A method for locating multiple moving targets based on bearing information

Abstract

A positioning method for multiple moving targets based on azimuth information belongs to the field of unmanned ship cooperative positioning. In order to solve the defects of the prior art, such as dependence on GNSS, unstable target state estimation, and lack of distributed multi-target cooperative sensing capability, the technical scheme provided by the present application comprises: outputting the global position, heading angle and speed information of each unmanned ship; determining the neighbor state information of each unmanned ship; collecting the target azimuth information obtained by each unmanned ship, and combining the self state information and the neighbor state information to generate target observation input; estimating the position state and speed state of the target in the global coordinate system based on the target observation input; adjusting the observer parameters according to the difference between the estimated state and the actual observation to make the error converge, and outputting the stable target state estimation result. It is suitable for real-time positioning and tracking of multiple moving targets in the cooperative operation of multiple unmanned ships in GNSS denial or complex environment.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of cooperative positioning of unmanned surface vehicles, and particularly relates to positioning of multiple moving targets based on bearing information. BACKGROUND

[0002] With the rapid development of unmanned surface vehicles (USVs), cluster cooperative operation has become an important means to achieve complex task objectives (such as target tracking, maritime search and rescue, intelligent patrol and monitoring, etc.). In these applications, real-time and high-precision positioning of targets is a fundamental prerequisite. Most traditional USV target positioning methods rely on global navigation satellite systems (GNSS), such as using GNSS and inertial navigation systems (INS) to achieve target and self-positioning. However, in near-shore, island areas or complex electromagnetic environments, GNSS signals often face obstructions, interference or denial, severely affecting the stability and accuracy of the positioning system.

[0003] To cope with GNSS denial environments, some studies have proposed methods for target positioning using relative measurement information, such as target tracking methods based on LiDAR or ultrasonic ranging. These methods can achieve certain results in specific scenarios, but they usually have high requirements for target reflection characteristics, communication environment or ranging accuracy, and are difficult to adapt to dynamic multiple targets or complex water surface environments.

[0004] In recent years, visual sensors have been widely used in the perception and positioning tasks of unmanned systems due to their low cost and rich information. Some scholars have conducted target detection and bearing estimation based on monocular or binocular vision systems, attempting to replace distance measurement with bearing angle to achieve target positioning. For example, some studies use visual SLAM technology to position in known maps, but for open water environments with no significant features, the SLAM method is limited in effectiveness. Other studies have proposed target bearing estimation methods based on single-boat visual observation, but when faced with multiple moving targets or targets frequently moving out of view, they cannot guarantee stable and continuous observation.

[0005] In addition, current positioning methods based on visual information mostly focus on single-boat observation and processing, without fully considering the possibility of multi-boat cooperation. Although some studies have explored cluster cooperative perception mechanisms, they mostly rely on global control centers or high-bandwidth communication conditions, lacking distributed solutions suitable for bandwidth-limited, angle-limited and center-node-unavailable scenarios. Especially in the case of dynamic targets frequently entering and leaving the field of view of unmanned surface vehicles and the lack of unified awareness of targets among unmanned surface vehicles, the performance of target positioning is significantly reduced.

[0006] In summary, the existing technology has the defects of relying on GNSS, limited visual coverage, unstable target state estimation, and lack of distributed multi-target cooperative perception capability. SUMMARY

[0007] To solve the defects of the prior art, such as dependence on GNSS, limited visual angle coverage, unstable target state estimation, and lack of distributed multi-target cooperative perception capability, the technical scheme provided by the present application is as follows:

[0008] A positioning method for multiple moving targets based on bearing information, comprising:

[0009] establishing a kinematic model describing the motion state of the unmanned ship in a two-dimensional plane, outputting the global position, heading angle and velocity information of each unmanned ship;

[0010] constructing a communication topology structure of the multi-unmanned ship cluster, determining the neighbor set of each unmanned ship based on a directed graph, and generating neighbor state information;

[0011] collecting the target bearing information obtained by each unmanned ship through a visual sensor, and combining the self-state information and neighbor state information of the unmanned ship to generate target observation input;

[0012] constructing a target state observer based on the target observation input, and estimating the position state and velocity state of the target in the global coordinate system;

[0013] establishing an observation error system according to the difference between the estimated state and the actual observation, adjusting the observer parameters to make the error converge, and outputting a stable target state estimation result.

[0014] Further, in a preferred embodiment, the kinematic model uses an earth-fixed coordinate system and a ship-fixed coordinate system for modeling, and the global position and velocity information of each unmanned ship are obtained through coordinate transformation.

[0015] Further, in a preferred embodiment, the communication topology structure is a directed graph structure, each node corresponds to an unmanned ship, and each directed edge represents an information transmissible neighbor relationship.

[0016] Further, in a preferred embodiment, each unmanned ship only observes the bearing information of one moving target at any time, and maps the bearing information to the earth coordinate system to form the observation input.

[0017] Further, in a preferred embodiment, the target state observer includes a position observer and a velocity observer, which are used to estimate the position state and velocity state of the target, respectively.

[0018] Further, in a preferred embodiment, the observation error system constructs a state error matrix by introducing control parameters.

[0019] A positioning device for multiple moving targets based on bearing information is also provided, comprising:

[0020] A module for establishing a kinematic model describing the motion state of the unmanned surface vehicle in a two-dimensional plane, outputting the global position, heading angle and velocity information of each unmanned surface vehicle;

[0021] A module for constructing a communication topology of the multi-unmanned surface vehicle cluster, determining the neighbor set of each unmanned surface vehicle based on a directed graph, and generating neighbor state information;

[0022] A module for collecting target bearing information obtained by each unmanned surface vehicle through a visual sensor, and combining the target bearing information with the self-state information and neighbor state information of the unmanned surface vehicle to generate target observation input;

[0023] A module for constructing a target state observer based on the target observation input, and estimating the position state and velocity state of the target in a global coordinate system;

[0024] A module for establishing an observation error system according to the difference between the estimated state and the actual observation, adjusting the observer parameters to make the error converge, and outputting a stable target state estimation result.

[0025] The computer storage medium for storing the computer program is also provided, and the computer executes the method when the computer program is read by the computer.

[0026] The computer including a processor and a storage medium is also provided, and the computer executes the method when the processor reads the computer program stored in the storage medium.

[0027] The computer program product is also provided as a computer program, and the method is implemented when the computer program is executed.

[0028] Compared with the prior art, the technical solution provided by the present application has the following advantages:

[0029] By constructing the kinematic model of the unmanned surface vehicle and introducing the earth-fixed coordinate system and the ship-fixed coordinate system, the two-dimensional motion state of the unmanned surface vehicle is accurately described at the mathematical modeling level, providing a rigorous theoretical basis for subsequent state estimation and error analysis. Unlike some existing researches that use simplified motion models, the present application establishes a more complete coordinate conversion and velocity expression system, which is beneficial to improving the accuracy of target state estimation and the reliability of system modeling.

[0030] By designing the target state observer based on bearing information, the position and velocity states of the target are estimated in real time by relying only on the bearing information obtained by the visual sensor and the interaction information of the neighboring unmanned surface vehicles (USVs) in the environment where GNSS signals are missing or disturbed. Compared with the traditional positioning methods that rely on GNSS positioning or distance measurement, this method significantly enhances the positioning capability of the system in a complex electromagnetic interference environment, and is particularly suitable for use in GNSS-unavailable areas such as near-shore or complex water surfaces.

[0031] By constructing the communication topology of the multi-USV cluster and modeling the information interaction relationship by using a directed graph, the system can cooperatively complete the target positioning task under a distributed architecture. Unlike some existing researches that rely on centralized control nodes or global communication links, this scheme uses local information fusion and distributed observation mechanism, effectively improving the scalability and robustness of the system, and adapting to the actual application requirements of limited communication bandwidth or unstable network.

[0032] To overcome the problem of limited visual angle of a single visual sensor, a strategy of target visual angle switching and observation data cooperative processing is proposed. By limiting each USV to observe only one target and designing the observation angle relationship between the USVs, the continuity of target state observation and the continuous tracking capability of the system for the target are enhanced. This method is more stable than the traditional method of observing multiple targets simultaneously but processing mixed information, and significantly reduces the positioning interruption problem caused by the frequent entering or exiting of the target from the visual field.

[0033] By introducing a state observation error system and proving the stability of the observer, it is ensured that under the condition of reasonable design parameters, the position and velocity estimation errors of the target can globally asymptotically converge to zero. This theoretical guarantee mechanism is different from most existing methods that only stay in the simulation verification stage of design, and provides a controllable and verifiable mathematical basis for the performance evaluation and parameter tuning in engineering implementation.

[0034] It is suitable for real-time positioning and tracking of multiple moving targets by multiple USVs in GNSS denial or complex environments. BRIEF DESCRIPTION OF DRAWINGS

[0035] Figure 1 It is a schematic diagram of a positioning method for multiple moving targets based on bearing information;

[0036] Figure 2 It is a curve graph of the observation error of USV 1 for multiple moving targets;

[0037] Figure 3 It is a curve graph of the observation error of USV 2 for multiple moving targets;

[0038] Figure 4 It is a curve graph of the observation error of USV 3 for multiple moving targets;

[0039] Figure 5 A plot of the observation error variation of the unmanned vehicle 4 for the multiple moving targets. DETAILED DESCRIPTION

[0040] In order to make the advantages and beneficial effects of the technical solutions provided by the present application more clear, the technical solutions provided by the present application are further described in detail below in combination with the drawings, and the specific embodiments are as follows:

[0041] Embodiment one, the embodiment provides a positioning method for multiple moving targets based on bearing information, comprising:

[0042] A step of establishing a kinematic model describing the motion state of the unmanned vehicle in a two-dimensional plane, outputting the global position, heading angle and velocity information of each unmanned vehicle;

[0043] A step of constructing a communication topology structure of the multi-unmanned vehicle cluster, determining the neighbor set of each unmanned vehicle based on a directed graph, and generating neighbor state information;

[0044] A step of collecting the target bearing information obtained by each unmanned vehicle through a visual sensor, and combining the self-state information and neighbor state information of the unmanned vehicle to generate a target observation input;

[0045] A step of constructing a target state observer based on the target observation input, estimating the position state and velocity state of the target in the global coordinate system;

[0046] A step of establishing an observation error system according to the difference between the estimated state and the actual observation, adjusting the observer parameters to make the error converge, and outputting a stable target state estimation result.

[0047] The kinematic model adopts an earth-fixed coordinate system and a ship-fixed coordinate system for modeling, and the global position and velocity information of each unmanned vehicle are obtained through coordinate transformation.

[0048] The communication topology structure is a directed graph structure, each node corresponds to an unmanned vehicle, and each directed edge represents an information transmissible neighbor relationship.

[0049] Each unmanned vehicle only observes the bearing information of one moving target at any time, and maps the bearing information to the earth coordinate system to form an observation input.

[0050] The target state observer includes a position observer and a velocity observer, which are used to estimate the position state and velocity state of the target, respectively.

[0051] The observation error system constructs a state error matrix by introducing a control parameter.

[0052] Embodiment two, the embodiment is further detailed description of the technical solutions provided by embodiment one, specifically:

[0053] A multi-motion target positioning method suitable for GNSS denial environment, mainly aiming at the problem of cooperative positioning of multiple dynamic targets in the task area by a cluster system composed of multiple unmanned ships. The method fully utilizes the target azimuth information obtained by the visual sensor, and combines the inter-ship communication topology and distributed observer design to realize real-time estimation of the target position and velocity state. Specifically, the following steps are included:

[0054] Firstly, a two-dimensional kinematic model of the unmanned ship is established, which is based on the earth-fixed coordinate system and the ship-fixed coordinate system, respectively describing the spatial position, heading angle, longitudinal velocity, lateral velocity and bow velocity of each unmanned ship. Through coordinate transformation method, the local state of each ship is mapped to the global reference coordinate, which lays the mathematical foundation for subsequent state cooperative observation and error fusion.

[0055] Secondly, the communication topology structure of the cluster system is constructed, and the neighbor relationship and information exchange path between each unmanned ship are described using a directed graph. In this topology, each node represents an unmanned ship, and the edge represents that the ship can share state information with the adjacent unmanned ship. Each unmanned ship only needs to exchange state information with its neighbor nodes, without global communication, thereby reducing the system bandwidth requirement and enhancing the system robustness.

[0056] Then, the visual sensor azimuth information is collected and the observation input is established. Since the visual sensor has a fixed viewing angle range, the system design assumes that each unmanned ship can only observe one moving target at any time. By identifying the angle between the target and the bow direction of the unmanned ship in the image, the target azimuth angle is obtained; combined with the heading angle and position information of the unmanned ship, the target azimuth data is unified to the global coordinate system. In addition, the system limits each target to be observed by at most one unmanned ship, ensuring that the target observation data does not conflict.

[0057] On this basis, a target state observer based on azimuth information is constructed, which includes two sub-modules: target position observer and target velocity observer. The input is: the azimuth angle information of the observed target, the position and attitude state of the unmanned ship itself, and the state information received from the neighbor ship. The observer estimates the position and velocity state of the target in the global coordinate system by fusing these information in the ship coordinate system.

[0058] Further, the target observation error system is constructed. By defining the error term between the estimated position of the target and the real position, and combining the bearing vector relationship between the USV and the target, the error state dynamic system is established. In this system, by adjusting the gain parameter of the observer, the state matrix of the error system is made to be a Hurwitz matrix, so as to ensure the error convergence. The theoretical derivation shows that as long as the parameter selection condition is met, the observation error will globally asymptotically converge to zero, ensuring the accuracy and stability of the target state estimation.

[0059] Finally, the effectiveness of the method is verified by simulation experiments. The simulation scene includes multiple dynamic targets randomly moving in a two-dimensional task area, and multiple USVs observing the bearing of the visible targets and forming a cooperative network through neighbor communication. In multiple rounds of simulation, the position and velocity estimation results of the USVs on the targets are highly consistent with the real trajectory, and the observation error gradually converges over time, fully verifying the adaptability and robustness of the method in actual complex environments.

[0060] Embodiment three, in combination Figures 1-5 To illustrate the embodiment, the technical solutions provided above are described in further detail through specific embodiments, and specifically:

[0061] (1) Establish the kinematic mathematical model of the USV.

[0062] Consider a network system composed of multiple USVs . Then, two related coordinate systems are introduced to describe the arbitrary motion of the USV: the earth-fixed coordinate system and the ship-fixed coordinate system. Both coordinate systems follow the Cartesian right-hand rule. The kinematic mathematical model of the USV is represented as:

[0063]

[0064] In the formula, represents the position of the i-th USV, represents the heading of the i-th USV. , and represent the longitudinal, lateral, and heading velocities of the USV, respectively. Define the position of the i-th USV as and the velocity as

[0065] . In addition, it can be obtained that:

[0066]

[0067] In the formula ​​​The rotation matrix from the ship body-fixed coordinate system to the earth-fixed coordinate system is expressed as

[0068]

[0069] The interaction topology between the unmanned ship cluster system is represented by a directed graph as , where represents the node set, represents the edge set. Define as the neighbor set of the th unmanned ship. For the graph , represents the adjacency matrix, where , .

[0070] For a multi-target system, define as the target node set, represents a matrix, where if the th unmanned ship can obtain information of the th target, then , otherwise . For the position state of the th target in the earth-fixed coordinate system is , and the velocity state is . The vector of the th unmanned ship pointing to the th target is . The unit vector of the th unmanned ship pointing to the th target is . The velocity of the th target in the ship body-fixed coordinate system of the th unmanned ship is .

[0071] (2) Construct a positioning method for multiple moving targets.

[0072] Define as the initial heading of the th unmanned ship, as the heading of the th unmanned ship at time . The heading error of the th unmanned ship at time and the initial heading is . The rotation matrix at the current time is defined as:

[0073]

[0074] Define Considering the limited field of view of each unmanned surface vessel's (USV) visual sensor, a single USV cannot simultaneously observe all moving targets within the mission area. To demonstrate the superiority of the proposed target localization method, it is assumed that each USV can observe the orientation information of at most one target. Each target can only be observed by at most one USV.

[0075] Definition of the first The unmanned surface vessel is the first The neighbor of two unmanned surface vessels (USVs), the angle difference between the two USVs is . , No. The fixed coordinate system of the unmanned surface vessel and the first The rotation matrix of the unmanned surface vessel in the fixed coordinate system is represented as follows: Based on the first The first goal in Position of the unmanned surface vessel in a fixed coordinate system , can obtain the first The first goal in The position of the unmanned surface vessel in the fixed coordinate system is ,in .

[0076] Assume the first The first goal in The position of the unmanned surface vessel in the fixed coordinate system is unknown, defined and They represent the first The first goal in Estimates of the position and velocity of an unmanned surface vessel in a fixed coordinate system.

[0077] Therefore, the target observer based on azimuth information can be obtained as follows:

[0078]

[0079] In the formula, and This indicates the positive design parameters.

[0080] The technical effect is as follows:

[0081] By using visual sensors to obtain the target azimuth angle and combining it with the constructed target observer and cluster collaborative information, it is possible to effectively observe and estimate the position and velocity status of moving targets in the ship coordinate system in complex mission environments where GNSS signals are lost or denied, significantly improving the autonomous operation capability of unmanned surface vessel clusters without satellite navigation support.

[0082] Multiple moving targets in the field of view can be simultaneously localized and tracked. The design of the target observer fully considers the inherent physical limitations of the visual sensor, especially the angle of view limitation. Through algorithm design, the robustness and practicality of the localization process are ensured, and the problem of localization failure caused by the limited field of view of the sensor is avoided.

[0083] By constructing the communication topology relationship of the multi-unmanned ship cluster system, the information interaction and collaborative observation within the cluster are realized. Each unmanned ship not only uses its own visual sensor to obtain target azimuth information, but also fuses the azimuth or state information transmitted by the neighbor unmanned ship. This distributed collaborative mechanism effectively overcomes the bottleneck of limited single-ship viewing angle, expands the spatial coverage of target positioning, and significantly improves the accuracy and reliability of target position and velocity state estimation. The whole scheme completely eliminates the dependence on GNSS signal, and provides a feasible scheme for target positioning in complex confrontation or restricted environment.

[0084] Each unmanned ship only relies on its own visual sensor and limited range neighbor communication, and can realize local estimation of target state by using the constructed target observer based on azimuth information. This architecture reduces the dependence on the central node or global communication network, improves the scalability and survivability of the system. At the same time, since it mainly relies on azimuth information and inter-ship cooperation, the demand for high-precision external positioning infrastructure is reduced, and the complexity and cost of system deployment and operation are significantly reduced.

[0085] The above further describes the technical solutions provided by the present application through several specific embodiments, in order to highlight the advantages and benefits of the technical solutions provided by the present application. However, the above several specific embodiments are not used as a limitation on the present application, and any reasonable modification and improvement of the present application, combination and equivalent replacement of the embodiments, etc. based on the spirit and principles of the present application should be included in the protection scope of the present application.

Claims

1. A method for locating multiple moving targets based on orientation information, characterized in that, include: The steps to establish a kinematic model describing the motion state of unmanned surface vessels in a two-dimensional plane and output the global position, heading angle and velocity information of each unmanned surface vessel; The steps for constructing the communication topology of a multi-unmanned surface vessel (USV) cluster, determining the neighbor set of each USV based on a directed graph, and generating neighbor state information are as follows: The steps involve collecting target orientation information obtained by each unmanned surface vessel (USV) through visual sensors, and combining this information with the USV's own state information and neighboring state information to generate target observation input. The steps to construct a target state observer based on the target observation input and estimate the target's position and velocity state in the global coordinate system; The steps are as follows: establish an observation error system based on the difference between the estimated state and the actual observation, adjust the observer parameters to make the error converge, and output a stable target state estimation result. The kinematic model uses a fixed Earth coordinate system and a fixed ship coordinate system for modeling, and obtains the global position and velocity information of each unmanned surface vessel through coordinate transformation.

2. The method for locating multiple moving targets based on orientation information according to claim 1, characterized in that, The communication topology is a directed graph structure, where each node corresponds to an unmanned surface vessel, and each directed edge represents a neighbor relationship where information can be transmitted.

3. The method for locating multiple moving targets based on orientation information according to claim 1, characterized in that, Each unmanned surface vessel observes the orientation information of only one moving target at any given time and maps this orientation information to a fixed Earth coordinate system to form the observation input.

4. The method for locating multiple moving targets based on orientation information according to claim 1, characterized in that, The target state observer consists of two parts: a position observer and a velocity observer, which are used to estimate the target's position and velocity states, respectively.

5. The method for locating multiple moving targets based on orientation information according to claim 1, characterized in that, The observation error system constructs a state error matrix by introducing control parameters.

6. A positioning device for multiple moving targets based on orientation information, characterized in that, include: A module is established to describe the motion state of unmanned surface vessels in a two-dimensional plane and output the global position, heading angle and velocity information of each unmanned surface vessel. A module for constructing the communication topology of a multi-unmanned surface vessel (USV) cluster, determining the neighbor set of each USV based on a directed graph, and generating neighbor state information. A module that collects target orientation information obtained by each unmanned surface vessel through visual sensors and combines it with the unmanned surface vessel's own state information and neighboring state information to generate target observation input; This module constructs a target state observer based on the target observation input, and estimates the target's position and velocity state in the global coordinate system. This module establishes an observation error system based on the difference between the estimated state and the actual observation, adjusts the observer parameters to make the error converge, and outputs a stable target state estimation result. The kinematic model uses a fixed Earth coordinate system and a fixed ship coordinate system for modeling, and obtains the global position and velocity information of each unmanned surface vessel through coordinate transformation.

7. A computer storage medium for storing computer programs, characterized in that, When the computer program is read by the computer, the computer executes the method of claim 1.

8. A computer, comprising a processor and a storage medium, characterized in that, When the processor reads the computer program stored in the storage medium, the computer executes the method of claim 1.

9. A computer program product, as a computer program, is characterized by: When the computer program is executed, it implements the method of claim 1.

Images