, and minimum distance that the unmanned swarm takes to complete its mission. The test steps are as follows:

[0054] (a) Count the total execution distance of all navigation tasks of the unmanned swarm;

[0055] (b) Divide the total execution distance by the total number of navigation tasks to obtain the average navigation distance;

[0056] (c) Find the largest execution distance, which is the maximum execution distance;

[0057] (d) Find the shortest execution distance, which is the minimum execution distance.

[0058] 4) Planning efficiency

[0059] An indicator used to evaluate the efficiency of the path planning algorithm in the unmanned swarm intelligent decision-making algorithm. The shorter the planning time, the higher the efficiency of the intelligent decision-making algorithm. The test steps are as follows:

[0060] (a) In the path planning task, record the time it takes for the algorithm to generate the final valid path. This time usually refers to the time from the start of path planning to the final acquisition of a feasible path;

[0061] (b) The recorded planning time is used as the value of the planning time.

[0062] (3) Evaluation indicators of collaborative perception

[0063] 1) Perception accuracy

[0064] Perception accuracy refers to the proportion of all detected targets that are correctly identified as targets. The higher the perception accuracy, the higher the accuracy of the system in correctly identifying targets. The test steps are as follows:

[0065] (a) Deploy a group of unmanned swarms, set multiple targets, and use the unmanned swarm intelligent perception algorithm to identify and detect the targets;

[0066] (b) Calculate the number of true positives (TP) and false positives (FP) based on the recognition detection results. True positives refer to the number of targets that are correctly detected and recognized, and false positives refer to the number of non-targets that are mistakenly identified as targets;

[0067] (c) Calculate the perception accuracy (Precision) = TP / (TP+FP), where TP is the number of true positives and FP is the number of false positives.

[0068] 2) Tracking accuracy

[0069] Tracking accuracy refers to the degree of deviation between the unmanned cluster and the actual position of the target during the tracking process. The higher the tracking accuracy, the smaller the error between the unmanned cluster and the actual position when tracking the target. The test steps are as follows:

[0070] (a) For each target tracking process, record the target’s true position and the position information output by the unmanned cluster;

[0071] (b) Calculate the absolute error of each target tracking process, that is, the distance difference between the target's true position and the position output by the unmanned cluster;

[0072] (c) Sum the absolute errors of all target tracking processes;

[0073] (d) The total absolute error value is divided by the number of targets to obtain the tracking accuracy (MAE).

[0074] MAE = Σ|true position - tracker output position| / number of targets.

[0075] 3) Perceived timeliness

[0076] Perception timeliness refers to the difference between the time when the unmanned swarm collaborative perception system detects and identifies a target and the time when the target actually appears or changes. It indicates the system's response speed to changes in the target. The test steps are as follows:

[0077] (a) During the operation of the collaborative perception system, the actual appearance time of each target and the time when the system issues a warning signal are recorded;

[0078] (b) Calculate the time delay for each target, that is, the time when the target appears minus the time when the system issues the warning signal;

[0079] (c) By performing a statistical analysis on the time delays of all targets, indicators such as the average time delay, maximum time delay, and minimum time delay can be calculated.

[0080] (4) Task allocation evaluation indicators

[0081] 1) Real-time distribution

[0082] The test is whether the cluster tasks can be given timely and effective instructions, with the task allocation time as the evaluation indicator. The shorter the time the algorithm takes to complete the task allocation, the higher the real-time performance of the task allocation. The test steps are as follows:

[0083] (a) Deploy a group of unmanned clusters, set multiple targets, and the task allocation algorithm generates a task allocation route plan for the unmanned cluster according to the requirements.

[0084] (b) The time required for the test task allocation algorithm to complete the allocation under this condition is the real-time performance of the allocation;

[0085] 2) Distribution effectiveness

[0086] Test whether the unmanned cluster algorithm can effectively assign tasks to the target, using the target assignment success rate as the evaluation indicator. The higher the assignment success rate, the better the assignment effectiveness. The test steps are as follows:

[0087] (a) Deploy a group of unmanned swarms and set multiple targets. There are randomly distributed obstacles between the unmanned swarms and the targets. The unmanned swarms start from the initial position and reach the target position.

[0088] (b) Using the unmanned cluster task allocation algorithm to allocate the target positions corresponding to each unmanned equipment; if all unmanned clusters reach all target positions, it means that the task allocation is successful;

[0089] (c) Change the cluster size and target number for multiple tests and calculate the task allocation success rate.

[0090] 3) Cluster execution efficiency

[0091] Test the path distance and execution time required by the cluster to complete the task. The path distance and execution time required by the cluster from the starting point to the target point according to the task allocation result are used as indicators. The shorter the path distance and execution time, the better the performance of the task allocation algorithm. The test steps are as follows:

[0092] (a) Deploy a group of unmanned swarms and set multiple targets. There are randomly distributed obstacles between the unmanned swarms and the targets. The unmanned swarms start from the initial position and reach the target position.

[0093] (b) Calculate the average execution distance and average execution time of the unmanned cluster under this task.

[0094] The present invention has the following beneficial technical effects:

[0095] Based on the virtual simulation module, the present invention can realize the construction of complex test and evaluation scenarios of unmanned clusters at sea, on land, in the air, and in the sky, as well as the simulation of various interferences. It can perform simulation tests and evaluations on the mission characteristics of unmanned cluster systems in any customized environment, breaking through the defects and limitations of real-machine simulation of unmanned clusters. The present invention uses real-time data transmission between the virtual simulation module, the hardware computing module, and the trusted evaluation module to comprehensively test the hardware computing embedded program in an efficient and low-cost manner. Confirming the feasibility of the task through simulation can avoid any risk of crash.

[0096] The present invention has the advantages of high efficiency, low cost, and repeatability. It can effectively solve the problems of uncertainty, incompleteness, and non-measurement in the actual machine test of unmanned clusters in actual environments, and provide important supplements and basic support for the actual machine test of unmanned clusters. The present invention constructs effective test evaluation indicators for cluster tasks such as formation control, path planning, collaborative perception, and task allocation of unmanned clusters, and can perform full-cycle testing and effective evaluation on unmanned cluster tasks. In addition, the unmanned cluster algorithm deploys a hardware computing module, which can realize the algorithm deployment under onboard computing resources, further approaching the actual unmanned cluster test. BRIEF DESCRIPTION OF THE DRAWINGS

[0097] Figure 1 This is a framework diagram of an unmanned swarm hardware-in-the-loop simulation test and evaluation system;

[0098] Figure 2 It is a multi-platform hardware-in-the-loop simulation test framework based on digital modeling;

[0099] Figure 3 It is the overall control framework of the UAV swarm simulation platform;

[0100] Figure 4 To test and evaluate the mission characteristics of unmanned swarms for swarm collaborative tasks (process diagram of the mission characteristics test of unmanned swarms for combat).

[0101] Figure 5 This is a schematic diagram of the UAV swarm mission characteristic testing framework;

[0102] Figure 6 This is a diagram of the task-driven in-field hardware-in-the-loop acceleration test architecture;

[0103] Figure 7 Figure 2 is a block diagram of the digital model evaluation indicators. DETAILED DESCRIPTION

[0104] In order to make the purpose, technical scheme and advantages of the embodiments of the present invention clearer, the following will be combined with the attached embodiments of the present invention. Figure 1-7 , the technical solutions in the implementation modes of the present invention are described clearly and completely. Obviously, the described implementation modes are only part of the implementation modes of the present invention, rather than all the implementation modes.

[0105] The unmanned cluster hardware-in-the-loop simulation test and evaluation system described in this embodiment includes:

[0106] An unmanned cluster hardware-in-the-loop simulation test and evaluation system, the test and evaluation system specifically includes a virtual simulation module, a hardware computing module and a trusted evaluation module. The virtual simulation module is mainly used to build the task execution scene of the unmanned cluster and display the unmanned cluster task execution process; the hardware computing module is used to execute the deployed unmanned cluster algorithm; the trusted evaluation module is used to evaluate the execution results of the unmanned cluster algorithm.

[0107] like Figure 1 As shown, the unmanned cluster hardware-in-the-loop simulation test and evaluation system integrates ROS operating system, 3D visual scene simulation platform, algorithm model library, sensor and communication modules to build a simulation environment that maps the real unmanned cluster execution mission scene; and builds an unmanned cluster indoor simulation hardware platform by integrating development board, ground station, data transmission terminal and other modules. Based on the indoor simulation system and hardware platform, the unmanned cluster hardware-in-the-loop accelerated test experiment is carried out. Finally, based on the multi-dimensional and multi-level evaluation standards and evaluation index system, the unmanned cluster hardware-in-the-loop simulation test results are effectively evaluated.

[0108] The unmanned cluster hardware-in-the-loop simulation test and evaluation system is suitable for virtual simulation and test evaluation of various unmanned systems such as unmanned aerial vehicles, unmanned vehicles, and unmanned ships.

[0109] The unmanned cluster hardware-in-the-loop simulation test and evaluation system, the simulation environment in the virtual simulation module is built by simulation platforms such as Gazebo and Unity, and specifically includes virtual scenes, virtual dynamic and static obstacles, virtual target positions or subjects, and virtual unmanned cluster equipment.

[0110] The unmanned cluster hardware-in-the-loop simulation test and evaluation system, wherein the hardware computing module is composed of one or more sets of hardware boards and is used to execute the unmanned cluster algorithm.

[0111] The hardware part of the unmanned cluster hardware-in-the-loop simulation test and evaluation system is as follows: Figure 2 As shown in the figure, the hardware computing unit, the three-dimensional simulation environment and the digital model of the unmanned cluster are connected through optical fiber links to achieve real-time communication between the nodes, realize distributed semi-physical simulation, and have good scalability. The number of simulation node computers can be increased as needed. The simulation node connects the virtual three-dimensional scene and the digital model through the serial port, which mainly verifies the cluster intelligence characteristics such as path planning, task allocation, formation control and collaborative perception of the unmanned cluster. Through the parallel connection of the simulation nodes, real-time interaction with the virtual simulation environment is realized to realize the distributed real-time simulation test of the unmanned cluster.

[0112] In the unmanned cluster hardware-in-the-loop simulation test and evaluation system, the virtual simulation module, the hardware computing module and the trusted evaluation module can communicate with each other and complete real-time data transmission. In the virtual simulation module, the virtual unmanned cluster senses environmental information through sensors and transmits it to the hardware computing platform. The unmanned cluster algorithm deployed on the hardware computing platform performs calculations based on the sensed environmental information, obtains action instructions or planning instructions for the unmanned cluster and transmits them to the virtual simulation module. The virtual unmanned cluster equipment in the virtual simulation module executes action instructions or planning instructions in the simulation environment.

[0113] In the unmanned cluster hardware-in-the-loop simulation test and evaluation system, the virtual unmanned cluster in the virtual simulation module transmits its own position coordinates, movement speed and other information as well as target information to the trusted evaluation module through real-time interaction, and the trusted evaluation module evaluates the execution effect of the unmanned cluster algorithm.

[0114] Taking the hardware-in-the-loop simulation test and evaluation of drones as an example, the simulation environment is built using a simulation platform. The overall control framework of the simulation platform includes three main components: simulator, controller, and sensor. The simulator is responsible for simulating the dynamics, environment, and sensor data of the unmanned cluster; the controller is responsible for designing and implementing various control algorithms to move the simulation platform to the specified location and achieve other tasks; the sensor is responsible for acquiring environmental information and feeding it back to the controller. The simulation platform provides an unmanned cluster control framework based on physical simulation, which can be used to test different control algorithms to make the unmanned cluster present better performance and reliability. The framework can simulate a variety of unmanned clusters and environments, which allows users to test and optimize in different scenarios and implement different applications. The overall control framework of the simulation platform is as follows Figure 3 shown.

[0115] The blue module is the simulation platform module, the yellow module is the PX4 SITL module, and the light orange module is other modules. The controller can receive messages from three modules: motion planning, state estimation, and ground station. Different operations are performed according to the received messages. At the same time, the simulation platform supports the independent design of the structure of the unmanned cluster, and the sensor module can also be imported to simulate the actual hardware. The perception part of the unmanned cluster mainly involves the autonomous positioning of the unmanned cluster, which mainly obtains the current position of the unmanned cluster through information such as images and point clouds. The motion planning part of the unmanned cluster mainly generates a smooth trajectory based on the current position and the position of the target point, so as to send the target point to the control unit to control the autonomous navigation flight of the unmanned cluster. The controller is responsible for designing and implementing control algorithms to control the unmanned cluster to complete various tasks. The simulation platform provides a variety of controllers, such as PID controller, LQR controller, and deep learning controller. In addition, the simulation platform also supports viewing the various status information of the unmanned cluster on the QGground platform to better control the flight of the unmanned cluster.

[0116] like Figure 4 As shown, the unmanned cluster hardware-in-the-loop simulation test and evaluation system can evaluate the cluster tasks such as formation control, path planning, collaborative perception and task allocation of the unmanned cluster. The evaluation indicators of formation control include: control success rate, formation efficiency, formation stability, formation collaborative accuracy and control effectiveness; the evaluation indicators of path planning include: planning success rate, execution time, execution distance and planning efficiency; the evaluation indicators of collaborative perception include: perception accuracy, tracking accuracy and perception timeliness; the evaluation indicators of task allocation include: allocation real-time, allocation effectiveness and cluster execution efficiency.

[0117] The test process of the evaluation indicators related to formation control, path planning, collaborative perception and task allocation is as follows:

[0118] (1) Formation control evaluation indicators

[0119] 1) Control success rate

[0120] Evaluate the control of the unmanned swarm formation in various scenarios, using the formation control success rate as the evaluation indicator. The higher the formation control success rate, the stronger the formation's obstacle avoidance capability. The test steps are as follows:

[0121] (a) Deploy a group of unmanned swarms, set up multiple randomly distributed static obstacles and multiple dynamic obstacles, and let the swarms reach the target position from the initial position;

[0122] (b) Dynamic obstacles move randomly to test the unmanned swarm formation’s ability to avoid obstacles and count the number of unmanned equipment that successfully reach the target point;

[0123] (c) Multiple experiments were conducted to calculate the success rate of formation control of the cluster under the formation obstacle avoidance algorithm.

[0124] 2) Formation efficiency

[0125] Test the ability of the cluster to form a team efficiently, using the number of algorithm iterations and path distance as evaluation indicators. The shorter the algorithm calculation time and the smaller the average path distance, the higher the efficiency of the algorithm. The test steps are as follows:

[0126] (a) Deploy a group of unmanned swarms, set the starting position and target position of each unmanned equipment in the swarm, and determine the swarm mission;

[0127] (b) Use the formation control algorithm to calculate the path trajectory of each unmanned equipment under the swarm task, record the time required for the algorithm to generate the path trajectory and the total path length of the drone group;

[0128] (c) Multiple experiments, the calculation algorithm calculates the distance between calculations and the average path distance under multiple experiments.

[0129] 3) Formation stability

[0130] The formation execution status is used as an evaluation indicator to test the formation stability. The test steps are as follows:

[0131] (a) Deploy a group of unmanned swarms, set the starting position and target position of each unmanned equipment in the swarm, and determine the swarm mission;

[0132] (b) Use the formation algorithm to form a formation, and record the speed, altitude, heading angle and other information between unmanned clusters every second during the execution process;

[0133] (c) Observe the formation control situation by the change of unmanned cluster information over time. If the change of unmanned cluster information over time tends to be stable, it is considered that the formation stability is high;

[0134] (d) After the formation is completed, the size of the unmanned cluster is increased or decreased, and the formation of the cluster is observed through a line graph of the unmanned cluster information changing over time. If the final line graph still tends to be stable, it is considered that the re-formation is successful and the formation is relatively stable.

[0135] 4) Formation coordination accuracy

[0136] This indicator measures the maximum number of movement commands output per minute when the unmanned swarm is in formation coordination control, reflecting the stability and sensitivity of the algorithm. The test steps are as follows:

[0137] (a) In the formation control task, for each unmanned equipment, record the number of commands output by the formation control algorithm per minute for maintaining formation coordination;

[0138] (b) The maximum number of movement commands output by all unmanned equipment is the formation coordination accuracy.

[0139] 5) Control effectiveness

[0140] Test whether the formation can be carried out effectively in certain scenarios. The test steps are as follows:

[0141] (a) Deploy a group of unmanned swarms, start and execute the formation control algorithm at different locations, and the unmanned swarms execute the trajectory planned by the algorithm;

[0142] (b) Record the trajectory of the unmanned swarm from the beginning to the completion of the formation, and check whether the formation is completed and the ability to maintain the formation based on the cluster trajectory.

[0143] (2) Path planning evaluation indicators

[0144] 1) Planning success rate

[0145] Test the rate at which the unmanned swarm successfully completes its navigation mission. The test steps are as follows:

[0146] (a) Deploy a group of unmanned swarms, set up multiple randomly distributed static obstacles and multiple dynamic obstacles, and let the swarms reach the target position from the initial position;

[0147] (b) Count the number of tasks performed in a complex scenario;

[0148] (c) Record the number of successful completions of cluster tasks, that is, the number of times the unmanned cluster successfully reaches the target point in complex scenarios;

[0149] (d) Calculate the success rate percentage.

[0150] 2) Execution time

[0151] The average time, maximum time, and minimum time it takes for an unmanned cluster to complete a task. The test steps are as follows:

[0152] (a) Count the total time it takes for the unmanned swarm to execute all tasks;

[0153] (b) Divide the total time by the total number of tasks to get the average execution time; average execution time = (total execution time of all tasks) / (total number of tasks);

[0154] (c) Find the longest execution time, which is the longest execution time;

[0155] (d) Find the shortest execution time among them, which is the shortest execution time.

[0156] 3) Execution distance

[0157] The average distance, maximum distance, and minimum distance that the unmanned swarm takes to complete its mission. The test steps are as follows:

[0158] (a) Count the total execution distance of all navigation tasks of the unmanned swarm;

[0159] (b) Divide the total execution distance by the total number of navigation tasks to obtain the average navigation distance;

[0160] (c) Find the largest execution distance, which is the maximum execution distance;

[0161] (d) Find the shortest execution distance, which is the minimum execution distance.

[0162] 4) Planning efficiency

[0163] An indicator used to evaluate the efficiency of the path planning algorithm in the unmanned swarm intelligent decision-making algorithm. The shorter the planning time, the higher the efficiency of the intelligent decision-making algorithm. The test steps are as follows:

[0164] (a) In the path planning task, record the time it takes for the algorithm to generate the final valid path. This time usually refers to the time from the start of path planning to the final acquisition of a feasible path;

[0165] (b) The recorded planning time is used as the value of the planning time.

[0166] (3) Evaluation indicators of collaborative perception

[0167] 1) Perception accuracy

[0168] Perception accuracy refers to the proportion of all detected targets that are correctly identified as targets. The higher the perception accuracy, the higher the accuracy of the system in correctly identifying targets. The test steps are as follows:

[0169] (a) Deploy a group of unmanned swarms, set multiple targets, and use the unmanned swarm intelligent perception algorithm to identify and detect the targets;

[0170] (b) Calculate the number of true positives (TP) and false positives (FP) based on the recognition detection results. True positives refer to the number of targets that are correctly detected and recognized, and false positives refer to the number of non-targets that are mistakenly identified as targets;

[0171] (c) Calculate the perception accuracy (Precision) = TP / (TP+FP), where TP is the number of true positives and FP is the number of false positives.

[0172] 2) Tracking accuracy

[0173] Tracking accuracy refers to the degree of deviation between the unmanned cluster and the actual position of the target during the tracking process. The higher the tracking accuracy, the smaller the error between the unmanned cluster and the actual position when tracking the target. The test steps are as follows:

[0174] (a) For each target tracking process, record the target’s true position and the position information output by the unmanned cluster;

[0175] (b) Calculate the absolute error of each target tracking process, that is, the distance difference between the target's true position and the position output by the unmanned cluster;

[0176] (c) Sum the absolute errors of all target tracking processes;

[0177] (d) The total absolute error value is divided by the number of targets to obtain the tracking accuracy (MAE).

[0178] MAE = Σ|true position - tracker output position| / number of targets.

[0179] 3) Perceived timeliness

[0180] Perception timeliness refers to the difference between the time when the unmanned swarm collaborative perception system detects and identifies a target and the time when the target actually appears or changes. It indicates the system's response speed to changes in the target. The test steps are as follows:

[0181] (a) During the operation of the collaborative perception system, the actual appearance time of each target and the time when the system issues a warning signal are recorded;

[0182] (b) Calculate the time delay for each target, that is, the time when the target appears minus the time when the system issues the warning signal;

[0183] (c) By performing a statistical analysis on the time delays of all targets, indicators such as the average time delay, maximum time delay, and minimum time delay can be calculated.

[0184] (4) Task allocation evaluation indicators

[0185] 1) Real-time distribution

[0186] The test is whether the cluster tasks can be given timely and effective instructions, with the task allocation time as the evaluation indicator. The shorter the time the algorithm takes to complete the task allocation, the higher the real-time performance of the task allocation. The test steps are as follows:

[0187] (a) Deploy a group of unmanned clusters, set multiple targets, and the task allocation algorithm generates a task allocation route plan for the unmanned cluster according to the requirements.

[0188] (b) The time required for the test task allocation algorithm to complete the allocation under this condition is the real-time performance of the allocation;

[0189] 2) Distribution effectiveness

[0190] Test whether the unmanned cluster algorithm can effectively assign tasks to the target, using the target assignment success rate as the evaluation indicator. The higher the assignment success rate, the better the assignment effectiveness. The test steps are as follows:

[0191] (a) Deploy a group of unmanned swarms and set multiple targets. There are randomly distributed obstacles between the unmanned swarms and the targets. The unmanned swarms start from the initial position and reach the target position.

[0192] (b) Using the unmanned cluster task allocation algorithm to allocate the target positions corresponding to each unmanned equipment; if all unmanned clusters reach all target positions, it means that the task allocation is successful;

[0193] (c) Change the cluster size and target number for multiple tests and calculate the task allocation success rate.

[0194] 3) Cluster execution efficiency

[0195] Test the path distance and execution time required by the cluster to complete the task. The path distance and execution time required by the cluster from the starting point to the target point according to the task allocation result are used as indicators. The shorter the path distance and execution time, the better the performance of the task allocation algorithm. The test steps are as follows:

[0196] (a) Deploy a group of unmanned swarms and set multiple targets. There are randomly distributed obstacles between the unmanned swarms and the targets. The unmanned swarms start from the initial position and reach the target position.

[0197] (b) Calculate the average execution distance and average execution time of the unmanned cluster under this task;

[0198] The unmanned cluster hardware-in-the-loop simulation test and evaluation system described in the present invention transmits real-time data through a virtual simulation module, a hardware computing module and a trusted evaluation module, and comprehensively tests the hardware computing embedded program in an efficient and low-cost manner. Confirm the feasibility of the task through simulation to avoid any risk of collapse. The process of using the system described in the present invention is as follows: 1) Test parameter setting stage: the parameters of the unmanned equipment itself, such as the number of unmanned equipment, maximum flight speed, target perception radius and collision threshold, need to be set; 2) Simulation environment construction stage: build the basic test environment based on the simulation platform, initialize the test environment in the form of hardware-in-the-loop simulation, and set the number and position of static / dynamic obstacles in the environment, the initial position of the cruise missile group and the initial position of the target to be perceived; 3) Test task execution stage: deploy the algorithm for the unmanned cluster to execute the task on the hardware computing platform, and communicate and interact with the simulation and virtual simulation environment in real time; 4) Test task result evaluation stage: evaluate the completion degree and completion efficiency of the task through multiple evaluation indicators of different tasks.

[0199] Example for drones:

[0200] 1. Task-driven in-field hardware-in-the-loop acceleration test

[0201] In view of the problems faced in the comprehensive performance test of unmanned clusters, such as high testing difficulty, high experimental cost and incomplete test information, a task-driven research on unmanned cluster in-field hardware-in-the-loop accelerated testing is carried out. By analyzing the unmanned cluster task process, integrating modules such as ros, gazebo, data transmission system, model algorithm library, etc., an unmanned cluster in-field simulation system that meets the actual task scenario is established. The unmanned cluster in-field simulation hardware platform is built by integrating modules such as ground control station, embedded development board and data transmission terminal. In response to the performance test requirements of unmanned clusters, the in-field hardware-in-the-loop accelerated testing of unmanned clusters is realized by combining the in-field simulation system and the in-field hardware platform, providing experimental support for the comprehensive performance evaluation of unmanned clusters in an all-round and multi-dimensional manner.

[0202] II. UAV Cluster Indoor and Outdoor Field Testing and Evaluation System Architecture

[0203] like Figure 5 As shown in the figure, the system architecture mainly consists of two parts: indoor simulation test and outdoor real machine test, so as to comprehensively and accurately evaluate the performance of unmanned swarm.

[0204] The unmanned swarm mission characteristic test framework is shown in the figure. The unmanned swarm mission execution mainly includes unmanned swarm ground reconnaissance, enemy target locking, target tracking and matching, task allocation, task execution, formation reorganization and return. For different mission links, the corresponding unmanned swarm mission characteristics are extracted, mainly including: track estimation, environmental perception, target detection, situational awareness, collaborative positioning, encirclement and obstacle avoidance, path planning and formation coordination. In order to evaluate the unmanned swarm mission characteristics, evaluation indicators such as trajectory deviation, fusion quality, recognition accuracy, perception consistency, allocation reliability, cluster efficiency, and control accuracy are constructed, and a multi-dimensional, full-process, and in-depth evaluation system is established to achieve effective evaluation of the unmanned swarm mission characteristics.

[0205] The main goal of the indoor simulation test is to evaluate the performance of the unmanned swarm in a controlled environment. This includes but is not limited to the flight performance, navigation system, target recognition and tracking capabilities of the unmanned swarm. Use advanced simulation software to simulate the flight environment and conditions of the unmanned swarm. This method can test the performance of the unmanned swarm in different environments and conditions, and can also simulate various possible failure conditions to evaluate the stability and reliability of the unmanned swarm. The main goal of the outdoor real machine test is to evaluate the performance of the unmanned swarm in the actual environment. This includes but is not limited to the flight performance, navigation system, target recognition and tracking capabilities of the unmanned swarm.

[0206] Conduct real-machine tests at designated test sites to simulate actual mission environments. This method can test the performance of unmanned swarms in actual environments and conditions, and can also evaluate the performance of unmanned swarms in actual mission environments.

[0207] The core functions of the unmanned swarm system include: autonomous mission planning, autonomous navigation, and autonomous perception. Analyze the different functions of the unmanned swarm system, extract the characteristics of autonomous mission planning and autonomous navigation functions based on the swarm platform, including track estimation, path planning, collaborative positioning, etc.; swarm mission characteristics, including autonomous obstacle avoidance, damage assessment, etc. In view of the mission characteristics of unmanned swarms, conduct mission characteristic evaluation research, and realize the test and evaluation of the mission characteristics of unmanned swarms by generating test cases, building test environments, and characterizing mission characteristics. Finally, verify the mission characteristic evaluation process through test evaluation software and internal and external field test systems.

[0208] 3. Task-driven in-field hardware-in-the-loop acceleration test

[0209] (1) Overall architecture of in-field hardware-in-the-loop acceleration testing

[0210] In order to solve the problems of high testing difficulty, high test cost, incomplete test information, etc. faced in testing the comprehensive performance of unmanned swarms, we designed a hardware-in-the-loop virtual test system for unmanned swarms and used a combination of virtual and real methods to test the intelligent characteristics of unmanned swarms. Using hardware-in-the-loop simulation environment testing, we can verify the performance of collaborative detection, collaborative tracking, and collaborative matching of unmanned swarms, and test the comprehensive performance of group path planning, formation control, and formation coordination of unmanned swarms, completing the intelligent characteristics test of the entire process of unmanned swarms. The overall framework is as follows: Figure 6 shown.

[0211] In the mission area, the main tasks performed by the unmanned swarm include flying the unmanned swarm according to the predetermined trajectory, online planning and mission changes. In view of the testing needs of unmanned swarm tasks, research on infield hardware-in-the-loop acceleration testing is carried out. By integrating ROS operating system, 3D visual scene simulation platform, algorithm model library, sensor and communication modules, a simulation environment that maps the real unmanned swarm mission scene is built; by integrating development boards, ground stations, data transmission terminals and other modules, an unmanned swarm infield simulation hardware platform is built. Based on the infield simulation system and hardware platform, unmanned swarm hardware-in-the-loop acceleration test experiments are carried out. Finally, based on multi-dimensional and multi-level evaluation standards and evaluation indicator systems, the results of the unmanned swarm infield hardware-in-the-loop acceleration test are effectively evaluated.

[0212] (2) Multi-platform in-the-loop simulation test framework based on digital modeling

[0213] 1) Overall plan of simulation test

[0214] The unmanned cluster in-field hardware-in-the-loop acceleration test connects the unmanned cluster controller hardware system to the virtual simulation system. Through real-time data transmission between the software controller and the software simulation system, the controller embedded program is fully tested in a highly efficient and low-cost manner. The test architecture is as follows Figure 2 As shown in the figure. Unmanned swarm maneuvering flight tests under hardware-in-the-loop simulation conditions help debug control and guidance algorithms and evaluate the models obtained from system identification. The feasibility of intelligent perception tasks can be confirmed by simulation to avoid any risk of crash. In addition, by arbitrarily reshaping and reusing the simulation environment, experiments involving complex external environments can be simulated.

[0215] The unmanned cluster in-field hardware-in-the-loop acceleration test system consists of a real-time network hub, a reflective memory card, and optical fiber. The real-time simulation network structure is shown in the figure below. The real-time simulation network connects the management computer and the simulation node computer through an optical fiber link to achieve real-time communication between the nodes, realize distributed semi-physical simulation, and have good scalability. The number of simulation node computers can be increased as needed. The simulation node connects the unmanned cluster virtual scene and the virtual unmanned cluster through the serial port. It mainly verifies the intelligent characteristics of the unmanned cluster such as trajectory planning, task allocation, formation control, and collaborative perception. Through the parallel connection of the simulation nodes, real-time interaction with the virtual simulation environment is achieved to realize the distributed real-time simulation test of the unmanned cluster.

[0216] 2) Digital modeling of interior scenes and unmanned swarms based on Gazebo

[0217] The indoor simulation environment is built using Gazebo. The overall control framework of Gazebo includes three main components: simulator, controller and sensor. The simulator is responsible for simulating the dynamics, environment and sensor data of the unmanned swarm; the controller is responsible for designing and implementing various control algorithms to move the unmanned swarm to the designated location and achieve other tasks; the sensor is responsible for acquiring environmental information and feeding it back to the controller. Gazebo provides an unmanned swarm control framework based on physical simulation, which can be used to test different control algorithms to make the unmanned swarm present better performance and reliability. The framework can simulate a variety of unmanned swarms and environments, which allows users to test and optimize in different scenarios and implement different applications. The overall control framework of the simulation platform is as follows Figure 3 shown.

[0218] The blue module is the Gazebo module, the yellow module is the PX4 SITL module, and the light orange module is other modules. This topic focuses on the verification of the perception and motion planning module algorithms. The controller can receive messages from three types of modules: motion planning, state estimation, and ground station. Different operations are performed according to the received messages. At the same time, the simulation platform supports the independent design of the structure of the unmanned cluster, and the sensor module can also be imported to simulate the actual hardware. The perception part of the unmanned cluster mainly involves the autonomous positioning of the unmanned cluster, which mainly obtains the current position of the unmanned cluster through information such as images and point clouds. The motion planning part of the unmanned cluster mainly generates a smooth trajectory based on the current position and the position of the target point, so as to send the target point to the control unit to control the autonomous navigation flight of the unmanned cluster. The controller is responsible for designing and implementing control algorithms to control the unmanned cluster to complete various tasks. Gazebo provides a variety of controllers, such as PID controllers, LQR controllers, and deep learning controllers. In addition, the simulation platform also supports viewing the various status information of the unmanned cluster on the QGground platform to better control the flight of the unmanned cluster.

[0219] The construction of unmanned swarm indoor simulation scenes should have the following characteristics: (1) It can realize the rapid and accurate reconstruction of realistic, multi-type three-dimensional unmanned swarm simulation test scenes. (2) It can realize the simulation of unmanned swarm flight and control process naturally and smoothly. (3) It supports the personalized construction of test scenes. To achieve the above goals, a simulation scheme as shown in the figure below is established. First, remote sensing measurement technology is used to quickly collect high-definition image data and three-dimensional laser point cloud data of various typical real scenes; then, digital geometry processing, computer vision, artificial intelligence and other technologies are used to realize the preprocessing of image and point cloud data, as well as the robust segmentation, accurate recognition and automatic reconstruction of three-dimensional models of major objects in three-dimensional scenes; finally, based on the automatically constructed three-dimensional model of the real scene and the three-dimensional models of various existing virtual auxiliary objects, guided by the needs of training task design, the personalized construction of unmanned swarm indoor simulation test scenes is realized through the virtual-real model fusion technology.

[0220] The data acquisition module is a process in which the deployed intelligent devices collect the real data of their own unmanned swarm through sensors during the unmanned swarm flight in a real environment, and the data transmission process of the communication system outputs the results to the data-driven model after passing through the unmanned swarm control model, presenting a digital model of the unmanned swarm. In order to test the realism of the unmanned swarm digital model, it is evaluated in four aspects: interactivity, security, accuracy and stability. Figure 7 shown.

[0221] Interactivity: It includes control interaction and test interaction. The control interaction consists of the operator controlling the flight process of the unmanned swarm in the virtual scene and the visual feedback results of the virtual scene. The test interaction is the process of the unmanned swarm interacting with other participating objects in the virtual scene test environment.

[0222] Safety: It includes operational safety and information security. Operational safety is to ensure that the operator and the unmanned cluster will not be affected by dangerous factors during the operation process in the system test environment, such as accident risks, distraction, etc.; information security is to prevent the risks of network and communication system intrusion and data leakage.

[0223] Accuracy: includes spatiotemporal consistency and state error. Spatiotemporal consistency means that the digital model of the unmanned swarm must form a consistent relationship with the state of the actual unmanned swarm control process in both time and space dimensions; result error refers to the error in the results presented by the digital model. Some error correction algorithms can be used to reduce the impact of this type of system error so that the final result is within an acceptable range.

[0224] Stability: includes system robustness. System robustness refers to the ability of the system to maintain normal operation under different degrees of network congestion, erroneous operation, overload and fault conditions, ensuring that the system can still perform normal testing under certain environmental pressure.

[0225] (3) Experimental method design and testing based on unmanned swarm mission characteristics

[0226] The design and testing process of the test method based on the unmanned swarm task characteristics are shown in the figure below. The unmanned swarm task characteristics test process includes 5 basic stages. 1) Test parameter setting stage: the parameters of the unmanned swarm itself, such as the number of unmanned swarms, maximum flight speed, target perception radius and unmanned swarm collision threshold, need to be set; 2) Simulation environment construction stage: the basic environment of the test needs to be built, the test environment is initialized in the form of hardware-in-the-loop simulation, and the number and position of static / dynamic obstacles in the environment, the initial position of the unmanned swarm and the initial position of the target to be perceived are set; 3) Test task execution stage: mainly the algorithms of the four major tasks of formation control, trajectory planning, collaborative perception and task allocation are simulated and verified; 4) Test task result evaluation stage: through a variety of evaluation indicators of different tasks, the completion degree and completion efficiency of the task are evaluated; 5) Test result comparison and output stage: execute a variety of different algorithms for the same task, use the same evaluation indicators to compare the execution advantages and disadvantages of different algorithms, and compare and output the final results. Software has been developed based on the present invention.

[0227] Although the preferred embodiments of the present disclosure have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications falling within the scope of the present disclosure.

[0228] Obviously, those skilled in the art can make various changes and modifications to the present disclosure without departing from the spirit and scope of the present disclosure. Thus, if these modifications and variations of the present disclosure fall within the scope of the claims of the present disclosure and their equivalents, the present disclosure is also intended to include these modifications and variations.

[0229] Those skilled in the art will appreciate that the embodiments of the present disclosure may be provided as methods, systems or computer program products. Therefore, the present disclosure may take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware. Moreover, the present disclosure may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0230] The method of the present invention has been verified through simulation experiments and practical applications, and the technical effects claimed by the present invention have been verified. The method proposed by the present invention solves the technical problems proposed by the present invention, and the method of the present invention has been verified through practical applications, and the technical effects and practicability claimed by the present invention have been verified.

[0231] The algorithm (method) proposed in the present invention is the underlying technical core of the present invention, and various products can be derived based on the algorithm.

[0232] The computer program of the developed system (software) is stored on a computer-readable storage medium, and the computer program is configured to implement the steps of the above-mentioned unmanned cluster hardware-in-the-loop simulation test and evaluation system when called by a processor, that is, the present invention is materialized on a carrier to become a computer program product.

[0233] The flowcharts and / or block diagrams of the methods, devices (systems) and computer program products according to the embodiments of the present invention are described. It should be understood that each process and / or block in the flowchart and / or block diagram and the combination of the processes and / or blocks in the flowchart and / or block diagram can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 These computer program instructions can also be stored in a computer-readable memory that can guide a computer or other programmable data processing device to work in a specific way, so that the instructions stored in the computer-readable memory produce a product including an instruction device, which implements the functions specified in the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0234] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0235] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present disclosure rather than to limit its protection scope. Although the present disclosure has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that after reading the present disclosure, those skilled in the art can still make various changes, modifications or equivalent substitutions to the specific implementation methods of the invention, but these changes, modifications or equivalent substitutions are all within the protection scope of the disclosed claims to be approved.

Claims

1. An unmanned cluster hardware-in-the-loop simulation test and evaluation system, characterized in that: The test and evaluation system includes a virtual simulation module, a hardware computing module and a trusted evaluation module; wherein the virtual simulation module is mainly used to build the task execution scenario of the unmanned cluster and display the unmanned cluster task execution process; the hardware computing module is used to execute the deployed unmanned cluster algorithm; the trusted evaluation module is used to evaluate the execution results of the unmanned cluster algorithm; The virtual simulation module and the hardware computing module, as well as the hardware computing module and the trusted evaluation module communicate with each other and complete real-time data transmission; In the virtual simulation module, the virtual unmanned cluster senses environmental information through sensors and transmits it to the hardware computing module (hardware computing platform). The unmanned cluster algorithm deployed on the hardware computing module (hardware computing platform) performs calculations based on the sensed environmental information, obtains the action instructions or planning instructions of the unmanned cluster and transmits them to the virtual simulation module; the virtual unmanned cluster in the virtual simulation module executes the action instructions or planning instructions in the simulation environment; The virtual unmanned cluster in the virtual simulation module transmits its own position coordinates, movement speed and target information to the trusted evaluation module through real-time interaction. The trusted evaluation module evaluates the execution effect of the unmanned cluster algorithm.

2. The unmanned cluster hardware-in-the-loop simulation test and evaluation system according to claim 1, characterized in that: It is suitable for virtual simulation and test evaluation of drones, unmanned vehicles and unmanned ships.

3. The unmanned cluster hardware-in-the-loop simulation test and evaluation system according to claim 1 or 2, characterized in that: The simulation environment in the virtual simulation module is built by Gazebo and Unity simulation platforms, and specifically includes virtual scenes, virtual dynamic and static obstacles, virtual target locations or subjects, and virtual unmanned cluster equipment.

4. The unmanned cluster hardware-in-the-loop simulation test and evaluation system according to claim 3 is characterized in that: The hardware computing module is composed of one or more sets of hardware boards, and the algorithm is deployed on the hardware boards to execute the unmanned cluster algorithm.

5. The unmanned cluster hardware-in-the-loop simulation test and evaluation system according to claim 1, characterized in that: The system can evaluate the formation control, path planning, collaborative perception and task allocation of unmanned swarms; The evaluation indicators of formation control include: control success rate, formation efficiency, formation stability, formation coordination accuracy and control effectiveness; the evaluation indicators of path planning include: planning success rate, execution time, execution distance and planning efficiency; the evaluation indicators of collaborative perception include: perception accuracy, tracking accuracy and perception timeliness; the evaluation indicators of task allocation include: allocation real-time, allocation effectiveness and cluster execution efficiency.

6. The unmanned cluster hardware-in-the-loop simulation test and evaluation system according to claim 5, characterized in that: The test process of the evaluation indicators related to formation control, path planning, collaborative perception and task allocation is as follows: (1) Formation control evaluation indicators 1) Control success rate Evaluate the control of the unmanned swarm formation in various scenarios, using the formation control success rate as the evaluation indicator. The higher the formation control success rate, the stronger the formation's obstacle avoidance capability. The test steps are as follows: (a) Deploy a group of unmanned swarms, set up multiple randomly distributed static obstacles and multiple dynamic obstacles, and let the swarms reach the target position from the initial position; (b) Dynamic obstacles move randomly to test the unmanned swarm formation’s ability to avoid obstacles and count the number of unmanned equipment that successfully reach the target point; (c) Conduct multiple experiments to calculate the success rate of formation control of the cluster under the formation obstacle avoidance algorithm; 2) Formation efficiency Test the cluster's ability to efficiently form a team, using the number of iterations and path distance as evaluation indicators. The shorter the calculation time and the smaller the average path distance, the higher the formation efficiency. The test steps are as follows: (a) Deploy a group of unmanned swarms, set the starting position and target position of each unmanned equipment in the swarm, and determine the swarm mission; (b) Use the formation control algorithm to calculate the path trajectory of each unmanned equipment under the swarm task, record the time required for the algorithm to generate the path trajectory and the total path length of the drone group; (c) Multiple experiments, calculate the distance between calculations and the average path distance under multiple experiments of the algorithm; 3) Formation stability The formation execution status is used as an evaluation indicator to test the formation stability. The test steps are as follows: (a) Deploy a group of unmanned swarms, set the starting position and target position of each unmanned equipment in the swarm, and determine the swarm mission; (b) Use the formation algorithm to form a formation, and record the speed, altitude, heading angle and other information between unmanned clusters every second during the execution process; (c) Observe the formation control situation by the change of unmanned cluster information over time. If the change of unmanned cluster information over time tends to be stable, it is considered that the formation stability is high; (d) After the formation is completed, the size of the unmanned cluster is increased or decreased, and the formation of the cluster is observed through a line graph of the unmanned cluster information changing over time. If the final line graph still tends to be stable, it is considered that the re-formation is successful and the formation is relatively stable; 4) Formation coordination accuracy This indicator measures the maximum number of movement commands output per minute during formation coordination control of the unmanned swarm, and is used to reflect stability and sensitivity. The test steps are as follows: (a) In the formation control task, for each unmanned equipment, record the number of commands output by the formation control algorithm per minute for maintaining formation coordination; (b) The maximum number of movement commands output by all unmanned equipment is the formation coordination accuracy; 5) Control effectiveness Test whether the formation can be effectively carried out in certain scenarios. The test steps are as follows: (a) Deploy a group of unmanned swarms, start and execute the formation control algorithm at different locations, and the unmanned swarms execute the trajectory planned by the algorithm; (b) Record the trajectory of the unmanned swarm from the beginning to the completion of the formation, and check whether the formation is completed and the ability to maintain the formation based on the swarm trajectory; (2) Path planning evaluation indicators 1) Planning success rate Test the rate at which the unmanned swarm successfully completes its navigation mission. The test steps are as follows: (a) Deploy a group of unmanned swarms, set up multiple randomly distributed static obstacles and multiple dynamic obstacles, and let the swarms reach the target position from the initial position; (b) Count the number of tasks performed in a complex scenario; (c) Record the number of successful completions of cluster tasks, that is, the number of times the unmanned cluster successfully reaches the target point in complex scenarios; (d) Calculate the success rate percentage; 2) Execution time The average time, maximum time, and minimum time it takes for an unmanned cluster to complete a task. The test steps are as follows: (a) Count the total time it takes for the unmanned swarm to execute all tasks; (b) Divide the total time by the total number of tasks to get the average execution time; average execution time = (total execution time of all tasks) / (total number of tasks); (c) Find the longest execution time, which is the longest execution time; (d) Find the shortest execution time among them, which is the shortest execution time; 3) Execution distance The average distance, maximum distance, and minimum distance that the unmanned swarm takes to complete its mission. The test steps are as follows: (a) Count the total execution distance of all navigation tasks of the unmanned swarm; (b) Divide the total execution distance by the total number of navigation tasks to obtain the average navigation distance; (c) Find the largest execution distance, which is the maximum execution distance; (d) Find the smallest execution distance, which is the minimum execution distance; 4) Planning efficiency The indicator used to evaluate the efficiency of the path planning algorithm in the unmanned swarm intelligent decision-making algorithm. The shorter the planning time, the higher the efficiency of the intelligent decision-making algorithm. The test steps are as follows: (a) In the path planning task, the time it takes for the algorithm to generate the final valid path is recorded; the time usually refers to the time from the start of path planning to the final acquisition of a feasible path; (b) taking the recorded planning time as the value of planning time; (3) Evaluation indicators of collaborative perception 1) Perception accuracy Perceptual accuracy refers to the proportion of all detected targets that are correctly identified as targets. The higher the perceptual accuracy, the higher the accuracy of the system in correctly identifying targets. The test steps are as follows: (a) Deploy a group of unmanned swarms, set multiple targets, and use the unmanned swarm intelligent perception algorithm to identify and detect the targets; (b) Calculate the number of true positives (TP) and false positives (FP) based on the recognition detection results. True positives refer to the number of targets that are correctly detected and recognized, and false positives refer to the number of non-targets that are mistakenly identified as targets. (c) Calculate the perception accuracy (Precision) = TP / (TP+FP), where TP is the number of true positives and FP is the number of false positives; 2) Tracking accuracy Tracking accuracy refers to the degree of deviation between the unmanned cluster and the actual position of the target during the tracking process. The higher the tracking accuracy, the smaller the error between the unmanned cluster and the actual position when tracking the target. The test steps are as follows: (a) For each target tracking process, record the target’s true position and the position information output by the unmanned cluster; (b) Calculate the absolute error of each target tracking process, that is, the distance difference between the target's true position and the position output by the unmanned cluster; (c) Sum the absolute errors of all target tracking processes; (d) Divide the total absolute error value by the number of targets to obtain the tracking accuracy (MAE); MAE = Σ|true position - tracker output position| / number of targets; 3) Perceived timeliness Perception timeliness refers to the difference between the time when the unmanned swarm collaborative perception system detects and identifies a target and the time when the target actually appears or changes. It indicates the system's response speed to changes in the target. The test steps are as follows: (a) During the operation of the collaborative perception system, the actual appearance time of each target and the time when the system issues a warning signal are recorded; (b) Calculate the time delay for each target, that is, the time when the target appears minus the time when the system issues the warning signal; (c) Statistical analysis of the time delays of all targets can be performed to calculate indicators such as average time delay, maximum time delay, and minimum time delay; (4) Task allocation evaluation indicators 1) Real-time distribution The test is whether the cluster tasks can be given task instructions in a timely and effective manner, with the task allocation time as the evaluation indicator. The shorter the time it takes for the algorithm to complete the task allocation, the higher the real-time performance of the task allocation. The test steps are as follows: (a) Deploy a group of unmanned clusters, set multiple targets, and the task allocation algorithm generates a task allocation route plan for the unmanned cluster according to the requirements. (b) The time required for the test task allocation algorithm to complete the allocation under this condition is the real-time performance of the allocation; 2) Distribution effectiveness Test whether the unmanned cluster algorithm can effectively assign tasks to the target. The target assignment success rate is used as an evaluation indicator. The higher the assignment success rate, the better the assignment effectiveness. The test steps are as follows: (a) Deploy a group of unmanned swarms and set multiple targets. There are randomly distributed obstacles between the unmanned swarms and the targets. The unmanned swarms start from the initial position and reach the target position. (b) Using the unmanned cluster task allocation algorithm to allocate the target positions corresponding to each unmanned equipment; if all unmanned clusters reach all target positions, it means that the task allocation is successful; (c) Change the cluster size and target number for multiple tests and calculate the success rate of task allocation; 3) Cluster execution efficiency The path distance and execution time required for the test cluster to complete the task are used as indicators. The shorter the path distance and execution time, the better the performance of the task allocation algorithm. The test steps are as follows: (a) Deploy a group of unmanned swarms and set multiple targets. There are randomly distributed obstacles between the unmanned swarms and the targets. The unmanned swarms start from the initial position and reach the target position. (b) Calculate the average execution distance and average execution time of the unmanned cluster under this task.

7. The unmanned cluster hardware-in-the-loop simulation test and evaluation system according to claim 1, characterized in that: The system is used to conduct comprehensive testing of hardware computing embedded programs, to confirm the feasibility of the task through simulation, and to avoid any risk of crash.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and the computer program is configured to implement the steps of an unmanned cluster hardware-in-the-loop simulation test and evaluation system according to any one of claims 1 to 6 when called by a processor.

Images