An unmanned aerial vehicle detection and recognition system and method based on multi-spectral lidar

Abstract

The present invention provides a drone detection and recognition system and method based on a multi-spectral lidar. The system includes a laser, a turntable control module, a transmitting optical system, a receiving optical system, a photon intensity collector, a spectrometer collector, and a host computer. The transmitting optical system is arranged on a rotating platform, and the rotating platform is used to drive the transmitting optical system to move in space. The laser emits laser light, which passes through the transmitting optical system to a micro-drone. The echo passes through the receiving optical system to the photon intensity collector and the spectrometer collector. The photon intensity collector and the spectrometer collector are connected to the host computer through a high-speed data acquisition card. The turntable control module is used to control the rotating platform. The present invention can obtain more intensity information and spectral information through the echo, making the understanding of the target characteristics refined. Thus, a method for detecting drone targets in a large amount of spectral data and point cloud data can be used for target detection in ground / sky anti-drone systems.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicle (UAV) detection, and particularly to a UAV detection and identification system and method based on multi-spectral lidar. Background Art

[0002] In recent years, some domestic units have carried out research on low-altitude defense systems, especially in UAV countermeasure systems. However, domestic defense equipment still has disadvantages such as a single detection means, low comprehensiveness of system functions, and weak networking and collaboration capabilities of equipment.

[0003] Currently, with the continuous development and expansion of the "low, slow, and small" flying object market, the phenomenon of "black flying" of "low, slow, and small" flying objects may have a serious adverse impact on public facilities. In recent years, people's concerns about the potential safety threats of "low, slow, and small" flying objects have been increasing. To address this issue, in order to protect social public personal safety, citizen personal privacy safety, property safety, etc., relevant research on anti-"low, slow, and small" flying object technologies is actively underway, and different anti-"low, slow, and small" flying object technology development strategies have been successively introduced. The new market for anti-"low, slow, and small" flying object technologies is also rapidly emerging.

[0004] Currently, the technical means for anti-"low, slow, and small" flying objects mainly include detection and tracking and early warning technologies, interference technologies, camouflage and deception technologies, and damage and capture technologies. By collecting the flight data and other information of "low, slow, and small" flying objects and combining interference technologies such as optoelectronic countermeasure technologies, control information interference technologies, and data link interference technologies, the "low, slow, and small" flying objects are made to reduce or even lose their flight capabilities. However, currently, the spatial identification and positioning means for intercepting "low, slow, and small" flying objects are relatively single, and are easily affected by various environmental uncertainties, resulting in poor accuracy and weak anti-interference ability. Therefore, it is very necessary to design a UAV detection and identification system and method based on multi-spectral lidar. Summary of the Invention

[0005] In order to overcome the deficiencies of the prior art, the purpose of the present invention is to provide a UAV detection and identification system and method based on multi-spectral lidar.

[0006] To achieve the above purpose, the present invention provides the following solutions:

[0007] The present invention provides a drone detection and identification system based on a multi-spectral lidar, including: a laser, a turntable control module, a rotating platform, a transmitting optical system, a receiving optical system, a photon intensity collector, a spectrometer collector, and a host computer. The transmitting optical system is arranged on the rotating platform, and the rotating platform is used to drive the transmitting optical system to move in space. The laser emits laser light, which passes through the transmitting optical system to a micro-drone, and the echo passes through the receiving optical system to the photon intensity collector and the spectrometer collector. The photon intensity collector and the spectrometer collector are connected to the host computer through a high-speed data acquisition card, and the turntable control module is used to control the rotating platform.

[0008] The present invention also provides a method for detecting and identifying drones based on a multi-spectral lidar, including:

[0009] Assemble and connect the drone detection and identification system;

[0010] Obtain drones on the market and collect their spectral data. The host computer performs SVM classification modeling based on the spectral data to obtain an SVM classification model;

[0011] Connect the laser and perform status verification and parameter initialization on it;

[0012] Connect the spectrometer collector and perform status verification and integration time setting on it;

[0013] Connect the high-speed data acquisition card and perform status verification on it, and then set the receiving channel and the sending channel;

[0014] Connect the turntable control module and perform status inspection and parameter initialization on it, and set the rotation rate and the turntable scanning range;

[0015] Start scanning to obtain the light intensity data, spectral data, and point cloud data of a certain point where there is a target in space;

[0016] Fuse the point cloud data of this point with the spectral data of this point;

[0017] Based on the SVM classification model, screen out the spectral data that conforms to the drone, and then screen out the point cloud data in space in sequence, and perform multi-source heterogeneous data fusion on them to obtain multi-source data;

[0018] Perform object recognition based on pointnet on the obtained multi-source data to obtain the position information and classification result of the target existing in space.

[0019] Preferably, the integration time is set to 200 ms.

[0020] Preferably, obtaining the point cloud data of a certain point is specifically:

[0021] Obtain the angular information (V_angle, H_angle) of the rotating platform at a certain point, where V_angle is the horizontal angle and H_angle is the pitch angle. According to the high-speed data acquisition card, the distance between the detection and recognition system and the target is DisValue. According to the distance formula (DisValue×cos(V_angle)×sin(H_angle), DisValue(i)×cos(V_angle)×cos(H_angle), DisValue(i)×sin(V_angle)), the point cloud data of the target in space, that is, (x, y, z), is obtained.

[0022] According to the specific embodiments provided by the present invention, the following technical effects are disclosed:

[0023] The present invention provides a drone detection and recognition system and method based on a multi-spectral lidar. The system includes a laser, a turntable control module, a rotating platform, a transmitting optical system, a receiving optical system, a photon intensity collector, a spectrometer collector, and a host computer. The method includes assembling and connecting the drone detection and recognition system, obtaining drones on the market, and collecting their spectral data. The host computer performs SVM classification modeling based on the spectral data to obtain an SVM classification model. Connect the laser and perform status verification and parameter initialization on it. Connect the spectrometer collector and perform status verification and integration time setting on it. Connect the high-speed data acquisition card and perform status verification on it. Then set the receiving channel and the transmitting channel. Connect the turntable control module and perform status inspection and parameter initialization on it. Set the rotation speed and the turntable scanning range, and start scanning to obtain the light intensity data, spectral data, and the point cloud data of a certain point in space where there is a target. Integrate the point cloud data of this point with the spectral data of this point, screen the spectral data that conforms to the drone based on the SVM classification model, and sequentially screen out the point cloud data in space, perform multi-source heterogeneous data fusion on it to obtain multi-source data, and perform object recognition based on pointnet on the obtained multi-source data to obtain the position information and classification result of the target existing in space. The present invention can obtain more intensity information and spectral information through the echo, making the understanding of the target characteristics refined. Therefore, the method for detecting drone targets in a large amount of spectral data and point cloud data can be used for target detection in ground / sky anti-drone systems. Description of the Drawings

[0024] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0025] Figure 1 This is the structural block diagram of the UAV detection and recognition system based on multi-spectral lidar provided by the embodiments of the present invention;

[0026] Figure 2 This is the flowchart of the UAV detection and recognition method based on multi-spectral lidar provided by the embodiments of the present invention;

[0027] Figure 3 This is the schematic diagram of the data processing flow;

[0028] Figure 4 This is the schematic diagram of an embodiment provided by the present invention;

[0029] Figure 5 This is the schematic diagram of the UAV three-dimensional point cloud result display. Specific embodiments

[0030] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without making creative efforts shall fall within the protection scope of the present invention.

[0031] The purpose of the present invention is to provide a UAV detection and recognition system and method based on multi-spectral lidar, which can obtain more intensity information and spectral information through echoes, make the understanding of the characteristics of the target refined, and thus can detect UAV targets in a large amount of spectral data and point cloud data. The method can be used for target detection in ground / sky anti-UAV systems.

[0032] To make the above objects, features, and advantages of the present invention more obvious and understandable, the present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0033] Figure 1 This is the structural block diagram of the UAV detection and recognition system based on multi-spectral lidar provided by the embodiments of the present invention, as Figure 1 shown, the present invention provides a UAV detection and recognition system based on multi-spectral lidar, including: a laser, a turntable control module, a rotating platform, a transmitting optical system, a receiving optical system, a photon intensity collector, a spectrometer collector, and a host computer;

[0034] The laser can use a white light laser.

[0035] The transmitting optical system is used to transmit the laser generated by the laser;

[0036] The emission optical system is arranged on a rotating platform, which is used to drive the emission optical system to move in space. The rotating platform is connected to the turntable control module, and the turntable control module is connected to the host computer. The host computer controls the movement of the rotating platform through the turntable control module. Among them, the rotation control module can adopt a motor control system. The emission optical system includes three reflecting mirrors and one optical rotating mirror.

[0037] The receiving optical system is used to receive the echo. The receiving optical system includes one optical rotating mirror, APD210, and one half mirror.

[0038] The photon intensity collector is used to collect the light intensity data of the echo.

[0039] The spectrometer collector is used to collect the spectral data of the echo.

[0040] The photon intensity collector and the spectrometer collector are connected to the host computer through a high-speed data acquisition card.

[0041] On the one hand, the host computer realizes the control of the turntable and the reading of the position information, obtains the spectral data and photon intensity data of the target's reflection of the light source at this position. On the other hand, it fuses the multi-source heterogeneous data, identifies and processes the point cloud data, and displays the results. On the third hand, it conducts the training of the UAV spectral data model and the identification of the point cloud data based on pointnet.

[0042] The laser emits laser light through the emission optical system to the micro UAV, and the echo passes through the receiving optical system to the photon intensity collector and the spectrometer collector.

[0043] Support Vector Machine (SVM) is a machine learning algorithm used for classification and regression analysis. It can effectively process linearly separable and linearly inseparable data and construct an optimal decision boundary in a high-dimensional space.

[0044] The core idea of SVM is to find a hyperplane that can separate the sample points of different classes as much as possible and maximize the interval between the two classes. This hyperplane is called the maximum margin hyperplane, which can be used for good classification prediction.

[0045] Specifically, SVM maps the samples to a high-dimensional feature space so that the data is linearly separable in this space. If a linear hyperplane cannot be found in the original input space to separate the data, SVM introduces a kernel function to transfer the computational complexity from the high-dimensional feature space to the original input space. Commonly used kernel functions include linear kernel, polynomial kernel, and Gaussian kernel, etc.

[0046] The training process of SVM is a convex optimization problem, and the goal is to minimize the structural risk of the model. During the solution process, SVM only focuses on those samples located near the decision boundary, which are called support vectors. This characteristic enables SVM to have good robustness and generalization ability.

[0047] Generally speaking, SVM is a powerful machine learning algorithm that can handle data in high-dimensional spaces and achieve classification and regression analysis by finding the optimal hyperplane. It has been widely applied in many fields such as image recognition, text classification, and bioinformatics.

[0048] As Figure 2 shown, the present invention also provides a method for detecting and identifying drones based on multi-spectral lidar, including:

[0049] Assemble and connect the drone detection and identification system;

[0050] Obtain drones on the market and collect their spectral data. The host computer conducts SVM classification modeling based on the spectral data to obtain an SVM classification model. Specifically:

[0051] Build an SVM classification model and train it based on the spectral data to obtain a trained SVM classification model;

[0052] Connect the laser and conduct status verification and parameter initialization for it;

[0053] Connect the spectrometer collector and conduct status verification and integration time setting for it;

[0054] Connect the high-speed data acquisition card, conduct status verification for it, and then set the receiving channel and the sending channel;

[0055] Connect the turntable control module, conduct status inspection and parameter initialization for it, and set the rotation rate and the turntable scanning range;

[0056] The data processing flow is as Figure 3 shown. Start scanning. The turntable will continuously scan the set range in a "zigzag" manner to obtain the light intensity data, spectral data, and point cloud data of a certain point where there is a target in space;

[0057] Fuse the point cloud data of this point with the spectral data of this point;

[0058] Based on the SVM classification model, screen the spectral data that conforms to the drone, and then screen out the point cloud data in space in sequence, and conduct multi-source heterogeneous data fusion on them to obtain multi-source data;

[0059] Conduct object recognition based on pointnet on the obtained multi-source data to obtain the position information and classification result of the target existing in space.

[0060] The integration time is set to 200 ms.

[0061] Obtain the point cloud data of a certain point, specifically:

[0062] Obtain the angle information (V_angle, H_angle) of the rotating platform at a certain point, where V_angle is the horizontal angle and H_angle is the pitch angle. According to the high-speed data acquisition card, the distance between the detection and recognition system and the target is DisValue. According to the distance formula (DisValue×cos(V_angle)×sin(H_angle), DisValue(i)×cos(V_angle)×cos(H_angle), DisValue(i)×sin(V_angle)), the point cloud data of the target in space, that is, (x, y, z), is obtained.

[0063] The present invention provides a schematic diagram of an embodiment, as Figure 4 shown. The xiaomo DS motor is adopted, the high-speed acquisition card is adq8, there are a total of 8 channels, and each channel has a sampling rate of 1 G. The selected target is a drone of the DJI Inspire series, and the spectrometer adopted is the ATP3330 series of OptoSky.

[0064] In the experiment of the present invention, the display diagram of the three-dimensional point cloud result of the drone is as Figure 5 shown.

[0065] The various embodiments in this specification are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. The same or similar parts among the various embodiments can be referred to each other.

[0066] Specific examples are applied in the present invention to elaborate on the principle and implementation manner of the present invention. The description of the above embodiments is only used to help understand the method and its core idea of the present invention; at the same time, for those of ordinary skill in the art, according to the idea of the present invention, there will be changes in the specific implementation manner and application scope. In summary, the content of this specification should not be construed as a limitation to the present invention.

Claims

1. A method for detecting and identifying unmanned aerial vehicles based on multispectral laser radar, characterized in that: The invention is applied to a UAV detection and identification system based on a multi-spectral laser radar, comprising: a laser, a turntable control module, a rotating platform, a transmitting optical system, a receiving optical system, a photon intensity collector, a spectrometer collector and a host computer, wherein the transmitting optical system is arranged on the rotating platform, and the rotating platform is used to drive the transmitting optical system to move in space, the laser emits laser light through the transmitting optical system to a micro UAV, and the echo is transmitted through the receiving optical system to the photon intensity collector and the spectrometer collector, the photon intensity collector and the spectrometer collector are connected to the host computer through a high-speed data acquisition card, and the turntable control module is used to control the rotating platform; The method includes: Assemble and connect the drone detection and identification system; Obtain drones on the market and collect spectral data from them. The host computer performs SVM classification modeling based on the spectral data to obtain an SVM classification model. Connect the laser and perform status check and parameter initialization; Connect the spectrometer collector and perform status check and integration time setting on it; Connect the high-speed data acquisition card, check its status, and then set the receiving channel and sending channel; Connect the turntable control module, perform status check and parameter initialization on it, and set the rotation rate and turntable scanning range; Start scanning to obtain the light intensity data, spectrum data and point cloud data of a certain point of the target in space; Fusion of the point cloud data of the point with the spectral data of the point; Based on the SVM classification model, the spectral data that meets the requirements of the UAV is screened, and the point cloud data in the space is screened in turn, and multi-source heterogeneous data is fused to obtain multi-source data; Based on the obtained multi-source data, pointnet-based object recognition is performed to obtain the location information and classification results of the targets in space.

2. The method according to claim 1, characterized in that: The integration time is set to 200 ms.

3. The method according to claim 2, characterized in that Get the point cloud data of a certain point, specifically: Get the angle information of the rotating platform at a certain point (V_angle, H_angle), where V_angle is the horizontal angle and H_angle is the pitch angle. According to the high-speed data acquisition card, the distance between the detection and recognition system and the target is DisValue. According to the distance formula (DisValue*cos(V_angle)*sin(H_angle), DisValue(i)*cos(V_angle)*cos(H_angle), DisValue(i)*sin(V_angle)), get the point cloud data of the target in space, that is, (x, y, z).

Images