WO2022057107A1 - 面向观测优化的多机异类传感器协同多目标跟踪方法 - Google Patents

面向观测优化的多机异类传感器协同多目标跟踪方法 Download PDF

Info

Publication number
WO2022057107A1
WO2022057107A1 PCT/CN2020/135190 CN2020135190W WO2022057107A1 WO 2022057107 A1 WO2022057107 A1 WO 2022057107A1 CN 2020135190 W CN2020135190 W CN 2020135190W WO 2022057107 A1 WO2022057107 A1 WO 2022057107A1
Authority
WO
WIPO (PCT)
Prior art keywords
target
uav
observation
formation
sensor
Prior art date
Application number
PCT/CN2020/135190
Other languages
English (en)
French (fr)
Inventor
孙顺
刘瑜
但波
郭晨
丁自然
谭大宁
姜乔文
Original Assignee
中国人民解放军海军航空大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 中国人民解放军海军航空大学 filed Critical 中国人民解放军海军航空大学
Publication of WO2022057107A1 publication Critical patent/WO2022057107A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G05CONTROLLING; REGULATING
    • G05DSYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
    • G05D1/00Control of position, course, altitude or attitude of land, water, air or space vehicles, e.g. using automatic pilots
    • G05D1/10Simultaneous control of position or course in three dimensions
    • G05D1/101Simultaneous control of position or course in three dimensions specially adapted for aircraft
    • G05D1/104Simultaneous control of position or course in three dimensions specially adapted for aircraft involving a plurality of aircrafts, e.g. formation flying

Definitions

  • the invention relates to the technical field of cooperative target tracking, in particular to a multi-machine heterogeneous sensor cooperative multi-target tracking method oriented to observation optimization.
  • the invention aims at optimizing the observation of multi-targets by heterogeneous sensors, studies the multi-machine multi-target cooperative tracking pattern, and combines the characteristics of the working mode of heterogeneous sensors and the observability conditions of multi-UAVs to optimize the multi-UAV coordination.
  • the tracking system conducts in-depth research on the global observation performance in the process of multi-target tracking, the observation configuration problem of multiple UAVs and multi-targets and the tight surround tracking problem, and proposes an effective and reliable new method.
  • the purpose of the present invention is to provide a multi-machine heterogeneous sensor cooperative multi-target tracking method oriented to observation optimization, which improves the global observation performance.
  • Mann filter (Bias-Compensated Pseudolinear Kalman Filter, BC-PLKF) is used to continuously estimate the target state.
  • the EKF and BC-PLKF methods are used for the target state tracking filtering of active/passive sensors respectively, which can unify the target state vector to be estimated.
  • the sequential The method obtains the local target state fusion estimation.
  • each UAV After each UAV obtains the state estimation result of the target, it communicates with the neighboring UAV through the communication link.
  • the communication content includes the observation result of the target, the UAV motion state information and the future control sequence information, and then can use
  • the distributed target state fusion estimation method based on information consistency obtains consistent target state estimation.
  • interactive multi-models can be combined to improve the accuracy of target state estimation.
  • target state fusion it is assumed that the target observations of each UAV have completed time registration, track association and system bias estimation.
  • the set of target nodes in the target formation p is expressed as The set of UAVs for positioning and tracking formation N p is In order to ensure the stability and continuity of multi-UAV tracking multiple targets, it is assumed that a target formation is observed by one and only one UAV formation, and a single UAV formation can observe multiple target formations. In order to ensure the stability and continuity of multi-UAV tracking multiple targets, it is assumed that a target formation is observed by one and only one UAV formation, and a single UAV formation can observe multiple target groups.
  • Figure 2 shows a multi-UAV-to-multi-target formation cooperative standoff tracking scenario.
  • M ij,k is the Fisher information matrix of the drone i to the target j
  • p represents the number of the target formation
  • q represents the number of the drone formation
  • N p represents the p-th target formation
  • M q represents the q-th unmanned aerial vehicle. aircraft formation.
  • H xij,k is the Jacobian matrix of the sensor measurement equation to the target state
  • R i,k is the sensor measurement error covariance matrix
  • a stable standoff distance r d is set, so a circle with the target as the center and the standoff distance as the radius is set as the rejection area for the target, so as to ensure that the UAV flies along the edge of the rejection area.
  • R min represents the minimum turning radius of the drone
  • v max represents the maximum speed of the drone
  • ⁇ max represents the maximum steering angular velocity of the drone.
  • the minimum distance between the formations is not less than 2r d , as shown in Figure 4, where the width d min of the added rectangle is determined by the width of the rectangle when the two circles are tangent.
  • the rejection area of the formation target is divided into and two parts, as shown in Figure 5.
  • the target j closest to the UAV i is selected, and its performance function is defined as:
  • the multi-target observation performance function of UAV system is:
  • N r is the length of the rolling time domain.
  • the optimized MPC method can be used to solve the UAV control variables in a decentralized framework, and to coordinately control the circular motion directions of multiple UAVs.
  • the distance between the targets in the formation is relatively close, and the distance between the formations is relatively far, so it is necessary to divide the multiple targets into multiple target formations (groups).
  • the working mode of the sensor based on the rotational scanning system and the influence of the sensor observation type on the target grouping and target mission planning are considered.
  • the airborne sensors can work in different modes according to their own characteristics and the requirements of the mission objectives.
  • the two common working modes are target search and target tracking.
  • the target tracking mode can be divided into two categories according to the priority of the target. Target capture, target tracking, memory tracking and target verification, etc.
  • phased array radar As an example, through multiple coherent accumulation, the signal-to-noise ratio of ranging can be improved, and the positioning and tracking accuracy of the target can be improved.
  • Using the camera sensor mounted on the gimbal can increase the sampling rate of the target observation by keeping the target in the field of view at all times, thereby improving the target tracking performance.
  • a tracking-search mode is defined.
  • the observation performance of the target can be improved by increasing the dwell time of the sensor field of view, increasing the sampling rate of the observation or improving the signal-to-noise ratio of the sensor; for continuous non-targets
  • the area or the secondary target area is based on the normal sampling rate, so as to ensure the search and detection of possible new targets, and at the same time to ensure the status update of the secondary target.
  • the observation period of the sensor changes from T s to T p,k ;
  • n TS represents the multiple of the dwell time of the sensor field of view relative to the original time.
  • the FIM at time t can be defined as:
  • Y t0 is the FIM at time t 0
  • H ⁇ is the linearization matrix of the observation equation
  • R is the sensor measurement error covariance matrix
  • the new FIM is equivalent to:
  • R TS represents the improved equivalent sensor measurement error matrix under the tracking working model.
  • the FIM is equivalent to:
  • g p,k is defined as the equivalent observation gain of the UAV to the target in the target formation N p at time k when the sensor is in the "track-search" working mode.
  • UAV i it is defined as:
  • g TS represents the improved sensor performance due to the sensor using the field-of-sight working mode
  • g TS 1, g ip,k ⁇ 1, that is, the sensor will reduce the observation ability of the secondary target.
  • This method includes two steps. The second is to configure the UAV to different targets according to the observation requirements and the current movement state of the UAV.
  • the UAV According to the ability to observe the target at the current position of the UAV, the UAV needs to be configured to different targets. For this reason, the optimal assignment method is usually introduced to obtain the optimal UAV-target formation observation configuration result for the total observation amount of all targets, but the influence of different working models of heterogeneous sensors on the overall observation performance is not considered, as The present invention proposes a method for solving the optimal observation configuration relationship between the UAV and the target, which is suitable for the cooperative multi-target tracking of heterogeneous sensors.
  • the newly added FIM at the current moment is used to define the observation amount of a single UAV on different target formations, namely
  • the number of UAVs in the UAV formation should not be less than 2 If the UAV is only equipped with active sensors or both active and passive sensors, a single UAV can realize the observation, positioning and tracking of the target. Assuming that the number of UAVs carrying only passive sensors is n P , and the number of other UAVs is n A , then the maximum number of UAV formations that meet the observable conditions is where [ ⁇ ] D represents the round-down operation.
  • the number of target formations is not more than the maximum number of UAV formations
  • the UAV formation can increase the number of UAVs in the formation and improve the observation performance of the target on the basis of satisfying the observability of the target. Therefore, for UAVs only equipped with passive sensors, the number of UAVs required to observe the same target formation is not less than two, namely
  • the number of UAVs observing the same target formation is not less than one, that is,
  • UAVs refer to UAVs equipped with active sensors.
  • each UAV formation only observes one target formation, namely
  • a UAV formation needs to observe at least one target formation in tracking mode, namely
  • Equation (23) the optimal UAV formation configuration results and the corresponding UAV-target configuration results can be obtained by using the 0-1 Integer Linear Programming method. Observe the configuration results.
  • the composition of the target formation may also vary.
  • the UAV needs to re-determine the composition of the UAV formation to ensure the optimal overall observation performance of the multi-UAV system for multiple targets.
  • the UAV When a single target formation is split into multiple formations, on the basis of ensuring the observability of the target, the UAV also needs to be split into multiple UAV formations, or when the number of target formations is large, other unmanned aerial vehicles may even be required.
  • the man-machine formation cooperates to observe the split target formation.
  • the UAV formation can be merged to improve the observation performance of the new target formation.
  • the UAV and the target are constantly moving, when the UAV is working in the tracking-search mode, the angular range of the target area in the UAV's field of view changes with time according to its relative geometric position relationship.
  • a drone formation may need to focus on observing multiple target formations.
  • the unmanned The UAV has poor observation ability, and other UAVs outside the formation may have better observation performance for the target. Therefore, when the target number closest to a UAV changes, the UAV will be re-optimized.
  • the observation configuration relationship for the target formation is particularly in the case of a relatively large number of target formations.
  • An observation-oriented optimization-oriented multi-machine heterogeneous sensor cooperative multi-target tracking method of the present invention includes:
  • the motion state x i,k of each UAV i at time k, the sensor measurement covariance matrix R i,k and the control sequence at the previous time are obtained Among them, j ⁇ N, i ⁇ M, N represents the set of all targets, and M represents the set of all UAVs;
  • the sensor measurement covariance matrix R i,k According to the motion state xi,k of each UAV i at time k , the sensor measurement covariance matrix R i,k and the control sequence at the previous time Obtain the estimated results x ij,k of the position states of all targets through the distributed heterogeneous sensor fusion method;
  • the estimation result x ij,k of the position state of the target determine whether the target closest to the UAV has changed or whether the formation composition of the target has split, merged or handed over;
  • the present invention discloses the following technical effects:
  • the proposed method can timely re-plan the observation configuration relationship according to the geometric position relationship between the UAV and the target and the dynamic change of the target formation composition under the "tracking-search" working model of heterogeneous sensors, thereby significantly improving the global Observe performance.
  • Fig. 1 is the implementation flow chart of the multi-machine heterogeneous sensor cooperative multi-target tracking method for observation optimization according to the present invention
  • FIG. 2 is a schematic diagram of multi-machine multi-target cooperative tracking according to the present invention.
  • Fig. 3 is the structural schematic diagram of the rejection area of the single target formation of the present invention.
  • FIG. 4 is a schematic diagram of the rejection area of multiple target formations of the present invention.
  • FIG. 5 is a schematic diagram of the division of the rejection area according to the present invention.
  • FIG. 2 shows a schematic diagram of a scene where multiple UAVs are used to coordinately track multiple targets.
  • each UAV can communicate with each other the UAV's own motion state, sensor performance, observation of the target, and the optimization results of the control sequence at the previous moment, and use distributed target state fusion. Estimation method to get the state estimation result about the target. Then, under the decentralized framework, each UAV determines whether it is necessary to re-plan the observation configuration relationship between the UAV and the target formation according to the above information, and finally, according to the observation configuration relationship, the envelope of the given rejection area is determined. The target formation is coordinated to track the target formation, so that the global observation results are optimal.
  • the communication between UAVs only occurs in the process of target state fusion estimation. In the subsequent process of observation configuration optimization and collaborative target tracking, the use of a decentralized framework can effectively reduce the demand for real-time communication capabilities and facilitate engineering practice.
  • Step 1 By communicating with neighboring UAVs, UAV i can obtain the motion state x i,k of all UAVs, the sensor measurement covariance matrix R i,k and the control sequence at the previous moment And through the distributed heterogeneous sensor fusion method, the estimated results of all target states are obtained as x ij,k , where j ⁇ N, i ⁇ M, N represents the set of all target nodes, and M represents the set of all UAV nodes;
  • Step 2 When the formation composition of the target changes, such as splitting, merging, handover, etc., or when the target closest to the UAV changes, re-plan the observation configuration relationship between the UAV and the target according to step 3, otherwise Skip step 3, and implement multi-machine and multi-target collaborative tracking according to step 4;
  • Step 3 According to the sensor configuration of the multi-UAV observation system, consider the influence of different working modes of different sensors on the observation performance, solve the optimal observation configuration relationship between the UAV and the target, and optimize the global observation performance;
  • Step 3.1 Target grouping, if the distance between any two adjacent targets in a target set is less than 2r d , and r d is the standoff distance, the target set is a target formation, and hierarchical clustering based on Euclidean distance is used. method to cluster all the targets, the termination threshold of the clustering is 2r d , and the obtained clustering result is the target formation result for There is r ij ⁇ 2r d , where n p is the number of target formations, and r ij represents the distance between target i and target j.
  • Step 3.2 The sensor works in the "track-search" mode. According to the sensor type carried by the UAV and the position distribution of the target to be observed, the equivalent observation gain of the UAV i to the target in the target formation N p can be calculated. coefficient g ip,k ;
  • ⁇ ip,k represents the angular domain where the observed target formation is located
  • n TS represents the multiple of the dwell time of the sensor field of view relative to the original time
  • g TS represents the performance gain of the corresponding sensor
  • d represents the number of improved observation dimensions
  • Step 3.3 Calculate the maximum number of UAV formations that meet the observable conditions:
  • n P the number of UAVs only equipped with passive sensors
  • n A the number of other UAVs
  • [ ] D represents the round-down operation
  • Step 3.4 Represent the UAV formation result as n q is the number of UAV formations. If the number of target formations is not more than n max , the solution model of the optimal observation configuration between UAVs and targets is:
  • [ ] U represents the round-up operation
  • the B ip at time k is defined as M ij,k is the Fisher information matrix of the drone i to the target j
  • J s is the global observation performance
  • the solution model of the optimal observation configuration between the UAV and the target is:
  • Step 4 UAVs are grouped into formations according to the relationship between the UAV and the target observation configuration.
  • the performance function of the UAV formation M q to which the UAV i belongs to the target formation that should be observed is:
  • the set of UAV numbers of sensors N r is the length of the rolling time domain, j 1 and j 2 represent When , the numbers of the two targets closest to the drone i in the dual target line, and Indicates different parts in the rejection area, and the division method is shown in Figure 5.
  • the model-based predictive control method can be used to solve the UAV control sequence Realize multi-machine heterogeneous sensor multi-target cooperative tracking.
  • An observation-oriented optimization-oriented multi-machine heterogeneous sensor cooperative multi-target tracking method of the present invention includes:
  • Step1 Obtain the motion state x i,k of each UAV i at time k, the sensor measurement covariance matrix R i,k and the control sequence at the previous time through the communication between the UAVs Among them, j ⁇ N, i ⁇ M, N represents the set of all targets, and M represents the set of all UAVs. .
  • Step2 According to the motion state xi,k of each UAV i at time k , the sensor measurement covariance matrix Ri ,k and the control sequence at the previous time The estimated results x ij,k of the position states of all targets are obtained by the distributed heterogeneous sensor fusion method.
  • Step 3 According to the estimation result x ij,k of the position state of the target, determine whether the target closest to the UAV has changed or whether the formation composition of the target has split, merged or handed over.
  • Step 4 According to the sensor configuration of the multi-UAV observation system and the influence of different working modes of different sensors on the observation performance, solve the optimal observation configuration relationship between the UAV and the target.
  • Step 5 According to the configuration relationship between the UAV and the target observation configuration, all UAVs are grouped into formation, and the model-based predictive control method is used to solve the performance function of the target formation observed by the UAV formation M q to which the UAV i belongs, and obtain no Human-machine control sequence Realize multi-machine heterogeneous sensor multi-target cooperative tracking.
  • the optimal observation configuration relationship between the UAV and the target is solved, specifically:
  • the method of hierarchical clustering based on Euclidean distance is used to cluster all the targets to obtain the target formation result; the termination threshold of the clustering is 2r d , and the target formation result is expressed as for have where n p is the number of target formations, represents the distance between target j 1 and target j 2 .
  • the equivalent observation gain coefficient g ip,k of the UAV i to the target in the target formation N p is calculated; the sensors carried by the UAV are in the "tracking- search" mode.
  • ⁇ ip,k represents the angular domain where the observed target formation is located
  • n TS represents the multiple of the dwell time of the sensor field of view relative to the original time
  • g TS represents the sensor performance gain
  • d represents the number of improved observation dimensions.
  • the number of UAVs equipped with only passive sensors is n P
  • the number of UAVs equipped with active sensors is n A
  • [ ⁇ ] D represents the round-down operation.
  • n q is the number of UAV formations. If the number of target formations is not more than n max , the solution model of the optimal observation configuration between UAVs and targets is:
  • [ ] U represents the round-up operation
  • the B ip at time k is defined as M ij
  • k is the Fisher information matrix of the drone i to the target j
  • J s is the global observation performance; if the number of target formations is more than n max , the solution model is based on the optimal observation configuration of the drone and the target
  • the proposed method can timely re-plan the observation configuration relationship according to the geometric position relationship between UAV-target and the dynamic change of target formation composition under the "tracking-search" working model of heterogeneous sensors, thereby significantly improving the global Observe performance.

Landscapes

  • Engineering & Computer Science (AREA)
  • Aviation & Aerospace Engineering (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Remote Sensing (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Automation & Control Theory (AREA)
  • Radar Systems Or Details Thereof (AREA)
  • Control Of Position, Course, Altitude, Or Attitude Of Moving Bodies (AREA)

Abstract

一种面向观测优化的多机异类传感器协同多目标跟踪方法,其针对多无人机异类传感器协同多目标跟踪问题,为协同目标跟踪技术。针对单目标编队,首先构造了拒止区域,得到适用于无人机连续运动的包络路径,并根据无人机的位置与传感器配置,构造无人机编队对目标编队跟踪的性能函数。令传感器在"跟踪-搜索"模式下工作,根据异类传感器的性能增益与可观测性条件建立模型,求解无人机与目标的最优观测配置关系。当目标编队组成变化时,适时地在线重规划无人机编队组成与观测配置关系,从而提出了全局观测优化的多目标协同跟踪方法。能够根据异类传感器性能、可观测性约束、目标编队动态变化有效改善了对多目标的全局观测性能。

Description

面向观测优化的多机异类传感器协同多目标跟踪方法
本申请要求于2020年9月18日提交中国专利局、申请号为2020109903945、发明名称为“面向观测优化的多机异类传感器协同多目标跟踪方法”的中国专利申请的优先权,其全部内容通过引用结合在本申请中。
技术领域
本发明涉及协同目标跟踪技术领域,特别是涉及面向观测优化的多机异类传感器协同多目标跟踪方法。
背景技术
受益于无人机信息融合技术与协同探测技术的快速发展,可使用多架搭载不同传感器的无人机在广域空间内对多运动目标、敏感区域长时间遂行侦察与跟踪任务。由于多无人机多目标协同跟踪是典型NP-hard问题,并且需要根据传感器特点提出有效鲁棒的解决方法。
本发明旨在针对异类传感器对多目标的观测优化问题,研究了多机多目标协同跟踪样式,并结合异类传感器工作模式特点、多无人机的可观测性条件,为优化多无人机协同跟踪系统对多目标跟踪过程中的全局观测性能,对多无人机多目标的观测配置问题与紧环绕跟踪问题进行深入研究,提出有效可靠的新方法。
发明内容
基于此,本发明的目的是提供面向观测优化的多机异类传感器协同多目标跟踪方法,提高了全局的观测性能。
1.多无人机协同standoff目标编队跟踪
(1)基于分布式异类传感器融合的目标状态估计
假设二维平面中使用多架无人机对编队运动的多个目标遂行侦察监视任务。无人机利用机载传感器能够感知一定范围内的目标,并利用状态滤波算法对目标进行连续跟踪。对于有源传感器的目标跟踪问题,已有大量的文献进行了相关研究,本文采用经典的扩展卡尔曼滤波器(Extended KalmanFilter,EKF)方法得到连续的目标状态估计结果。考虑到多无人机编队系统中存在无源传感器的情况,为使得后续目标状态的分布式融合估 计的顺利进行,需要使用无源传感器得到目标估计结果,因此本文引入了偏差补偿的伪线性卡尔曼滤波器(Bias-Compensated Pseudolinear Kalman Filter,BC-PLKF)用于连续估计目标状态。选用EKF和BC-PLKF方法分别用于有/无源传感器的目标状态跟踪滤波,可统一待估计的目标状态向量,一方面,当无人机搭载有多个异类传感器时,可直接使用序贯方法得到局部的目标状态融合估计,另一方面,当无人机编队中存在异类传感器时,方便使用现有分布式目标状态融合估计方法。
在各个无人机得到目标的状态估计结果后,通过通信链路与邻近无人机进行通信,通信内容包括对目标的观测结果、无人机运动状态信息以及未来的控制序列信息,而后可利用基于信息一致性的分布式目标状态融合估计方法得到一致性的目标状态估计。针对复杂的机动目标,则可以结合交互多模型提高目标状态估计精度。目标状态融合过程中假设各个无人机的目标观测结果已完成时间配准、航迹关联和系统偏差估计。
(2)多机standoff目标编队跟踪性能函数
假设通过分布式目标状态融合估计后,k时刻无人机i对目标j的位置状态结果为x ij,k=[x ij,k,y ij,k] T,其中N={j}表示所有目标节点的集合,M={i}表示所有无人机节点的集合。设无人机的运动状态为
Figure PCTCN2020135190-appb-000001
分别表示无人机k时刻的位置p i,k=[x i,k,y i,k] T、航向、速度和转向角速度。量测方程统一表示为z i,k=h(p i,k,x ij,k)。由于目标中可能存在多个编队的情况,则目标编队p中的目标节点的集合表示为
Figure PCTCN2020135190-appb-000002
对编队N p进行定位跟踪无人机的集合为
Figure PCTCN2020135190-appb-000003
为保证多无人机对多目标跟踪时的稳定性和连续性,假设一个目标编队有且仅被一个无人机编队观测,而单个无人机编队可以观测多个目标编队。为保证多无人机对多目标跟踪时的稳定性和连续性,假设一个目标编队有且仅被一个无人机编队观测,而单个无人机编队可以观测多个目标群。图2展示了多无人机对多目标编队协同standoff跟踪场景。
无人机i∈M q的视角下,k时刻,无人机编队M q对目标编队N p中的目标的观测能力可以描述为:
Figure PCTCN2020135190-appb-000004
其中M ij,k为无人机i对目标j的Fisher信息矩阵,p表示目标编队的编号,q表示无人机编队编号,N p表示第p个目标编队,M q表示第q个无人机编队。
Figure PCTCN2020135190-appb-000005
其中,H xij,k为传感器量测方程对目标状态的雅克比矩阵,R i,k为传感器量测误差协方差矩阵。
(3)拒止区域
为使无人机与目标之间保持稳定的安全距离或受限于某些传感器的观测视野,使得固定翼无人机能够持续稳定的对目标进行连续观测,要求无人机与目标之间保持稳定的standoff距离r d,因此设置以目标为圆心,standoff距离为半径的圆为对目标的拒止区域,从而保证无人机沿着拒止区域的边缘飞行。当多个目标彼此靠近,其对应的拒止区域相互重叠,从而形成一个目标群/编队,对应拒止区域为所有目标拒止区域的并集。将传统standoff目标跟踪的概念推广到多目标场景,则要求无人机与距其最近目标之间的距离不小于预设的standoff距离,形成无人机沿着多个目标圆的外包络环绕飞行的standoff多目标跟踪样式,又被称作紧环绕(Tight Circumnavigation)跟踪。当目标之间的间距较大(但小于2r d)时,拒止区域会产生陡峭的凹陷,介于无人机对长时控制序列的预测能力有限,无人机无法理想地直接沿着拒止区域的包络运动。为提高无人机控制的稳定性和控制优化效率,根据无人机运动性能,使用无人机的最小转弯半径圆填补陡峭的凹陷,得到修补后的拒止区域,从而得到适用于无人机连续运动的包络路径,其中最小转弯半径使用无人机的最大速度计算得到
Figure PCTCN2020135190-appb-000006
其中,R min表示无人机的最小转弯半径,v max表示无人机的最大速度,ω max表示无人机的最大转向角速度。
在不显著影响观测性能的条件下,为简化计算,直接在两圆的中部使用矩形填补凹陷区域成较为规则的图形,双目标拒止区域的构造示意图如 图3所示。
当无人机观测编队的数目较少时,需要部分无人机对多个目标编队进行观测,采取相似的思路构建拒止区域。以两个目标编队为例,其中编队之间的最小距离不小于2r d,如图4所示,其中添加矩形的宽度d min由两圆相切时的矩形宽度确定。
当无人机位于拒止区域内时,应保证无人机控制的优化方向径向地指向拒止区域外,同时尽可能保证性能函数及其导数在拒止区域附近有着相近的尺度。因此将编队目标的拒止区域分为
Figure PCTCN2020135190-appb-000007
Figure PCTCN2020135190-appb-000008
两部分,如图5所示。在区域
Figure PCTCN2020135190-appb-000009
中,为保证控制量的优化方向径向地远离目标,选择距离无人机i最近的目标j,定义其性能函数为:
Figure PCTCN2020135190-appb-000010
其中
Figure PCTCN2020135190-appb-000011
当无人机在区域
Figure PCTCN2020135190-appb-000012
中,则需要使控制量的优化方向垂直于目标连线向外,选择距离无人机i最近的目标连线
Figure PCTCN2020135190-appb-000013
定义性能函数为:
Figure PCTCN2020135190-appb-000014
其中,令
Figure PCTCN2020135190-appb-000015
表示无人机到目标连线的距离,即
Figure PCTCN2020135190-appb-000016
综上所述,对无人机i∈M q,k时刻无人机编队M q对目标编队N p的性能函数可以表述为:
Figure PCTCN2020135190-appb-000017
注意到,对于仅测角无源传感器,其FIM(Fisher Information Matrix Fisher信息矩阵)不满秩,无法求导,本方法中其FIM行列式及其导数 的计算使用FIM的迹代替。
结合基于模型预测控制的方法,对于无人机i,无人机系统对多目标的观测性能函数为:
Figure PCTCN2020135190-appb-000018
其中N r为滚动时域的长度。
可采用优化后的MPC方法在分散式框架下求解无人机控制变量,并对多无人机的圆周运动方向进行协同控制。
2.多无人机编队协同跟踪
当目标有多个编队时,编队内目标间隔较近,而编队间的距离相对较远,因此需要对多目标进行划分,划分为多个目标编队(群)。同时,为了提高对编队目标的观测、定位、跟踪能力,需要对多无人机的跟踪任务目标进行规划,以获得最佳的战场态势感知能力。此外,考虑基于旋转扫描体制传感器的工作模式、传感器观测类型对目标分群、目标任务规划的影响。
(1)传感器工作模式
机载传感器可根据其自身特点和任务目标的要求工作在不同的模式下,常见的两种工作模式分别为目标搜索与目标跟踪,其中目标跟踪工作模式又可以根据目标的优先级程度,划分为目标捕获、目标跟踪、记忆跟踪与目标验证等。
不同的传感器有不同的工作模式。以相控阵雷达为例,通过多次相参积累,可以提高测距的信噪比,改善对目标的定位跟踪精度。使用云台搭载的相机传感器,可以通过将目标始终置于视野之中提高对目标观测的采样率,进而提高目标跟踪性能。
为兼顾目标跟踪观测质量和对新目标的搜索发现,定义一种跟踪-搜索工作模式。该工作模式下,假设传感器对已知的重点目标区域,通过提高传感器视野的驻留时间,增加观测的采样率或提高传感器信噪比的方式,提高对目标的观测性能;对连续的非目标区域或次要目标区域则按照正常的采样率,从而保证对可能出现的新目标进行搜索并检测,同时保证对次要目标的状态更新。设k时刻目标编队N p所在区域的角域范围为β p,k,则传感器的观测周期由T s变为T p,k
Figure PCTCN2020135190-appb-000019
其中n TS表示传感器视野驻留时间相对于原有时间的倍数。n TS=1表示没有提高驻留时间,传感器按照在全角域搜索的工作模式感知或跟踪目标,易知,此时T p,k=T s
在连续时间域上,不考虑目标过程噪声,仅考虑目标状态的Fisher信息矩阵,则t时刻的FIM可定义为:
Figure PCTCN2020135190-appb-000020
其中Y t0为t 0时刻的FIM,H τ为观测方程的线性化矩阵,R表示传感器量测误差协方差矩阵。
离散化后,得到k时刻的FIM为:
Figure PCTCN2020135190-appb-000021
若传感器采用跟踪的工作模式对部分重点密集目标进行观测,则新的FIM等价为:
Figure PCTCN2020135190-appb-000022
其中,R TS表示跟踪工作模型下得到改善的等效传感器量测误差矩阵。
若传感器采用搜索的工作模式对次要目标进行观测,则FIM等价为:
Figure PCTCN2020135190-appb-000023
整合式(13)和式(14),得到不同工作模式下的新增FIM的行列式为:
Figure PCTCN2020135190-appb-000024
其中g p,k定义为传感器在“跟踪-搜索”工作模式下,k时刻无人机对目标编队N p内目标的等效观测量增益。对无人机i,其定义为:
Figure PCTCN2020135190-appb-000025
其中g TS表示由于传感器采用视野驻留工作模式,所提高的传感器性 能,d表示改善的传感器维度数。若传感器通过驻留可以提高测距和测角的采样率,则g TS≈n TS,d=2。以相控阵雷达为例,当其工作在增程模式时,使用相参积累可以有效提高测距的信噪比,则g TS=n TS,d=1;若使用非相参积累,则一般g TS<n TS。若传感器工作在搜索模式下对次要目标进行观测,则g TS=1,g ip,k<1,即传感器会降低对次要目标的观测能力。若传感器工作在跟踪模式下对重点目标进行观测,则g TS>1,一般情况下,有g ip,k>1,即传感器可以有效提高对重点目标的观测能力。若传感器对全角域按照搜索的模型进行探测,则g ip,k=1,传感器对目标的观测能力不变。若传感器对全角域按照跟踪的模型进行探测,且g TS<n TS,则g ip,k<1,反而会降低整体的观测能力。因此有区别地针对不同目标区域采取不同的工作模式,是有效提高整个无人机系统观测能力的关键。
(2)多无人机与多目标的观测配置
为更好地使用无人机遂行跟踪任务,得到持续稳定的最佳观测状态,需要将不同的目标群分配给不同的目标,该方法包含两个步骤,一是对多目标进行分割,得到多个目标编队,二是根据观测要求和无人机当前运动状态,将无人机配置到不同的目标上。
1)目标分群
首先需要对多个目标进行分割,得到多个目标编队。根据standoff跟踪的要求,若一个目标集合中任意邻近的两个目标之间的距离小于2r d,则认为该目标集合为一个目标编队。因此可以使用基于欧氏距离的分级聚类的方法对所有目标进行聚类,聚类的终止门限为2r d,从而得到目标编队结果N p,p=1,…,n p,n p为目标编队的数量。
2)无人机观测配置
根据无人机当前位置上对目标的观测能力,需要将无人机配置到不同的目标。为此,通常引入最优分配(optimal assignment)方法得到对全部目标的总观测量最优的无人机-目标编队观测配置结果,但没有考虑异类传感器不同工作模型对整体观测性能的影响,为此本发明提出了一种适用于异类传感器协同多目标跟踪的无人机与目标之间的最优观测配置关系求解方法。
忽略先验FIM,使用当前时刻的新增的FIM定义单无人机对不同目标编队的观测量,即
Figure PCTCN2020135190-appb-000026
忽略时间标签,观测配置结果的目标函数表示为:
Figure PCTCN2020135190-appb-000027
其中{x ip}表示值为0或1的二值分配矩阵。若x ip=1,则表示第i个无人机在跟踪模式下对目标编队N p进行观测;若x ip=0,则第i个无人机在搜索模式下对目标编队N p进行观测。
在讨论分配矩阵的约束前,需要考虑无人机搭载的传感器类型。若无人机仅搭载无源传感器时,为保证无人机编队在不进行特殊机动的条件下对目标仍具有可观测性,其所在无人机编队中的无人机数量应不少于2架;若无人机仅搭载主动传感器或同时搭载主/被动传感器,则单个无人机便能够实现对目标的观测、定位和跟踪。假设仅搭载无源传感器的无人机数量为n P,其他无人机数目为n A,则在满足可观测条件下的最大无人机编队数目为
Figure PCTCN2020135190-appb-000028
其中[·] D表示向下取整操作。
下面分两种情况讨论分配矩阵的约束。
a)目标编队数目不多于最大无人机编队数目
当目标编队数目小于或等于最大无人机编队数目时,无人机编队可以在满足对目标可观测性的基础上,增加编队中无人机的数量,提高对目标的观测性能。因此,对于仅搭载无源传感器的无人机,要求对同一目标编队观测的无人机数量不少于两架,即
Figure PCTCN2020135190-appb-000029
相似地,对其他无人机则要求,对同一目标编队观测的无人机数量不少于一架,即
Figure PCTCN2020135190-appb-000030
其他无人机指搭载有源传感器的无人机。
此外,当无人机数目较多时,在资源配置时,为避免多数无人机配置到单个目标编队的情况,对观测同一个目标编队的无人机数量上界进行约 束,即
Figure PCTCN2020135190-appb-000031
其中[·] U表示向上取整操作。
同时,在跟踪模式下,每一个无人机编队仅对一个目标编队进行观测,即
Figure PCTCN2020135190-appb-000032
综上所述,整合式-中的约束条件,得到无人机与目标的最优观测配置的求解模型为:
Figure PCTCN2020135190-appb-000033
Figure PCTCN2020135190-appb-000034
Figure PCTCN2020135190-appb-000035
x ip={0,1},i∈M,p=1,…,n p
b)目标编队数目多于最大无人机编队数目
当目标编队数目多于最大无人机编队数目时,则要求所有无人机按照满足观测性情况下的最小规模组成编队,即
Figure PCTCN2020135190-appb-000036
同时,一个无人机编队则需要在跟踪模式下至少观测一个目标编队,即
Figure PCTCN2020135190-appb-000037
则无人机与目标的最优观测配置的求解模型为:
Figure PCTCN2020135190-appb-000038
Figure PCTCN2020135190-appb-000039
Figure PCTCN2020135190-appb-000040
x ip={0,1},i∈M,p=1,…,n p
根据式(23)和式(26)中的两类模型,使用0-1整数线性规划(Zero-One IntegerLinearProgramming)方法可得到最佳的无人机编队配置结果以及对应的无人机-目标的观测配置结果。
(3)在线重规划
由于目标是不断运动,目标编队的组成也可能是变化的。
1)当目标由一个目标编队跨越到另一个编队(目标交接)时,无人机需要重新确定无人机编队组成,以确保多无人机系统对多目标的总体观测性能最优。
2)当单个目标编队分裂成多个编队时,在保证对目标可观测性的基础上,无人机也需要分裂成多个无人机编队,或者目标编队数目较多时,甚至可能需要其他无人机编队协作地对分裂出去的目标编队进行观测。
3)当多个目标编队合并成一个编队时,无人机编队可以通过合并,提高对新目标编队的观测性能。
注意若某一目标编队进行上述的交接、分裂与合并过程时,对应的无人机编队并非简单地进行相应的分裂与合并,而是需要根据当前时刻无人机与目标的位置关系,根据最优观测性这一优化目标重新确定无人机的编队构成,以及无人机编队与目标编队的对应关系。因此,当目标编队构成发生变化时,需重新规划无人机-目标的观测配置关系。
此外,由于无人机与目标均在不断运动,当其工作在跟踪-搜索模式时,目标区域在无人机视野中的角度范围根据其相对几何位置关系随时间不断变化。尤其对于目标编队数目相对较多的情况,一个无人机编队可能需要重点观测多个目标编队,当该无人机编队中的某个无人机距离某一目标编队较远时,该无人机对其观测能力较差,而该编队外的其他无人机可能对该目标具有更优的观测性能,因此当距离某一无人机最近的目标编号发生变化时,则重新优化无人机对目标编队的观测配置关系。
本发明一种面向观测优化的多机异类传感器协同多目标跟踪方法包括:
通过各无人机之间的通信,获得各无人机i在k时刻的运动状态x i,k、传感器量测协方差矩阵R i,k和上时刻的控制序列
Figure PCTCN2020135190-appb-000041
其中,j∈N,i∈M,N表示所有目标的集合,M表示所有无人机的集合;
根据各无人机i在k时刻的运动状态x i,k、传感器量测协方差矩阵R i,k和上时刻的控制序列
Figure PCTCN2020135190-appb-000042
通过分布式异类传感器融合方法得到所有目标的位置状态的估计结果x ij,k
根据目标的位置状态的估计结果x ij,k,判断距离无人机最近的目标是否发生变化或者目标的编队组成是否发生分裂、合并或交接;
若是,则根据多无人机观测系统的传感器配置和不同传感器的不同工作模式对观测性能的影响,求解无人机与目标之间的最优观测配置关系;
若否,则根据无人机与目标观测配置关系对所有无人机进行分组编队,使用基于模型预测控制方法对无人机i所属的无人机编队M q观测的目标编队的性能函数求解,获得无人机控制序列
Figure PCTCN2020135190-appb-000043
实现多机异类传感器多目标协同跟踪。
根据本发明提供的具体实施例,本发明公开了以下技术效果:
大量实验结果表明了所述方法具有更高的全局观测性能,并对不同的无人机数目、目标数目、目标编队数目具有较好的适应性。同时,所提方法能够在异类传感器的“跟踪-搜索”工作模型下,根据无人机-目标的几何位置关系、目标编队组成的动态变化适时地重规划观测配置关系,从而显著改善了全局的观测性能。
说明书附图
为了更清楚地说明本发明实施例或现有技术中的技术方案,下面将对 实施例中所需要使用的附图作简单地介绍,显而易见地,下面描述中的附图仅仅是本发明的一些实施例,对于本领域普通技术人员来讲,在不付出创造性劳动性的前提下,还可以根据这些附图获得其他的附图。
图1为本发明面向观测优化的多机异类传感器协同多目标跟踪方法实施流程图;
图2为本发明多机多目标协同跟踪示意图;
图3为本发明单个目标编队的拒止区域的构造示意图;
图4为本发明多个目标编队的拒止区域示意图;
图5为本发明拒止区域的划分示意图。
具体实施方式
下面将结合本发明实施例中的附图,对本发明实施例中的技术方案进行清楚、完整地描述,显然,所描述的实施例仅仅是本发明一部分实施例,而不是全部的实施例。基于本发明中的实施例,本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例,都属于本发明保护的范围。
为使本发明的上述目的、特征和优点能够更加明显易懂,下面结合附图和具体实施方式对本发明作进一步详细的说明。
结合图1所述的本发明实施流程图,对本发明作进一步详细描述。
利用多架无人机协同跟踪多个目标的场景示意图如图2所示。通过通信链路或中继通信等方式,各个无人机可以相互传递无人机自身的运动状态、传感器性能、对目标的观测以及上一时刻的控制序列优化结果,并利用分布式目标状态融合估计方法,得到关于目标的状态估计结果。而后,在分散式框架下,各个无人机根据上述信息,判断是否需要重新规划无人机与目标编队之间的观测配置关系,最后根据观测配置关系,在给定的拒止区域的包络上协同跟踪目标编队,使全局观测结果最优。无人机之间的通信只发生在目标状态融合估计过程中,在后续的观测配置优化、协同目标跟踪过程中,则使用分散式框架,能够有效降低对实时通信能力的需求,易于工程实践。
无人机i在k时刻运行本方法时的具体实施步骤如下。
步骤1:通过与邻近无人机进行通信,无人机i可获得所有无人机的 运动状态x i,k、传感器量测协方差矩阵R i,k和上时刻的控制序列
Figure PCTCN2020135190-appb-000044
并通过分布式异类传感器融合方法得到所有目标状态的估计结果为x ij,k,其中,j∈N,i∈M,N表示所有目标节点的集合,M表示所有无人机节点的集合;
步骤2:当目标的编队组成发成分裂、合并、交接等变化时,或者距离无人机最近的目标发生变化时,则按照步骤3重新规划无人机与目标之间的观测配置关系,否则跳过步骤3,按照步骤4实施多机多目标协同跟踪;
步骤3:根据多无人机观测系统的传感器配置,考虑不同传感器的不同工作模式对观测性能的影响,求解无人机与目标之间的最优观测配置关系,优化全局观测性能;
步骤3.1:目标分群,若某一目标集合中任意邻近的两个目标之间的距离小于2r d,r d为standoff距离,则该目标集合为一个目标编队,使用基于欧氏距离的分级聚类的方法对所有目标进行聚类,聚类的终止门限为2r d,得到的聚类结果即目标编队结果
Figure PCTCN2020135190-appb-000045
对于
Figure PCTCN2020135190-appb-000046
有r ij<2r d,其中,n p为目标编队的数量,r ij表示目标i和目标j之间的距离。
步骤3.2:传感器在“跟踪-搜索”模式下工作,根据无人机所搭载的传感器类型与所需观测目标的位置分布,可计算无人机i对目标编队N p中目标的等效观测增益系数g ip,k
Figure PCTCN2020135190-appb-000047
其中,β ip,k表示被观测的目标编队所在的角域范围,n TS表示传感器视野驻留时间相对于原有时间的倍数,g TS表示对应传感器性能增益,d表示改善的观测维度数;
步骤3.3:计算满足可观测条件下的最大无人机编队数目:
Figure PCTCN2020135190-appb-000048
其中,仅搭载无源传感器的无人机数量为n P,其他无人机数目为n A,[·] D表示向下取整操作;
步骤3.4:将无人机编队结果表示为
Figure PCTCN2020135190-appb-000049
n q为无人机编队的数量,若目标编队数目不多于n max,则无人机与目标的最优观测配置的求解模型为:
Figure PCTCN2020135190-appb-000050
Figure PCTCN2020135190-appb-000051
Figure PCTCN2020135190-appb-000052
x ip={0,1},i∈M,p=1,…,n p
其中,[·] U表示向上取整操作,k时刻的B ip定义为
Figure PCTCN2020135190-appb-000053
M ij,k为无人机i对目标j的Fisher信息矩阵,J s为全局观测性能;
若目标编队数目多于n max,则无人机与目标的最优观测配置的求解模型为:
Figure PCTCN2020135190-appb-000054
Figure PCTCN2020135190-appb-000055
Figure PCTCN2020135190-appb-000056
x ip={0,1},i∈M,p=1,…,n p
求解上述模型便可得到使全局观测最优的无人机与目标观测配置关系;
步骤4:无人机根据无人机与目标观测配置关系进行分组编队,无人机i所属的无人机编队M q对其应观测的目标编队的性能函数为:
Figure PCTCN2020135190-appb-000057
Figure PCTCN2020135190-appb-000058
其中,
Figure PCTCN2020135190-appb-000059
表示k时刻无人机i对目标j的Fisher信息矩阵,
Figure PCTCN2020135190-appb-000060
为传感器量测方程对目标状态的雅克比矩阵,R i,k为传感器量测误差协方差矩阵,F q表示无人机编队M q观测的目标编队的编号集合,M P表示仅搭载无源传感器的无人机编号的集合,N r为滚动时域的长度,j 1和j 2表示
Figure PCTCN2020135190-appb-000061
时,双目标连线距离无人机i最近的两个目标的编号,
Figure PCTCN2020135190-appb-000062
Figure PCTCN2020135190-appb-000063
表示拒止区域中的不同部分,划分方式如图5所示。基于上述所提性能函数,使用基于模型预测控制方法可以求解无人机控制序列
Figure PCTCN2020135190-appb-000064
实现多机异类传感器多目标协同跟踪。
本发明一种面向观测优化的多机异类传感器协同多目标跟踪方法包括:
Step1:通过各无人机之间的通信,获得各无人机i在k时刻的运动状态x i,k、传感器量测协方差矩阵R i,k和上时刻的控制序列
Figure PCTCN2020135190-appb-000065
其中,j∈N,i∈M,N表示所有目标的集合,M表示所有无人机的集合。.
Step2:根据各无人机i在k时刻的运动状态x i,k、传感器量测协方差矩阵R i,k和上时刻的控制序列
Figure PCTCN2020135190-appb-000066
通过分布式异类传感器融合方法得到所有目标的位置状态的估计结果x ij,k
Step3:根据目标的位置状态的估计结果x ij,k,判断距离无人机最近的目标是否发生变化或者目标的编队组成是否发生分裂、合并或交接。
若是,执行Step4。
Step4:根据多无人机观测系统的传感器配置和不同传感器的不同工作模式对观测性能的影响,求解无人机与目标之间的最优观测配置关系。
若否,执行Step5。
Step5:根据无人机与目标观测配置关系对所有无人机进行分组编队,使用基于模型预测控制方法对无人机i所属的无人机编队M q观测的目标编队的性能函数求解,获得无人机控制序列
Figure PCTCN2020135190-appb-000067
实现多机异类传感器多目标协同跟踪。
所述根据多无人机观测系统的传感器配置和不同传感器的不同工作模式对观测性能的影响,求解无人机与目标之间的最优观测配置关系,具体为:
采用基于欧氏距离的分级聚类的方法对所有目标进行聚类获得目标编队结果;聚类的终止门限为2r d,所述目标编队结果表示为
Figure PCTCN2020135190-appb-000068
对于
Figure PCTCN2020135190-appb-000069
Figure PCTCN2020135190-appb-000070
其中,n p为目标编队的数量,
Figure PCTCN2020135190-appb-000071
表示目标j 1和目标j 2之间的距离。
根据无人机所搭载的传感器类型与观测目标的位置分布,计算无人机i对目标编队N p中目标的等效观测增益系数g ip,k;无人机所搭载的传感器在“跟踪-搜索”模式下工作。
Figure PCTCN2020135190-appb-000072
其中,β ip,k表示被观测的目标编队所在的角域范围,n TS表示传感器视野驻留时间相对于原有时间的倍数,g TS表示传感器性能增益,d表示改善的观测维度数。
计算满足可观测条件下的最大无人机编队数目
Figure PCTCN2020135190-appb-000073
其中,仅搭载无源传感器的无人机数量为n P,搭载有有源传感器的无人机数目为n A,[·] D表示向下取整操作。
将无人机编队结果表示为
Figure PCTCN2020135190-appb-000074
n q为无人机编队的数量,若目标编队数目不多于n max,则无人机与目标的最优观测配置的求解模型为:
Figure PCTCN2020135190-appb-000075
Figure PCTCN2020135190-appb-000076
Figure PCTCN2020135190-appb-000077
x ip={0,1},i∈M,p=1,…,n p
其中,[·] U表示向上取整操作,k时刻的B ip定义为
Figure PCTCN2020135190-appb-000078
M ij,k为无人机i对目标j的Fisher信息矩阵,J s为全局观测性能;若目标编队数目多于n max,根据无人机与目标的最优观测配置的求解模型
Figure PCTCN2020135190-appb-000079
求解无人机与目标之间的最优观测配置关系。
大量实验结果表明了所述方法具有更高的全局观测性能,并对不同的无人机数目、目标数目、目标编队数目具有较好的适应性。同时,所提方法能够在异类传感器的“跟踪-搜索”工作模型下,根据无人机-目标的几何位置关系、目标编队组成的动态变化适时地重规划观测配置关系,从而显著改善了全局的观测性能。
本说明书中各个实施例采用递进的方式描述,每个实施例重点说明的都是与其他实施例的不同之处,各个实施例之间相同相似部分互相参见即 可。
本文中应用了具体个例对本发明的原理及实施方式进行了阐述,以上实施例的说明只是用于帮助理解本发明的方法及其核心思想;同时,对于本领域的一般技术人员,依据本发明的思想,在具体实施方式及应用范围上均会有改变之处。综上所述,本说明书内容不应理解为对本发明的限制。

Claims (4)

  1. 面向观测优化的多机异类传感器协同多目标跟踪方法,其特征在于,各个无人机分别运行本方法实现对多目标的协同跟踪,无人机k在k时刻运行本方法时包括以下步骤:
    步骤1,通过与其他无人机进行通信,无人机i可获得所有无人机的运动状态x i,k、传感器量测协方差矩阵R i,k和上时刻的控制序列
    Figure PCTCN2020135190-appb-100001
    并通过分布式异类传感器融合方法得到所有目标状态的估计结果为x ij,k,其中,j∈N,i∈M,N表示所有目标节点的集合,M表示所有无人机节点的集合;
    步骤2,当目标的编队组成发成分裂、合并、交接变化时,或者距离无人机最近的目标发生变化时,则按照步骤3重新规划无人机与目标之间的观测配置关系,否则跳过步骤3,按照步骤4实施多机多目标协同跟踪;
    步骤3,根据多无人机观测系统的传感器配置,考虑不同传感器的不同工作模式对观测性能的影响,求解无人机与目标之间的最优观测配置关系;
    步骤4,无人机根据无人机与目标观测配置关系进行分组编队,而后根据无人机i所属的无人机编队M q对其应观测的目标编队的性能函数,使用基于模型预测控制方法可以求解无人机控制序列
    Figure PCTCN2020135190-appb-100002
    实现多机异类传感器多目标协同跟踪。
  2. 如权利要求1所述的面向观测优化的多机异类传感器协同多目标跟踪方法,其特征在于,步骤3具体为:
    步骤3.1,使用基于欧氏距离的分级聚类的方法对所有目标进行聚类,聚类的终止门限为2 r d,得到的聚类结果即目标编队结果
    Figure PCTCN2020135190-appb-100003
    对于
    Figure PCTCN2020135190-appb-100004
    Figure PCTCN2020135190-appb-100005
    其中,n p为目标编队 的数量,
    Figure PCTCN2020135190-appb-100006
    表示目标j 1和目标j 2之间的距离;
    步骤3.2,传感器在“跟踪-搜索”模式下工作,根据无人机所搭载的传感器类型与所需观测目标的位置分布,计算无人机i对目标编队N p中目标的等效观测增益系数g ip,k
    Figure PCTCN2020135190-appb-100007
    其中,β ip,k表示被观测的目标编队所在的角域范围,n TS表示传感器视野驻留时间相对于原有时间的倍数,g TS表示对应传感器性能增益,d表示改善的观测维度数;
    步骤3.3,计算满足可观测条件下的最大无人机编队数目
    Figure PCTCN2020135190-appb-100008
    其中,仅搭载无源传感器的无人机数量为n P,其他无人机数目为n A,[·] D表示向下取整操作;
    步骤3.4,将无人机编队结果表示为
    Figure PCTCN2020135190-appb-100009
    n q为无人机编队的数量,若目标编队数目不多于n max,则无人机与目标的最优观测配置的求解模型为
    Figure PCTCN2020135190-appb-100010
    Figure PCTCN2020135190-appb-100011
    Figure PCTCN2020135190-appb-100012
    x ip={0,1},i∈M,p=1,…,n p
    其中,[·] U表示向上取整操作,k时刻的B ip定义为
    Figure PCTCN2020135190-appb-100013
    M ij,k为无人机i对目标j的Fisher信息矩阵,J s为全局观测性能;若目标编队数目多于n max,则无人机与目标的最优观测配置的求解模型为
    Figure PCTCN2020135190-appb-100014
    Figure PCTCN2020135190-appb-100015
    Figure PCTCN2020135190-appb-100016
    x ip={0,1},i∈M,p=1,…,n p
    求解上述模型便可得到使全局观测最优的无人机与目标观测配置关系。
  3. 如权利要求1所述的面向观测优化的多机异类传感器协同多目标跟踪方法,其特征在于,步骤4中的性能函数具体为:
    Figure PCTCN2020135190-appb-100017
    Figure PCTCN2020135190-appb-100018
    其中,
    Figure PCTCN2020135190-appb-100019
    表示k时刻无人机i对目标j的Fisher信息矩阵,
    Figure PCTCN2020135190-appb-100020
    为传感器量测方程对目标状态的雅克比矩阵,R i,k为传感器量测误差协方差矩阵,F q表示无人机编队M q观测的目标编队的编号集合,M P表示仅搭载无源传感器的无人机编号的集合,
    Figure PCTCN2020135190-appb-100021
    Figure PCTCN2020135190-appb-100022
    表示拒止区域中的不同部分,j 1和j 2表示
    Figure PCTCN2020135190-appb-100023
    时,双目标连线距离无人机i最近的两个目标的编号,N r为滚动时域的长度。
  4. 面向观测优化的多机异类传感器协同多目标跟踪方法,其特征在 于,所述方法包括:
    通过各无人机之间的通信,获得各无人机i在k时刻的运动状态x i,k、传感器量测协方差矩阵R i,k和上时刻的控制序列
    Figure PCTCN2020135190-appb-100024
    其中,j∈N,i∈M,N表示所有目标的集合,M表示所有无人机的集合;
    根据各无人机i在k时刻的运动状态x i,k、传感器量测协方差矩阵R i,k和上时刻的控制序列
    Figure PCTCN2020135190-appb-100025
    通过分布式异类传感器融合方法得到所有目标的位置状态的估计结果x ij,k
    根据目标的位置状态的估计结果x ij,k,判断距离无人机最近的目标是否发生变化或者目标的编队组成是否发生分裂、合并或交接;
    若是,则根据多无人机观测系统的传感器配置和不同传感器的不同工作模式对观测性能的影响,求解无人机与目标之间的最优观测配置关系;
    若否,则根据无人机与目标观测配置关系对所有无人机进行分组编队,使用基于模型预测控制方法对无人机i所属的无人机编队M q观测的目标编队的性能函数求解,获得无人机控制序列
    Figure PCTCN2020135190-appb-100026
    实现多机异类传感器多目标协同跟踪。
PCT/CN2020/135190 2020-09-18 2020-12-10 面向观测优化的多机异类传感器协同多目标跟踪方法 WO2022057107A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202010990394.5 2020-09-18
CN202010990394.5A CN112051862B (zh) 2020-09-18 2020-09-18 面向观测优化的多机异类传感器协同多目标跟踪

Publications (1)

Publication Number Publication Date
WO2022057107A1 true WO2022057107A1 (zh) 2022-03-24

Family

ID=73603466

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/135190 WO2022057107A1 (zh) 2020-09-18 2020-12-10 面向观测优化的多机异类传感器协同多目标跟踪方法

Country Status (2)

Country Link
CN (1) CN112051862B (zh)
WO (1) WO2022057107A1 (zh)

Cited By (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116380148A (zh) * 2023-04-06 2023-07-04 中国人民解放军93209部队 多传感器目标跟踪系统的两级时空误差标校方法及装置
CN116380502A (zh) * 2023-04-07 2023-07-04 济南大学 一种eps系统多点同步nvh数据采集布局方法及系统
CN116400715A (zh) * 2023-03-02 2023-07-07 中国人民解放军战略支援部队信息工程大学 模型误差条件下基于CNN+BiLSTM神经网络的多无人机协同直接跟踪方法
CN116489602A (zh) * 2023-06-21 2023-07-25 北京航空航天大学 一种分布式容错目标跟踪方法、系统、设备及介质
CN116974297A (zh) * 2023-06-27 2023-10-31 北京五木恒润科技有限公司 基于多目标优化的冲突消解方法、装置、介质及电子设备
CN117148730A (zh) * 2023-11-01 2023-12-01 北京航空航天大学 一种时变分组编队跟踪控制方法、系统及电子设备
CN117270393A (zh) * 2023-10-07 2023-12-22 重庆大学 智能机器人集群协同控制系统
CN117687309A (zh) * 2024-02-04 2024-03-12 北京航空航天大学 用于异构集群系统的异步约束输出编队跟踪方法及系统

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111427379B (zh) * 2020-04-19 2022-05-31 中国人民解放军海军航空大学 观测驱动的多无人机协同standoff目标跟踪方法
CN112051862B (zh) * 2020-09-18 2021-09-14 中国人民解放军海军航空大学 面向观测优化的多机异类传感器协同多目标跟踪
CN113655811B (zh) * 2021-08-25 2024-05-07 北京理工大学 隐式通信下无人机协同跟踪地面目标的期望队形获得方法

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5793931A (en) * 1996-11-18 1998-08-11 The United States Of America As Represented By The Secretary Of The Army Method and apparatus for multi-sensor, multi-target tracking using intelligent search techniques
CN102997911A (zh) * 2012-12-13 2013-03-27 中国航空无线电电子研究所 一种被动传感器组网探测多目标方法
CN103235315A (zh) * 2013-04-17 2013-08-07 苏州麦杰智能科技有限公司 一种多机动目标跟踪系统
CN104252579A (zh) * 2014-09-18 2014-12-31 中国人民解放军海军航空工程学院 扩展广义s-维分配编队目标机动模式判断方法
CN111427379A (zh) * 2020-04-19 2020-07-17 中国人民解放军海军航空大学 观测驱动的多无人机协同standoff目标跟踪方法
CN112051862A (zh) * 2020-09-18 2020-12-08 中国人民解放军海军航空大学 面向观测优化的多机异类传感器协同多目标跟踪

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109116349B (zh) * 2018-07-26 2022-12-13 西南电子技术研究所(中国电子科技集团公司第十研究所) 多传感器协同跟踪联合优化决策方法
CN109508000B (zh) * 2018-12-16 2021-09-03 西南电子技术研究所(中国电子科技集团公司第十研究所) 异构多传感器多目标跟踪方法

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5793931A (en) * 1996-11-18 1998-08-11 The United States Of America As Represented By The Secretary Of The Army Method and apparatus for multi-sensor, multi-target tracking using intelligent search techniques
CN102997911A (zh) * 2012-12-13 2013-03-27 中国航空无线电电子研究所 一种被动传感器组网探测多目标方法
CN103235315A (zh) * 2013-04-17 2013-08-07 苏州麦杰智能科技有限公司 一种多机动目标跟踪系统
CN104252579A (zh) * 2014-09-18 2014-12-31 中国人民解放军海军航空工程学院 扩展广义s-维分配编队目标机动模式判断方法
CN111427379A (zh) * 2020-04-19 2020-07-17 中国人民解放军海军航空大学 观测驱动的多无人机协同standoff目标跟踪方法
CN112051862A (zh) * 2020-09-18 2020-12-08 中国人民解放军海军航空大学 面向观测优化的多机异类传感器协同多目标跟踪

Cited By (15)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116400715A (zh) * 2023-03-02 2023-07-07 中国人民解放军战略支援部队信息工程大学 模型误差条件下基于CNN+BiLSTM神经网络的多无人机协同直接跟踪方法
CN116380148B (zh) * 2023-04-06 2023-11-10 中国人民解放军93209部队 多传感器目标跟踪系统的两级时空误差标校方法及装置
CN116380148A (zh) * 2023-04-06 2023-07-04 中国人民解放军93209部队 多传感器目标跟踪系统的两级时空误差标校方法及装置
CN116380502A (zh) * 2023-04-07 2023-07-04 济南大学 一种eps系统多点同步nvh数据采集布局方法及系统
CN116380502B (zh) * 2023-04-07 2024-03-08 济南大学 一种eps系统多点同步nvh数据采集布局方法及系统
CN116489602A (zh) * 2023-06-21 2023-07-25 北京航空航天大学 一种分布式容错目标跟踪方法、系统、设备及介质
CN116489602B (zh) * 2023-06-21 2023-08-18 北京航空航天大学 一种分布式容错目标跟踪方法、系统、设备及介质
CN116974297A (zh) * 2023-06-27 2023-10-31 北京五木恒润科技有限公司 基于多目标优化的冲突消解方法、装置、介质及电子设备
CN116974297B (zh) * 2023-06-27 2024-01-26 北京五木恒润科技有限公司 基于多目标优化的冲突消解方法、装置、介质及电子设备
CN117270393A (zh) * 2023-10-07 2023-12-22 重庆大学 智能机器人集群协同控制系统
CN117270393B (zh) * 2023-10-07 2024-05-17 重庆大学 智能机器人集群协同控制系统
CN117148730A (zh) * 2023-11-01 2023-12-01 北京航空航天大学 一种时变分组编队跟踪控制方法、系统及电子设备
CN117148730B (zh) * 2023-11-01 2024-01-16 北京航空航天大学 一种时变分组编队跟踪控制方法、系统及电子设备
CN117687309A (zh) * 2024-02-04 2024-03-12 北京航空航天大学 用于异构集群系统的异步约束输出编队跟踪方法及系统
CN117687309B (zh) * 2024-02-04 2024-04-16 北京航空航天大学 用于异构集群系统的异步约束输出编队跟踪方法及系统

Also Published As

Publication number Publication date
CN112051862A (zh) 2020-12-08
CN112051862B (zh) 2021-09-14

Similar Documents

Publication Publication Date Title
WO2022057107A1 (zh) 面向观测优化的多机异类传感器协同多目标跟踪方法
CN112000130B (zh) 一种无人机的多机协同高精度建图定位系统
CN109613931B (zh) 基于生物社会力的异构无人机集群目标跟踪系统及方法
CN108196575B (zh) 一种无人机任务分配及路线规划方法
CN107844129B (zh) 基于路径规划和信息融合的多无人机协同搜索方法及系统
US8718838B2 (en) System and methods for autonomous tracking and surveillance
CN111831008A (zh) 一种基于分布式架构的无人机编队协同控制系统及其方法
CN108020226B (zh) 一种固定翼无人机自主避障的航迹规划方法
CN110032214B (zh) 基于矢量场的快速Standoff目标跟踪方法
CN114047785B (zh) 一种无人机集群协同搜索多运动目标的方法和系统
Ke et al. Cooperative path planning for air–sea heterogeneous unmanned vehicles using search-and-tracking mission
Zhao et al. Design of leader's path following system for multi-vehicle autonomous convoy
Benkrid et al. Multi-robot coordination for energy-efficient exploration
CN114186859B (zh) 复杂未知环境多机协同多目标任务分配方法
Guan et al. Autonomous exploration using UWB and LiDAR
Fu et al. Obstacle avoidance and collision avoidance of UAV swarm based on improved VFH algorithm and information sharing strategy
Lee et al. Uwb based relative navigation and leader-follower formation for uavs using maneuvering of a follower
Hirsch et al. Decentralized cooperative urban tracking of multiple ground targets by a team of autonomous UAVs
Nelson et al. Initial experiments in cooperative control of unmanned air vehicles
Zhang et al. Cooperative Tactical Planning Method for UAV Formation
Lebedeva et al. Method for distributed mapping of terrain by a heterogeneous group of robots based on google cartographer
Zheng et al. The uav path planning method based on lidar
Savkin et al. Decentralized formation flocking and stabilization for networks of unicycles
Zehua MTSP UAV detection problem based on ant algorithm
Ziye et al. Multi-objective task assignment and autonomous approach research based on multiple unmanned vehicles

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20953963

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20953963

Country of ref document: EP

Kind code of ref document: A1