Unmanned test method and system based on ground mobile platform

Abstract

The invention provides an unmanned test method and system based on a ground mobile platform, and belongs to the technical field of unmanned detection. The method comprises the following steps: constructing a ground mobile platform, and integrating a multi-modal sensor, a test instrument and the ground mobile platform by using a consistency interface protocol; in combination with data collected by a multi-modal sensor, real-time construction of a scene map and path planning of a ground mobile platform are respectively carried out by adopting a map real-time construction algorithm and a path planning navigation algorithm, so that the autonomous test capability of a test instrument is enhanced, and the execution efficiency of a complex test task is improved; according to the constructed scene map and path, the ground mobile platform is controlled to arrive at the test area through voice recognition control and gesture recognition, and the test instrument on the mobile platform is controlled to complete the test operation, so that the man-machine non-contact intelligent collaborative test capability of the test instrument in the complex and dangerous environment is further enhanced; the purpose of autonomous testing of the testing instrument is further achieved.

Description

Technical Field

[0001] The present invention belongs to the technical field of unmanned detection, and in particular relates to an unmanned testing method and system based on a ground mobile platform. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] Under the guidance of the innovation of advanced artificial intelligence technology, various types of unmanned intelligent equipment have been widely used in various industries, and their task execution efficiency and accuracy have been greatly improved. In the traditional test and assurance model, the test process mostly relies on manual operation, and the deployment and movement of test instruments mainly rely on manpower, lacking the ability of autonomous movement and autonomous operation. Testing in outdoor and dangerous environments is limited, and it cannot meet the development of unmanned intelligent equipment and the demand for unmanned, fast and accurate testing and assurance in new combat modes. For example, when using complex electromagnetic environments to complete the anti-interference capability assessment task of electronic countermeasure equipment, the signal environment in the test field is mainly described by the number of electromagnetic pulses, and there is a lack of more sophisticated evaluation methods for electromagnetic signals, such as evaluation in the time, frequency, energy, modulation and other domains. The current test method mainly relies on a single monitoring receiver to obtain the signal characteristics of a single fixed position. It is difficult to compare and analyze the signal characteristics of different positions and directions of the test site at the same time. If it needs to be moved, it needs to be moved and fixed by manpower. If multiple monitoring receivers are deployed in different positions to simultaneously measure the environmental noise level, frequency band occupancy, time occupancy, signal field strength, signal type, spectrum density, etc., the effectiveness of the jammer / jammer combination can be fully evaluated. However, the deployment of multiple monitoring receivers often requires manual handling, and the traditional method cannot be flexibly deployed according to the changes in the on-site electromagnetic environment. At the same time, for large and heavy desktop instruments, there is sometimes a risk of injury to the movers. Secondly, the operation and parameter setting of the test instrument require professional personnel to operate in person. The test instrument generally contains a large number of operation buttons, and its operation method is single, the operation process is cumbersome, and the test efficiency is low. Summary of the invention

[0004] In order to overcome the above-mentioned deficiencies of the prior art, the present invention provides an unmanned testing method and system based on a ground mobile platform.

[0005] To achieve the above objectives, one or more embodiments of the present invention provide the following technical solutions:

[0006] The first aspect of the present invention provides an unmanned testing method and system based on a ground mobile platform;

[0007] An unmanned testing method based on a ground mobile platform, comprising:

[0008] Build a ground mobile platform and integrate multimodal sensors and test instruments with the ground mobile platform using a consistent interface protocol;

[0009] In combination with the data collected by the multimodal sensor, a real-time map construction algorithm and a path planning navigation algorithm are used to respectively perform real-time construction of a scene map and path planning of a ground mobile platform;

[0010] According to the constructed scene map and path, voice recognition is used to control the ground mobile platform to reach the test area.

[0011] Or, use gesture recognition to control the ground mobile platform to reach the test area;

[0012] Control the test equipment on the mobile platform to reach the test area to complete the test operation.

[0013] As a further technical solution, the consistent interface protocol is: the data from multimodal sensors, test instruments and ground platforms are uniformly processed by the onboard processor, the onboard processor is configured with Ubuntu and ROS systems, and the Ubuntu and ROS systems establish a consistent interface protocol for data transmission and data processing.

[0014] As a further technical solution, a signal monitoring unit and a signal positioning unit are provided in the test instrument.

[0015] As a further technical solution, the process of using the real-time map construction algorithm to construct the scene map in real time is as follows:

[0016] The A-LOAM algorithm is used to segment ground points and optimize the feature extraction process to obtain more reliable feature points for inter-frame matching. The loop detection module is introduced to further suppress error accumulation.

[0017] Use radar and inertial sensors to perceive the surrounding environment and obtain environmental data; fuse radar data with inertial sensor data to reduce cumulative errors.

[0018] As a further technical solution, a path planning navigation algorithm is used to plan the path of the ground mobile platform, specifically:

[0019] The path planning of the ground mobile platform includes global path planning and local obstacle avoidance path planning;

[0020] The global path planning uses the A* algorithm to obtain the correct shortest path in a short time;

[0021] The local obstacle avoidance path planning adopts the VFH algorithm, establishes a binary histogram and changes the obstacle expansion range so that the mobile platform can maintain a safe distance and complete local obstacle avoidance.

[0022] As a further technical solution, the use of speech recognition to control the ground mobile platform to reach the test area is specifically as follows: transmitting the sound data acquired by the microphone array to the onboard processor, and preprocessing the collected speech signal;

[0023] Then the preprocessed speech signal is converted into a feature vector, and the feature vector is input into the funasr speech recognition model for recognition to obtain text information;

[0024] Match the control instructions of the ground mobile platform according to the text information to realize voice control of the mobile platform.

[0025] As a further technical solution, the use of gesture recognition to control the ground mobile platform to reach the test area is specifically: the image data acquired by the high-definition camera is transmitted to the airborne processor, the gesture image in the image data is recognized based on the YOLO recognition algorithm, the gesture is matched to the control command of the mobile platform, and the control of the mobile platform by gesture is realized.

[0026] A second aspect of the present invention provides an unmanned testing system based on a ground mobile platform.

[0027] An unmanned test system based on a ground mobile platform, comprising:

[0028] The ground mobile platform building module is configured to: build a ground mobile platform and integrate the multi-modal sensor and the test instrument with the ground mobile platform using a consistent interface protocol;

[0029] The map construction and path planning module is configured to: combine the data collected by the multimodal sensor, use the real-time map construction algorithm and the path planning navigation algorithm to respectively perform real-time construction of the scene map and path planning of the ground mobile platform;

[0030] The human-computer interaction module is configured to: control the ground mobile platform to reach the test area using voice recognition according to the constructed scene map and path,

[0031] Or, use gesture recognition to control the ground mobile platform to reach the test area;

[0032] The test module is configured to control the test instrument on the mobile platform that arrives at the test area to complete the test operation.

[0033] A third aspect of the present invention provides a computer-readable storage medium having a program stored thereon, which, when executed by a processor, implements the steps of an unmanned testing method based on a ground mobile platform as described in the first aspect of the present invention.

[0034] The fourth aspect of the present invention provides an electronic device, including a memory, a processor, and a program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps of an unmanned testing method based on a ground mobile platform as described in the first aspect of the present invention are implemented.

[0035] One or more of the above technical solutions have the following beneficial effects:

[0036] (1) The present invention integrates a ground platform with strong autonomous mobility capabilities with traditional test instruments, and combines real-time map construction algorithms and path planning and navigation algorithms to enhance the autonomous testing capabilities of the test instruments, making them more flexible and autonomous in completing test tasks, thereby improving the execution efficiency of complex test tasks (especially in field test tasks).

[0037] (2) The present invention uses speech recognition algorithms and gesture recognition algorithms to enhance the human-machine contactless intelligent collaborative testing capabilities of test instruments in complex and dangerous environments through consistent mapping of multimodal semantic understanding such as speech and gestures with test instrument parameters, and further achieves the goal of autonomous testing of test instruments.

[0038] Advantages of additional aspects of the present invention will be given in part in the following description, and in part will become obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0040] Figure 1 This is a flow chart of the method of the first embodiment.

[0041] Figure 2 This is the overall architecture diagram of the unmanned test of the ground mobile platform in the first embodiment.

[0042] Figure 3 It is a system structure diagram of the second embodiment. DETAILED DESCRIPTION

[0043] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.

[0044] It should be noted that the terms used herein are for describing specific embodiments only and are not intended to be limiting of exemplary embodiments according to the present invention.

[0045] In the absence of conflict, the embodiments of the present invention and the features of the embodiments may be combined with each other.

[0046] In the traditional testing process, the deployment and movement of test instruments mainly rely on manpower, and testing is limited in the wild and dangerous environments. In particular, when performing field electromagnetic environment monitoring tasks, multiple test instruments need to be flexibly deployed in a distributed manner. Traditional methods cannot perform optimal and flexible deployment based on changes in the on-site electromagnetic environment. For large and heavy desktop instruments, there is sometimes a risk of injury to personnel who move them; in addition, instrument operation and parameter setting mainly rely on professional personnel to operate in person. The operation method is single and the process is cumbersome, making the test efficiency low. Based on the above problems, the present application provides an unmanned testing method and system based on a ground mobile platform, which is described in detail below in conjunction with specific embodiments.

[0047] Embodiment 1

[0048] This embodiment discloses an unmanned testing method based on a ground mobile platform;

[0049] like Figure 1 and Figure 2 As shown, an unmanned testing method based on a ground mobile platform includes:

[0050] Step S1, constructing a ground mobile platform, and integrating a multimodal sensor, a test instrument and the ground mobile platform using a consistent interface protocol;

[0051] In step S1, the constructed ground mobile platform can be wheeled or tracked, and needs to have certain stability, durability and mobility, and can carry multimodal sensors and test instruments. Among them, the multimodal sensors and test instruments are integrated on the ground mobile platform through a consistent interface protocol. Specifically, the integration of multimodal sensors, test instruments and ground mobile platforms must ensure the real-time and reliability of data transmission, and the consistency of data format and interface calling method. The data of multimodal sensors, test instruments and ground platforms are uniformly processed by the airborne processor. The airborne processor is configured with Ubuntu and ROS systems. Ubuntu and ROS systems establish a consistent interface protocol for data transmission and data processing. Data transmission is performed by defining subscribers and publishers of the ROS system. The subscriber requests to connect to the publisher and transmits the subscribed topic name, message type and communication protocol (TCP / UDP). After receiving the request, the publisher confirms the connection information to the subscriber and sends its own TCP / UDP address information. The subscriber establishes a TCP / UDP connection with the publisher based on the publisher's TCP / UDP address information. After the connection is established, the publisher begins to publish topic message data to the subscriber. The subscriber receives the message and processes the data through the callback function. The perception data of multimodal sensors, the processing data of test instruments, and the mobile platform data are all shared through ROS system nodes, and the corresponding module data can be obtained by subscribing to the corresponding topics.

[0052] In this embodiment, the multimodal sensor includes a laser radar, a high-definition camera, an inertial measurement unit, a Beidou navigation system and a microphone array. The laser radar can perform distance measurement and terrain mapping in real time; the high-definition camera is used to capture information such as the surrounding environment, faces, gestures, etc.; the inertial measurement unit and the Beidou navigation system are used to capture the real-time position and posture information of the mobile platform.

[0053] Step S2, combining the data collected by the multimodal sensor, using a real-time map construction algorithm and a path planning navigation algorithm to respectively perform real-time construction of a scene map and path planning of a ground mobile platform;

[0054] Real-time construction of scene maps is the process of sensors sensing the environment and visualizing it. In this embodiment, the real-time construction algorithm uses the A-LOAM algorithm, which optimizes the feature extraction process by segmenting ground points to obtain more reliable feature points for inter-frame matching. The A-LOAM algorithm includes point cloud dedistortion, feature extraction and matching, inter-frame pose estimation, etc. to reduce elevation errors.

[0055] Among them, point cloud dedistortion is to correct the point cloud distortion caused by radar movement. The process is as follows:

[0056] Assume that the 3D laser radar pose at the start time 0 of a frame of point cloud is known to be:

[0057]

[0058] r is the rotation amount, t is the translation amount, and the 3D laser radar pose at the i-th moment is:

[0059]

[0060] The relative position of the radar at two moments is:

[0061]

[0062] Where: T0 is the posture at time 0, T i is the posture at time i, and R is the relative posture between two moments.

[0063] The initial point coordinates of the 3D laser radar point cloud are:

[0064] P0=[p x p y p z ];

[0065] Then the coordinates of the 3D laser radar point cloud at the i-th moment are:

[0066] P i =RP0;

[0067] The 3D LiDAR point cloud is expressed in local coordinates. Any position can be regarded as the starting position T0, and the starting position T0 can be regarded as the unit matrix. At this time, we only need to obtain the position T at time i. i , that is, to obtain r i and t i .

[0068] The relationship between rotation, translation and posture is:

[0069] r i =wd time ;

[0070] t i =Vd time ;

[0071] The average speed and average angular velocity per unit time are used to obtain the relative posture, where w is the average angular velocity, v is the average speed, and d time As unit time.

[0072] Feature extraction: Scan the latest frame of point cloud, calculate the curvature of the frame of point cloud in the surrounding area, and judge the features of the points by the curvature. After sorting the curvature, define the points with curvature greater than a certain threshold as edge point features, otherwise define them as plane point features. Assume that at time k is the set of edge points, is a set of plane points, and the feature point set F k :

[0073]

[0074] The curvature calculation formula is:

[0075]

[0076] Where N represents the set of surrounding points on the same line as point i. is the coordinate of point i, is the coordinate of point j.

[0077] Feature matching: The previous part extracted the features of the point cloud. Next, we match the feature points. By matching between frames and between frames and local maps, we calculate the pose transformation between point clouds and map the pose transformation into a distance minimization problem.

[0078] Assume that the edge point set scanned in the k+1th frame is Select a point i from among them, and the set of edge points scanned in the kth frame is Select a point j that is closest to i in this set, and select a point l that is closest to j on the scan line adjacent to j.

[0079] Therefore, the pose transformation is mapped into a distance minimization problem, and the straight line distance from point i to point j and point l is expressed as:

[0080]

[0081] For plane point matching, similar to edge point matching, the plane point set of the k+1th frame Select a point i from this set and assume the plane point set of the kth frame In this set, select a point j that is closest to i, select a point l on the same line as j, and select a point m on the scan line adjacent to j. The distance from point i to the plane is calculated by the following formula:

[0082]

[0083] Inter-frame pose estimation: Using the feature points in the point cloud as constraints, the pose calculation is transformed into a distance minimization problem. The distance error equation is constructed based on the feature point matching constraints, and the total loss function for the line and surface constraints is:

[0084]

[0085] Obtained by the following formula

[0086]

[0087] The iteration is performed using the Levenberg-Marquardt Algorithm (LM) as follows:

[0088]

[0089] Where:

[0090]

[0091] Where T k+1 is the pose at time k+1. J is the Jacobian matrix, and λ is the damping factor used to adjust the convergence speed and stability of the algorithm.

[0092] Assuming that the laser radar moves at a uniform speed in this scanning frame, the current pose is solved, and then the pose of each laser point in this frame can be inferred by interpolation, which can be used to match the map and establish the point cloud at the current moment.

[0093]

[0094] In addition, a loop detection module is added to further suppress error accumulation. The loop detection module converts the LiDAR point cloud data into descriptors, which can reflect the line and surface features of the point cloud. After extracting the descriptors, the similarity between the current frame and the historical frame descriptors is compared. When the similarity between the descriptors of the two frames is high, consistency verification, geometric constraints and other methods are used to confirm whether a loop is formed. If a loop is formed, the robot's posture is adjusted to reduce the accumulated error.

[0095] Use two-dimensional laser radar, three-dimensional laser radar, inertial sensors, etc. to perceive the surrounding environment and obtain environmental data. Fuse radar data with inertial sensor data to reduce cumulative errors. The laser radar sensor perceives data as point clouds. According to the operating principle of the laser radar, each frame of the point cloud of the radar in motion will be distorted. First, remove the motion distortion of the point cloud, and then extract features from the point cloud. According to the different curvatures of the point cloud, extract the feature points of the point cloud, construct point-line constraints and line-surface constraints for feature points, and perform inter-frame matching and pose estimation. The inertial sensor data is pre-integrated, inertial solution, and state prediction, and then information is fused with the laser radar through the extended Kalman filter. Finally, the accumulated error is eliminated through loop detection, the loop points are closed, and the local map is not incremented to the global map.

[0096] For the path planning of the ground mobile platform, in this embodiment, a path planning navigation algorithm is used to perform path planning of the ground mobile platform, including using the A* algorithm to obtain the correct shortest path in a short time to achieve global path planning of the ground mobile platform. The A* algorithm has a search direction and can obtain the correct shortest path in a relatively short time without traversing all nodes. The A* algorithm constructs a cost function based on the heuristic function and guides the search process by estimating the cost from the current node to the target node. The calculation formula of the node cost value is:

[0097] F = G + H;

[0098] In the formula, F represents the cost of the current node, G represents the cost from the starting point to the current node, and H represents the cost from the current node to the target node. There are many ways to estimate H. Assuming the coordinates of the starting point are (x g ,y g ), the end point coordinate is (x n ,y n ). The most common method is the Manhattan method, which only considers the horizontal and vertical distances when estimating the distance between two points as shown below:

[0099] h(n)=|x g -x n |+|y g -y n |;

[0100] Or use the Euclidean distance, that is, the shortest line length between two points as shown below.

[0101]

[0102] This strategy enables the A* algorithm to give priority to paths that are more likely to reach the target node during the search process, thereby reducing unnecessary searches and achieving higher efficiency.

[0103] In addition, the VFH algorithm is also used to complete the local obstacle avoidance path planning for local obstacle avoidance. The VFH algorithm enables the mobile platform to maintain a safe distance by establishing a binary histogram and changing the obstacle expansion range, thus completing the local obstacle avoidance path planning for local obstacle avoidance.

[0104] Step S3, based on the constructed scene map and path, use voice recognition to control the ground mobile platform to reach the test area,

[0105] Or, use gesture recognition to control the ground mobile platform to reach the test area;

[0106] The microphone array, high-definition camera and ground mobile platform are all connected to the airborne processing, and the sound data and image data are transmitted to the airborne processor for processing by the consistent interface protocol. Based on this, in the process of using voice recognition to control the ground mobile platform to reach the test area, the sound data obtained by the microphone array is first transmitted to the airborne processor to pre-process the collected voice signal;

[0107] Then the preprocessed speech signal is converted into a feature vector, and the feature vector is input into the funasr speech recognition model for recognition to obtain text information;

[0108] Match the control instructions of the ground mobile platform according to the text information to realize voice control of the mobile platform.

[0109] In the process of using gesture recognition to control the ground mobile platform to reach the test area, the image data obtained by the high-definition camera is transmitted to the onboard processor, the gesture image in the image data is recognized based on the YOLO recognition algorithm, and the gesture is matched to the control command of the mobile platform to realize the control of the mobile platform by gesture.

[0110] The YOLO algorithm process is:

[0111] The input image is divided into S×S grids, each of which is responsible for predicting whether there is an object in the grid and the category and location information of the object. Each grid predicts multiple bounding boxes, as well as the confidence and category probability of these bounding boxes. The non-maximum suppression algorithm is used to remove redundant bounding boxes to obtain the final object detection result.

[0112] YOLO's loss function consists of three parts: coordinate loss, confidence loss, and category probability loss. These three losses together measure the accuracy of the model in predicting bounding box locations, confidence, and category probabilities.

[0113] Coordinate loss function:

[0114] L coord =∑[(x pred -x true ) 2 +(y pred -y true ) 2 +(w pred -w true ) 2 +(h pred -h true ) 2 ];

[0115] Among them, (x pred ,y pred ,w pred ,hpred ) represents the center coordinates and width and height of the predicted bounding box, (x true ,y true ,w true ,h true ) represents the center coordinates and width and height of the real bounding box.

[0116] Confidence loss function:

[0117] L conf =-∑[t·log(pred)+(1-t)·log(1-pred)];

[0118] Among them, t represents the true label (1 if it contains the target, and 0 if it does not), and pred represents the prediction confidence.

[0119] Category probability loss function:

[0120] L class =-∑(c true ·log(c pred ));

[0121] Among them, c true represents the true category probability distribution, c pred Represents the predicted class probability distribution.

[0122] Step S4, controlling the testing instrument on the mobile platform that arrives at the testing area to complete the testing operation;

[0123] The test instrument is equipped with a signal monitoring unit and a signal positioning unit. With the help of various sensors and receivers, it is used to capture signals in the environment, sense signals within a specific frequency range, and convert them into electrical signals for processing. By measuring the strength of the signal, the existence of the signal and its approximate strength can be determined. Using tools such as spectrum analyzers, the signal is decomposed into components of different frequencies to identify its spectral characteristics, and the captured signal is matched with known signal patterns to determine its type or source. Signal recognition is performed by storing and retrieving known signal patterns through a signal feature database or recognition algorithm. Signal positioning is the process of determining the location of the signal source, and the signal propagation characteristics and measurement data are used to determine the location of the signal source.

[0124] Embodiment 2

[0125] This embodiment discloses an unmanned testing system based on a ground mobile platform;

[0126] like Figure 3 As shown, an unmanned test system based on a ground mobile platform includes:

[0127] A ground mobile platform building module is configured to: integrate multimodal sensors and test instruments with a ground mobile platform using a consistent interface protocol;

[0128] The map construction and path planning module is configured to: combine the data collected by the multimodal sensor, use the real-time map construction algorithm and the path planning navigation algorithm to respectively perform real-time construction of the scene map and path planning of the ground mobile platform;

[0129] The human-computer interaction module is configured to: control the ground mobile platform to reach the test area using the speech recognition algorithm and the gesture recognition algorithm according to the constructed scene map and path;

[0130] The test module is configured to complete the test operation using the test instrument on the ground mobile platform.

[0131] Embodiment 3

[0132] The purpose of this embodiment is to provide a computer-readable storage medium.

[0133] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of an unmanned testing method based on a ground mobile platform as described in Example 1 of the present disclosure.

[0134] Embodiment 4

[0135] The purpose of this embodiment is to provide an electronic device.

[0136] An electronic device comprises a memory, a processor and a program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps of an unmanned testing method based on a ground mobile platform as described in Example 1 of the present disclosure are implemented.

[0137] The steps involved in the apparatuses of the above embodiments 2, 3 and 4 correspond to the method embodiment 1, and the specific implementation methods can refer to the relevant description part of embodiment 1. The term "computer-readable storage medium" should be understood as a single medium or multiple media including one or more instruction sets; it should also be understood to include any medium that can store, encode or carry an instruction set for execution by a processor and enable the processor to execute any method in the present invention.

[0138] Those skilled in the art should understand that the modules or steps of the present invention described above can be implemented by a general-purpose computer device, or alternatively, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. The present invention is not limited to any specific combination of hardware and software.

[0139] Although the above describes the specific implementation mode of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solution of the present invention without creative work are still within the scope of protection of the present invention.

Claims

1. An unmanned testing method based on a ground mobile platform, characterized in that: include: Build a ground mobile platform and integrate multimodal sensors and test instruments with the ground mobile platform using a consistent interface protocol; In combination with the data collected by the multimodal sensor, a real-time map construction algorithm and a path planning navigation algorithm are used to respectively perform real-time construction of a scene map and path planning of a ground mobile platform; According to the constructed scene map and path, voice recognition is used to control the ground mobile platform to reach the test area. Or, use gesture recognition to control the ground mobile platform to reach the test area; Control the test equipment on the mobile platform to reach the test area to complete the test operation.

2. The unmanned testing method based on a ground mobile platform as claimed in claim 1, characterized in that: The consistent interface protocol is: the data from the multimodal sensors, test instruments and ground platforms are uniformly processed by the onboard processor, the onboard processor is configured with Ubuntu and ROS systems, and the Ubuntu and ROS systems establish a consistent interface protocol for data transmission and data processing.

3. The unmanned testing method based on a ground mobile platform as claimed in claim 1, characterized in that: The test instrument is provided with a signal monitoring unit and a signal positioning unit.

4. The unmanned testing method based on a ground mobile platform as claimed in claim 1, characterized in that: The real-time map construction algorithm is used to construct the scene map in real time, specifically: The A-LOAM algorithm is used to segment ground points and optimize the feature extraction process to obtain more reliable feature points for inter-frame matching. The loop detection module is introduced to further suppress error accumulation. Use radar and inertial sensors to perceive the surrounding environment and obtain environmental data; fuse radar data with inertial sensor data to reduce cumulative errors.

5. The unmanned testing method based on a ground mobile platform as claimed in claim 1, characterized in that: The path planning and navigation algorithm is used to plan the path of the ground mobile platform, specifically: The path planning of the ground mobile platform includes global path planning and local obstacle avoidance path planning; The global path planning uses the A* algorithm to obtain the correct shortest path in a short time; The local obstacle avoidance path planning adopts the VFH algorithm, establishes a binary histogram and changes the obstacle expansion range so that the mobile platform can maintain a safe distance and complete local obstacle avoidance.

6. The unmanned testing method based on a ground mobile platform as claimed in claim 1, characterized in that: The method of using voice recognition to control the ground mobile platform to reach the test area is specifically as follows: The sound data acquired by the microphone array is transmitted to the onboard processor to pre-process the collected speech signal; Then the preprocessed speech signal is converted into a feature vector, and the feature vector is input into the funasr speech recognition model for recognition to obtain text information; Match the control instructions of the ground mobile platform according to the text information to realize voice control of the mobile platform.

7. The unmanned testing method based on a ground mobile platform as claimed in claim 1, characterized in that: The method of controlling the ground mobile platform to reach the test area by using gesture recognition is specifically as follows: The image data acquired by the high-definition camera is transmitted to the onboard processor, the gesture images in the image data are recognized based on the YOLO recognition algorithm, the gestures are matched to the control commands of the mobile platform, and the control of the mobile platform by gestures is realized.

8. An unmanned testing system based on a ground mobile platform, characterized in that: include: The ground mobile platform building module is configured to: build a ground mobile platform and integrate the multi-modal sensor and the test instrument with the ground mobile platform using a consistent interface protocol; The map construction and path planning module is configured to: combine the data collected by the multimodal sensor, use the real-time map construction algorithm and the path planning navigation algorithm to respectively perform real-time construction of the scene map and path planning of the ground mobile platform; The human-computer interaction module is configured to: control the ground mobile platform to reach the test area using voice recognition according to the constructed scene map and path, Or, use gesture recognition to control the ground mobile platform to reach the test area; The test module is configured to control the test instrument on the mobile platform that arrives at the test area to complete the test operation.

9. A computer-readable storage medium having a program stored thereon, characterized in that: When the program is executed by a processor, the steps of an unmanned testing method based on a ground mobile platform as described in any one of claims 1 to 7 are implemented.

10. An electronic device comprising a memory, a processor, and a program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps of an unmanned testing method based on a ground mobile platform as described in any one of claims 1-7 are implemented.

Images