Moving target trajectory prediction method and related equipment

Abstract

The invention discloses a moving target trajectory prediction method and related equipment, and relates to the technical field of trajectory prediction.The method comprises the steps that multi-source dynamic sensing data of a moving target are obtained, and the multi-source dynamic sensing data comprise target motion state parameters, environment disturbance parameters and biological behavior characteristic parameters; generating an initial trajectory prediction result based on the target motion state parameters; dynamically correcting the initial trajectory prediction result based on the environmental disturbance parameters to obtain a disturbance compensation trajectory; and performing behavior pattern matching on the disturbance compensation trajectory based on the biological behavior characteristic parameters to generate a target trajectory prediction result. By fusing the multi-source dynamic sensing data and combining a dual optimization mechanism of environment disturbance dynamic correction and biological behavior pattern matching, the trajectory prediction precision in a complex scene is improved.

Description

Technical Field

[0001] This application relates to the technical field of trajectory prediction, and particularly to a method for predicting the trajectory of a moving target and related devices. Background Art

[0002] In the prior art, the method for predicting the trajectory of a moving target only relies on data of a single dimension. For example, a traditional model obtains the real-time position and speed of a moving target through a radar or a vision sensor, and calculates the trajectory based on a kinematic model. In addition, for a moving target with biological characteristics, the prior art is difficult to accurately predict sudden turning or speed changes due to ignoring biological behavior characteristics, especially the problem of error accumulation is prominent in a long-term tracking scenario. Therefore, there is an urgent need for a method for predicting the trajectory of a moving target to solve the above-mentioned technical problems. Summary of the Invention

[0003] A series of simplified concepts are introduced in the Summary of the Invention section, which will be further described in detail in the Detailed Description section. The Summary of the Invention section of this application does not mean to attempt to limit the key features and essential technical features of the claimed technical solution, nor does it mean to attempt to determine the protection scope of the claimed technical solution.

[0004] In a first aspect, this application provides a method for predicting the trajectory of a moving target, including: Obtaining multi-source dynamic perception data of a moving target, where the multi-source dynamic perception data includes target motion state parameters, environmental disturbance parameters, and biological behavior characteristic parameters; Generating an initial trajectory prediction result based on the target motion state parameters; Dynamically correcting the initial trajectory prediction result based on the environmental disturbance parameters to obtain a disturbance compensation trajectory; Performing behavior pattern matching on the disturbance compensation trajectory based on the biological behavior characteristic parameters to generate a target trajectory prediction result.

[0005] In some embodiments, obtaining the multi-source dynamic perception data of the moving target includes: Obtaining the real-time speed, acceleration, and motion direction of the moving target through a millimeter-wave radar as the target motion state parameters; Collecting the air flow speed, air pressure gradient, and terrain undulation around the moving target through a distributed air pressure sensor array as the environmental disturbance parameters; Extracting the body surface temperature distribution and limb swing frequency of the moving target through an infrared thermal imager as the biological behavior characteristic parameters.

[0006] In some embodiments, generating the initial trajectory prediction result based on the target motion state parameters includes: Calculate the instantaneous trajectory vector of a moving target based on real-time speed, acceleration, and direction of motion; Construct a target motion state matrix according to the instantaneous trajectory vector and a preset kinematic model; Input the target motion state matrix into a trajectory prediction neural network to generate an initial trajectory prediction result, where the trajectory prediction neural network is trained with historical motion data and is used to output a trajectory sequence matching the current motion state.

[0007] In some embodiments, dynamically correct the initial trajectory prediction result based on environmental perturbation parameters to obtain a perturbation compensation trajectory, including: Determine the fluid resistance coefficient of the moving target based on the air flow speed and direction of motion; Determine the terrain elevation change rate based on the terrain undulation degree; Determine a perturbation compensation model based on the terrain elevation change rate and the air pressure gradient; Dynamically correct the initial trajectory prediction result based on the fluid resistance coefficient and the perturbation compensation model to determine the perturbation compensation trajectory.

[0008] In some embodiments, perform behavior pattern matching on the perturbation compensation trajectory based on biological behavior characteristic parameters to generate a target trajectory prediction result, including: Determine the energy consumption rate of the moving target based on the body surface temperature distribution; Construct a motion rhythm map of the moving target based on the limb swing frequency; Determine the trajectory segment correction intensity based on the first comparison result between the energy consumption rate and a preset energy threshold; Match the reference trajectory segments in the historical behavior pattern library based on the motion rhythm map, where the similarity between the reference trajectory segments and the motion rhythm map is greater than a preset similarity threshold; Perform spatio-temporal superposition on the perturbation compensation trajectory based on the trajectory segment correction intensity and the reference trajectory segments to generate a target trajectory prediction result.

[0009] In some embodiments, it further includes: Generate a trajectory visualization map based on the target trajectory prediction result, where the trajectory visualization map includes dynamic risk area markings; Determine an obstacle avoidance instruction or a path optimization suggestion corresponding to the moving target based on the dynamic risk area markings.

[0010] In some embodiments, it further includes: Based on the target trajectory prediction result, real-time monitor the deviation degree between the actual motion path of the moving target and the target trajectory prediction result; When the deviation degree is greater than a first preset threshold, trigger a re-acquisition instruction for multi-source dynamic perception data; Based on the re-acquisition instruction, re-obtain multi-source dynamic perception data.

[0011] In a second aspect, the present application proposes a moving target trajectory prediction device, including: A multi-source data acquisition unit, configured to acquire multi-source dynamic perception data of a moving target, where the multi-source dynamic perception data includes target motion state parameters, environmental disturbance parameters, and biological behavior characteristic parameters; An initial trajectory generation unit, configured to generate an initial trajectory prediction result based on the target motion state parameters; A compensation trajectory determination unit, configured to dynamically correct the initial trajectory prediction result based on the environmental disturbance parameters to obtain a disturbance compensation trajectory; A target trajectory prediction unit, configured to perform behavior pattern matching on the disturbance compensation trajectory based on the biological behavior characteristic parameters to generate a target trajectory prediction result.

[0012] In a third aspect, an electronic device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program stored in the memory, it implements the steps of the moving target trajectory prediction method according to any one of the first aspects.

[0013] In a fourth aspect, the present application proposes a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the moving target trajectory prediction method according to any one of the first aspects.

[0014] In summary, the present application improves the trajectory prediction accuracy in complex scenarios by fusing multi-source dynamic perception data and combining a dual optimization mechanism of dynamic correction based on environmental disturbances and behavior pattern matching of biological behaviors. Specifically, the compensation model constructed based on environmental disturbance parameters can real-time offset external interferences such as air resistance and terrain undulations, and the introduction of biological behavior characteristics can simulate the physiological constraints and behavior preferences of the target, so as to achieve a trajectory prediction that more conforms to the actual motion law in the fields of UAV navigation, intelligent security, etc., effectively reducing the risk of path deviation and enhancing the adaptive ability of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] By reading the detailed description of the preferred embodiments below, various other advantages and benefits will become clear to those of ordinary skill in the art. The drawings are only for the purpose of showing the preferred embodiments and are not considered to limit this specification. Moreover, throughout the drawings, the same reference numerals are used to represent the same components. In the drawings: Figure 1 is a schematic flowchart of a moving target trajectory prediction method provided by an embodiment of the present application; Figure 2 is a schematic structural diagram of a moving target trajectory prediction device provided by an embodiment of the present application; Figure 3 This is a structural diagram of an electronic device for predicting the trajectory of a moving target provided by an embodiment of the present application. Detailed implementation manners

[0016] The terms "first", "second", "third", "fourth", etc. (if any) in the specification, claims and drawings of the present application are used to distinguish similar objects, and do not necessarily describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances so that the embodiments described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or units does not necessarily limit to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices. The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments.

[0017] Please refer to Figure 1 , which is a schematic flowchart of a method for predicting the trajectory of a moving target provided by an embodiment of the present application, and specifically may include: S110. Obtain multi-source dynamic perception data of the moving target, where the multi-source dynamic perception data includes target motion state parameters, environmental disturbance parameters, and biological behavior characteristic parameters; Exemplarily, obtaining multi-source dynamic perception data of the moving target is the core basis of trajectory prediction. By fusing target motion state parameters, environmental disturbance parameters, and biological behavior characteristic parameters, it realizes the all-round perception of the target motion characteristics. Among them, the target motion state parameters (such as speed, acceleration, direction) directly reflect the instantaneous motion law of the target; the environmental disturbance parameters (such as air flow speed, terrain undulation) depict the dynamic interference of the external environment on the target trajectory; and the biological behavior characteristic parameters (such as body surface temperature distribution, limb swing frequency) are used to reveal the behavior pattern of the target under physiological constraints, so as to provide multi-dimensional data support for prediction.

[0018] The synergistic effect of multi-source dynamic perception data breaks through the limitations of single-dimensional data. Through the combination of millimeter-wave radar, distributed air pressure sensor arrays, and infrared thermal imagers, the system can synchronously capture the physical motion, environmental disturbance, and biological behavior characteristics of the target, forming a multi-dimensional data fusion framework. This comprehensive perception mechanism not only provides high-precision input for subsequent initial trajectory prediction, but also lays a data foundation for environmental disturbance compensation and biological behavior pattern matching, thus improving the robustness and adaptability of trajectory prediction in complex scenarios.

[0019] S120. Generate an initial trajectory prediction result based on the target motion state parameters; Exemplarily, generating an initial trajectory prediction result based on the target motion state parameters aims to construct its basic motion model through physical quantities such as the real-time speed, acceleration, and motion direction of the target. These parameters directly reflect the instantaneous motion law of the target. Through a preset kinematic model or neural network algorithm, they are mapped into a trajectory sequence for a future period of time to form a preliminary prediction framework. The initial prediction result serves as a benchmark for subsequent correction and optimization, providing the core input for environmental disturbance compensation and behavior pattern matching.

[0020] The key to initial trajectory prediction lies in converting discrete motion state parameters into a mathematical representation of a continuous trajectory. By integrating the dynamic motion characteristics of the target (such as the direction of acceleration change), its short-term motion trend can be deduced, and the neural network trained with historical data is used to capture the potential correlations in complex motion patterns. This process not only establishes a physical logic basis for trajectory prediction but also reserves an interface for the dynamic correction of subsequent multi-source data fusion, ensuring the adjustability and expandability of the prediction result in complex scenarios.

[0021] S130. Dynamically correct the initial trajectory prediction result based on the environmental disturbance parameters to obtain a disturbance compensation trajectory; Exemplarily, dynamically correcting the initial trajectory prediction result based on the environmental disturbance parameters aims to construct a disturbance compensation model by quantifying the influence of external environmental factors such as air flow velocity, terrain undulation degree, and air pressure gradient on the target motion. This model adjusts the curvature and speed components of the initial predicted trajectory in real time by calculating key parameters such as the fluid resistance coefficient and terrain elevation change rate, thereby offsetting the trajectory deviation caused by environmental disturbances and enhancing the physical rationality of the prediction result.

[0022] The core of the dynamic correction process lies in the coupled analysis of environmental parameters and the target motion state. By quantifying dynamic interferences such as air flow resistance and terrain changes into computable compensation factors, the idealized assumptions in the initial trajectory prediction are iteratively optimized to generate a disturbance compensation trajectory that better fits the actual environmental conditions. This mechanism effectively solves the problem of cumulative prediction errors caused by traditional methods ignoring external disturbances and provides a trajectory input with environmental adaptability for subsequent behavior pattern matching.

[0023] S140. Perform behavior pattern matching on the disturbance compensation trajectory based on the biological behavior characteristic parameters to generate a target trajectory prediction result.

[0024] Exemplarily, behavior pattern matching is performed on the disturbance compensation trajectory based on biobehavioral characteristic parameters, aiming to analyze the motion preference and behavior pattern of the target through its physiological characteristics (such as energy consumption rate, limb swing frequency). By comparing the body surface temperature distribution with a preset energy threshold, the energy reserve state of the target is inferred, so as to predict the persistence or turning tendency of its subsequent motion; at the same time, the motion rhythm map constructed based on the limb swing frequency can map the periodic behavior pattern of the target, providing a physiological constraint basis for trajectory correction.

[0025] The core of behavior pattern matching lies in associating biometric features with historical behavior data to achieve personalized optimization of trajectory prediction. By matching the motion rhythm map with high-similarity reference trajectory segments in the historical behavior pattern library, a trajectory correction strategy that conforms to the target's physiological characteristics is extracted, and combined with the correction intensity determined by the energy consumption rate, spatio-temporal superposition is performed on the disturbance compensation trajectory. The finally generated trajectory prediction result not only retains the physical rationality of environmental disturbance correction but also incorporates the dynamic adaptability driven by biobehavior, improving the prediction accuracy in complex scenarios.

[0026] In summary, in the embodiments of the present application, by integrating the target motion state parameters, environmental disturbance parameters, and biobehavioral characteristic parameters, a multi-dimensional data-driven trajectory prediction framework is constructed, improving the prediction accuracy and adaptability in complex scenarios. Dynamically correcting the initial trajectory prediction result based on environmental disturbance parameters effectively offsets the influence of external disturbances such as air resistance and terrain undulation on the target motion, enhancing the robustness of the physical model; combining biobehavioral characteristic parameters to perform behavior pattern matching on the compensated trajectory, integrating physiological constraints such as energy consumption rate and motion rhythm map into the prediction logic, realizing dual optimization from physical motion laws to biobehavioral preferences. The embodiments of the present application break through the limitations of traditional single data sources. In the fields of UAV autonomous navigation, intelligent security tracking, etc., it can accurately predict sudden turning or speed changes, reduce the risk of path deviation, and at the same time, through dynamic risk area marking and closed-loop feedback mechanisms, provide high-confidence data support for real-time obstacle avoidance decision-making and path planning, with broad application value and technological innovation.

[0027] In some instances, multi-source dynamic perception data of a moving target is obtained, including: The real-time speed, acceleration, and motion direction of the moving target are obtained through a millimeter-wave radar as target motion state parameters; The air flow velocity, air pressure gradient, and terrain undulation around the moving target are collected through a distributed air pressure sensor array as environmental disturbance parameters; The body surface temperature distribution and limb swing frequency of the moving target are extracted through an infrared thermal imager as biobehavioral characteristic parameters.

[0028] Exemplarily, a millimeter-wave radar determines the real-time speed, acceleration, and motion direction of a target by transmitting high-frequency electromagnetic waves in a specific frequency band and receiving the target reflection signal. For the measurement of real-time speed, the radar directly calculates the speed component of the target in the radial direction of the radar by analyzing the Doppler frequency shift amount of the reflection signal and combining the frequency difference between the transmitted wave and the received wave. To obtain the motion direction in three-dimensional space, the radar uses a multi-antenna array to receive signals, calculates the azimuth angle and elevation angle of the target relative to the radar through the angle-of-arrival estimation technology, and combines the coordinate transformation algorithm to decompose the radial speed into a speed vector in three-dimensional space, thereby determining the motion direction of the target. The determination of acceleration is achieved through the calculation of speed difference within a continuous time window, that is, capturing the instantaneous speed sequence of the target within a preset sampling period, and using the least squares method or the Kalman filter algorithm to fit the speed change rate, and outputting a smooth acceleration value after eliminating the measurement noise interference. The above parameters rely on the timing synchronization and spatial calibration of the radar signal processing link to ensure the consistency of the target motion state parameters in the time and space dimensions, providing input for subsequent trajectory prediction.

[0029] The distributed barometric pressure sensor array synchronously collects the barometric pressure data around the target through multiple barometric pressure sensor nodes distributed according to a preset topology, and determines the air flow velocity by combining the principles of fluid mechanics and spatial interpolation algorithms. Specifically, each node measures the barometric pressure value within the same time window, calculates the local pressure gradient through the barometric pressure difference between adjacent nodes, and based on Bernoulli's equation, converts the pressure gradient into an air flow velocity vector. For the determination of the air flow direction, the array reconstructs the motion trend of the air flow in three-dimensional space by analyzing the temporal correlation of barometric pressure changes among multiple nodes and combining the spatial position relationship. The determination of the barometric pressure gradient depends on the spatial distribution density and measurement accuracy of the array nodes. Using the finite difference method, the discrete barometric pressure measurement values are fitted into a continuous barometric pressure field, and then the barometric pressure change rate per unit distance, that is, the barometric pressure gradient, is calculated. The measurement of the terrain undulation degree is achieved by integrating a laser rangefinder module or an inertial measurement unit. The laser rangefinder emits pulsed laser to the ground and receives the reflection signal, calculates the vertical distance between the sensor node and the ground through the time difference of flight, and combines the relative position data between nodes to construct a terrain elevation model; the inertial measurement unit detects the attitude change of the sensor itself through an accelerometer and a gyroscope, and deduces the terrain undulation degree in combination with the initial elevation reference. The determination of the above parameters is based on the time and space synchronization calibration of the sensor array and the multi-source data fusion algorithm to ensure that the measurement results of the air flow velocity, barometric pressure gradient, and terrain undulation degree have high spatial resolution and physical consistency.

[0030] The infrared thermal imager generates a thermal imaging image that reflects the temperature distribution by receiving the infrared energy radiated from the target body surface. The determination of the body surface temperature distribution is based on the radiation intensity data captured by the infrared sensor array of the thermal imager. The radiation intensity of each pixel is converted into a corresponding temperature value by the blackbody radiation law to form a two-dimensional temperature matrix. The discrete temperature points are smoothed by a spatial interpolation algorithm, and a continuous temperature distribution map is output after eliminating environmental noise and measurement errors. The preset physiological model further associates the temperature distribution with the target metabolic rate. For example, high-temperature areas correspond to areas with high metabolic activity, and low-temperature areas represent areas with low energy consumption, thereby quantifying the spatial heterogeneity of body surface temperature.

[0031] The determination of limb swing frequency is achieved through the analysis of continuous frame thermal imaging sequences. Within the preset sampling period, the thermal imager captures the dynamic thermal image sequence of the target limb, and locates the key temperature feature area of ​​the joint or limb end through the feature point recognition algorithm. Based on the displacement vector of the feature area between adjacent frames, the swing amplitude and direction change of the limb in unit time are calculated, and the main frequency component of the periodic motion signal is extracted using Fourier transform or autocorrelation algorithm. By counting the number of swing cycles within the preset time window, the limb swing frequency is finally determined to characterize the target's motion rhythm characteristics. The analysis of the above parameters is based on the time resolution and spatial calibration accuracy of the thermal imager to ensure the temporal and spatial consistency of the temperature distribution and swing frequency data.

[0032] The preset physiological model is an empirical or theoretical model based on the known relationship between the physiological characteristics of organisms and the distribution of body surface temperature. It is used to map the body surface temperature data obtained by the thermal imager to the energy metabolism state. The model is set to show that the body surface temperature is positively correlated with the internal metabolic activity, blood circulation intensity and tissue heat production efficiency. For example, high temperature areas may correspond to areas with frequent muscle activity or high energy consumption (such as limb joints during exercise), while low temperature areas may represent static tissues or energy reserve areas. Through the correspondence between the temperature distribution and the measured value of the metabolic rate under a specific motion state, the model establishes a quantitative mapping rule between the temperature gradient and the energy consumption rate, thereby supporting the inference of the target's real-time energy metabolism level from non-invasive temperature measurement and providing physiological constraint parameters for trajectory prediction.

[0033] In some examples, generating an initial trajectory prediction result based on the target motion state parameters includes: Calculate the instantaneous trajectory vector of the moving target based on real-time speed, acceleration and direction of movement; Construct the target motion state matrix according to the instantaneous trajectory vector and the preset kinematic model; The target motion state matrix is ​​input into the trajectory prediction neural network to generate an initial trajectory prediction result, wherein the trajectory prediction neural network is trained through historical motion data and is used to output a trajectory sequence that matches the current motion state.

[0034] Exemplarily, when determining the instantaneous trajectory vector of a moving target based on real-time speed, acceleration, and direction of motion, the velocity vector and acceleration vector are first decomposed and synthesized in a three-dimensional Cartesian coordinate system. The vector direction of the real-time speed is determined by the azimuth angle and elevation angle measured by the millimeter-wave radar, and it is converted into speed components in three-dimensional space through trigonometric functions. The vector direction of the acceleration is derived based on the instantaneous change rate of the velocity vector, and the acceleration components are calculated through the velocity difference within a continuous time window. The speed components and acceleration components are superimposed at time steps to generate an instantaneous trajectory vector, which is specifically represented as a linear combination of the velocity vector and acceleration vector within a time increment. This vector maps the velocity and acceleration at the current moment to the displacement increment at the next moment through integral operations or kinematic equations, thereby characterizing the instantaneous motion trend of the target in three-dimensional space. The instantaneous trajectory vector characterizes the instantaneous motion trend of the target at the current moment, including the combined effects of linear velocity, angular velocity, and acceleration, providing basic data for subsequent construction of the motion state matrix.

[0035] When constructing the target motion state matrix, the preset kinematic model defines the motion constraint conditions of the target based on Newtonian mechanics or rigid body kinematics principles. For example, the uniform linear motion model assumes a constant target speed, the uniformly accelerated motion model introduces an acceleration vector, and the curvilinear motion model combines the angular velocity and curvature parameters. By mapping the real-time speed, acceleration, and direction of motion to the model parameter space, a multi-dimensional state matrix is generated. Specifically, the speed components are decomposed into the speed in the X-axis direction, the speed in the Y-axis direction, and the speed in the Z-axis direction in a three-dimensional coordinate system; the acceleration components correspond to the acceleration in the X-axis direction, the acceleration in the Y-axis direction, and the acceleration in the Z-axis direction; the direction angle is characterized by the azimuth angle and elevation angle, and combined with model parameters (such as mass, moment of inertia), a matrix is constructed according to the time series. Each row of the matrix corresponds to a sampling time point, and the column vectors sequentially include the position coordinates, speed components, acceleration components, direction angle, and model parameters of the target in three-dimensional space, forming a structured data set describing the motion characteristics of the target.

[0036] The determination of the initial trajectory prediction result is achieved by inputting the target motion state matrix into a trajectory prediction neural network, which is a deep learning model based on a Long Short-Term Memory network (LSTM) or a Convolutional Neural Network (CNN). The trajectory prediction neural network is trained with historical motion data, including historical motion state matrices and corresponding true trajectory sequences. The motion state matrix consists of parameters such as velocity, acceleration, and direction, and the trajectory sequence is a series of position coordinate points in time series form. During the training process, the network optimizes the weight parameters through the backpropagation algorithm to learn the non-linear mapping relationship between motion state parameters and future trajectories. The loss function uses the Mean Squared Error (MSE) or Dynamic Time Warping (DTW) to minimize the deviation between the predicted trajectory and the actual trajectory. The model hyperparameters (such as the number of network layers, the number of neurons, and the learning rate) are determined through cross-validation to ensure a balance between the generalization ability and prediction accuracy of the model. In the prediction stage, the current target motion state matrix is used as the input, and after forward propagation calculation by the neural network, a sequence of trajectory point coordinates within a preset future time window is output to form the initial trajectory prediction result. This result includes the predicted positions and velocity distributions of the target at consecutive time steps, serving as the reference trajectory for subsequent correction and optimization.

[0037] In some instances, based on environmental perturbation parameters, the initial trajectory prediction result is dynamically corrected to obtain a perturbation compensation trajectory, including: Based on the air flow velocity and the motion direction, determine the fluid resistance coefficient of the moving target; Based on the terrain undulation degree, determine the terrain elevation change rate; Based on the terrain elevation change rate and the air pressure gradient, determine the perturbation compensation model; Based on the fluid resistance coefficient and the perturbation compensation model, dynamically correct the initial trajectory prediction result to determine the perturbation compensation trajectory.

[0038] Exemplarily, the determination of the fluid resistance coefficient is based on the vector relationship between the air flow velocity and the target motion direction. The air flow velocity is collected by a distributed air pressure sensor array, and the three-dimensional air flow velocity field is reconstructed through air pressure gradient analysis and spatial interpolation algorithms to obtain the air flow velocity vector around the target; the target motion direction is determined by the direction of the real-time velocity vector measured by a millimeter-wave radar. The calculation of the fluid resistance coefficient first calculates the angle between the air flow velocity vector and the target motion direction vector through the vector dot product formula, specifically the arccosine value of the dot product of the two vectors divided by the product of their moduli, expressed as: Where, is the angle; is the air flow velocity vector; is the target motion direction vector.

[0039] Based on this angle, combined with the Reynolds number and the target geometric feature parameters, determine the fluid resistance coefficient through a preset empirical model or a look-up table method , expressed as:

[0040] Wherein, is the Reynolds number, which is calculated from the air density , air flow velocity , target characteristic length (such as the diameter of the windward side or the equivalent length) and the air viscosity coefficient . The target characteristic length is obtained by matching the target type through a pre-set database, such as the standard geometric parameters of an unmanned aerial vehicle, a vehicle or an organism; the target geometric characteristics include the windward area, surface roughness and shape factor, which are obtained by matching the target type through a pre-set database. For example, the windward area of a spherical target is , while the windward area of a streamlined object needs to be calculated based on a 3D model; the pre-set empirical model correlates the above parameters to the fluid drag coefficient. For example, for a target with a specific shape, the variation law of the fluid drag coefficient with the Reynolds number and the included angle can be fitted as a piecewise function; the look-up table method is implemented through a pre-generated multi-dimensional look-up table. The indexes in the table are the Reynolds number interval, the included angle interval and the target type, and the values measured experimentally are stored correspondingly. According to the Reynolds number, included angle and target type calculated in real time, the accurate value is obtained through interpolation or nearest neighbor matching. The fluid drag coefficient comprehensively characterizes the hindering effect of the air flow direction, flow state and target geometric characteristics on the movement, and provides a quantitative drag parameter input for the subsequent disturbance compensation model.

[0041] When determining the terrain elevation change rate based on the terrain undulation degree, first, discrete elevation data on the target movement path is obtained through a laser rangefinder module or an inertial measurement unit. The laser rangefinder emits pulsed laser to the ground and receives the reflected signal, calculates the vertical distance between the sensor node and the ground, and combines the relative position coordinates between the nodes to generate a discrete elevation point set; the inertial measurement unit detects the attitude change of the sensor itself through an accelerometer and a gyroscope, and combines the initial elevation reference to deduce a continuous elevation change sequence. The above discrete elevation data is constructed into a 3D point cloud according to the spatial coordinates, and a digital elevation model (DEM) is generated through the Delaunay triangulation algorithm to form a continuous terrain surface representation.

[0042] The calculation of the terrain elevation change rate is based on the elevation gradient in the target movement direction. Along the arc length sequence and the corresponding elevation values of the target movement path, the elevation difference and arc length difference between adjacent sampling points are calculated through the sliding window method, expressed as: Wherein, is the elevation value of the th sampling point; is the arc length from the starting point to the The cumulative arc length of each sampling point. To eliminate the interference of measurement noise, the Kalman filter algorithm is used to smooth the original differential results and output a stable sequence of elevation change rates. This parameter quantifies the dynamic change of the terrain slope along the target movement direction, directly reflecting the real-time impact of terrain undulation on the target movement energy consumption and trajectory curvature, and providing an accurate terrain interference quantization input for subsequent disturbance compensation.

[0043] When determining the disturbance compensation model based on the terrain elevation change rate and the air pressure gradient, the model construction is achieved through the linear superposition of the terrain disturbance component and the air pressure disturbance component. The calculation of the terrain disturbance component is based on the terrain elevation change rate, which combines the discrete elevation data obtained by the laser ranging module or the inertial measurement unit, and uses the digital elevation model and the sliding window differential algorithm to calculate the elevation gradient along the target movement direction. . The terrain slope angle is determined by and combined with the target mass and the acceleration due to gravity . The terrain resistance component is calculated through the formula , representing the impact of terrain undulation on the energy consumption of target movement. The calculation of the air pressure disturbance component is based on the spatial interpolation result of the distributed air pressure sensor array to obtain the air pressure gradient , whose direction is consistent with the airflow velocity vector. The air pressure gradient force is calculated by the formula , where is the effective force area of the target, which is obtained by matching the preset database or real-time measurement. Finally, the disturbance compensation model superimposes the terrain resistance component and the air pressure gradient force according to the direction to generate a comprehensive disturbance compensation vector , expressed as: where, is the unit vector in the slope direction; is the unit vector in the airflow direction; this model provides a physical compensation basis for trajectory correction by quantifying the dynamic interference of terrain and air pressure on target movement.

[0044] When dynamically correcting the initial trajectory prediction result based on the fluid resistance coefficient and the disturbance compensation model, first substitute the fluid resistance coefficient into the air resistance formula to calculate the airflow resistance , expressed as: where, is the air density; this resistance represents the dynamic hindrance effect of the airflow on target movement.

[0045] Combined with the comprehensive disturbance compensation vector output by the disturbance compensation model, calculate the total disturbing force , expressed as: Among them, It is generated by the linear superposition of the terrain resistance component and the pressure gradient force, corresponding to the influence of the terrain elevation change rate and the pressure gradient respectively.

[0046] Input the total disturbing force into the kinematic equation, and iteratively correct the velocity component and acceleration component of the initial trajectory prediction result. At the discrete time step within, the corrected velocity and position are expressed as: Among them, is the acceleration of the initial prediction; by updating the velocity and position parameters step by step in time, a disturbance compensation trajectory is generated. This trajectory eliminates the error generated by the initial prediction due to ignoring external disturbances through the comprehensive disturbance compensation of dynamically superposing air flow resistance, terrain slope and pressure gradient, thereby improving the physical rationality and prediction accuracy of the trajectory in a complex environment.

[0047] In some examples, based on the biological behavior characteristic parameters, the behavior pattern of the disturbance compensation trajectory is matched to generate the target trajectory prediction result, including: Determine the energy consumption rate of the moving target based on the body surface temperature distribution; Construct the motion rhythm map of the moving target based on the limb swing frequency; Determine the trajectory segmentation correction intensity based on the first comparison result between the energy consumption rate and the preset energy threshold; Match the reference trajectory segment in the historical behavior pattern library based on the motion rhythm map, where the similarity between the reference trajectory segment and the motion rhythm map is greater than the preset similarity threshold; Based on the trajectory segmentation correction intensity and the reference trajectory segment, perform spatio-temporal superposition on the disturbance compensation trajectory to generate the target trajectory prediction result.

[0048] Exemplarily, when determining the energy consumption rate of a moving target based on the body surface temperature distribution, an infrared thermal imager captures the radiation intensity data of the target's body surface through its infrared sensor array. The radiation intensity of each pixel is converted into a corresponding temperature value according to the blackbody radiation law, forming a two-dimensional temperature matrix. The preset physiological model is constructed based on the experimental calibration data of the organism temperature and metabolic rate. The local temperature value is converted into the energy consumption rate through the mapping relationship between the temperature gradient and the metabolic rate. Specifically, the model corresponds the high-temperature regions (such as the limb joints during movement) to the high-metabolic activity regions, and the low-temperature regions (such as static tissues) to the low-energy consumption regions, and establishes a linear or non-linear functional relationship between the temperature gradient and the local metabolic rate based on the spatial differences in the temperature distribution. After the temperature value of each pixel is mapped to the local energy consumption rate through this functional relationship, a weighted average algorithm is used to integrate and calculate the metabolic rate of the entire temperature region of the image, and finally the overall energy consumption rate of the target is output. This parameter provides a physiological constraint basis for subsequent trajectory correction by quantifying the energy consumption intensity of the target in the current motion state.

[0049] When constructing the motion rhythm map of a moving target based on the limb swing frequency, the periodic motion characteristics of the target limb are analyzed through a continuous frame thermal imaging sequence. The infrared thermal imager captures a dynamic thermal image sequence of the target limb at a preset sampling frequency, and locates the high-temperature regions at the joints or the ends of the limbs through a feature point recognition algorithm. The determination of the high-temperature regions is based on the thermal imaging pixel analysis of the body surface temperature distribution, and the feature points with temperatures significantly higher than the surrounding tissues are selected by combining a preset temperature threshold, generating an initial feature point set containing spatial coordinates.

[0050] Based on the initial feature point set, the displacement vectors between adjacent frames are calculated through the optical flow method or the template matching algorithm. For each feature point, the pixel coordinate change amount in the continuous frames is extracted, and combined with the spatial resolution and time interval of the thermal imager, the pixel displacement is converted into an actual physical displacement vector. The direction and amplitude of the displacement vector respectively represent the direction change and motion intensity of the limb swing, generating a displacement vector data set containing a time series.

[0051] Based on the displacement vector data set, the main frequency component of the displacement sequence is extracted using the fast Fourier transform (FFT). The displacement amplitude sequence in the time domain is converted into a frequency domain signal, and the frequency value corresponding to the main frequency peak is determined through spectrum analysis. The extraction of the main frequency needs to meet the preset signal-to-noise ratio threshold to exclude the false peaks caused by noise interference, and the main frequency value and the corresponding spectrum energy distribution are output.

[0052] Based on the main frequency value and the displacement sequence, the periodicity is verified through the autocorrelation algorithm. The autocorrelation calculation is performed on the displacement amplitude sequence to generate an autocorrelation function curve. The determination of periodicity is based on the uniformity of the peak intervals and the attenuation rate of the peak amplitudes in the curve. If the autocorrelation curve exhibits periodic peaks and the attenuation rate is lower than the preset threshold, it is determined that the displacement sequence has periodicity, and the periodicity verification flag and the peak interval time parameter are output.

[0053] Based on the main frequency value, the periodicity verification result, and the displacement vector data, a motion rhythm map is generated. The map is stored in the form of a time and frequency matrix. The rows of the matrix correspond to the time series, and the columns include the main frequency value, the mean displacement amplitude, and the phase angle parameter. The phase angle is calculated from the initial phase offset of the displacement sequence, and the mean displacement amplitude is obtained by smoothing the norm of the displacement vector through a sliding window method. The construction of the map fills the data gaps through an interpolation algorithm to ensure the continuity in the time dimension, and a standardized motion rhythm map is output.

[0054] The finally generated motion rhythm map fully characterizes the periodic law of the target limb swing, including frequency stability, amplitude fluctuation range, and phase synchronization. The storage format of the map is a structured multi-dimensional array, which supports efficient query and matching operations and provides standardized input data for subsequent behavior pattern matching.

[0055] When determining the trajectory segmentation correction intensity based on the first comparison result between the energy consumption rate and the preset energy threshold, the preset energy threshold is set through the statistics of historical motion data, and different energy threshold intervals are specifically set for different biological types (such as humans, animals). For example, the energy threshold for the jogging state of humans is set to 5 kcal / min, while that for the running state of dogs is set to 3 kcal / min. The real-time energy consumption rate is obtained by inverse deduction from the body surface temperature distribution through a preset physiological model and is compared with the corresponding threshold in real time. When the real-time rate exceeds the threshold, it is determined that the target is in a high energy consumption state, and the motion intensity needs to be reduced or the path needs to be adjusted, indicating that the greater the energy consumption of the target, the more it needs to rely on historical behavior patterns; if it does not exceed, the current motion mode is maintained. The correction intensity α is quantified according to the percentage exceeding the threshold and is expressed as: α = 1 - [(real-time energy consumption rate - preset energy threshold) / preset energy threshold] Among them, if the real-time energy consumption rate is 120% of the threshold, it is linearly mapped to an intensity coefficient of 0.8 (range 0 - 1), indicating a low correction intensity, and the main body of the compensation trajectory is retained; if it exceeds 50%, it is mapped to 0.5; if it exceeds 100%, then α corresponds to 0, indicating a high correction intensity, and the historical trajectory is fully referenced. The trajectory segmentation correction intensity directly determines the intervention degree of subsequent trajectory correction, ensuring the physical rationality of the predicted trajectory under energy constraints and avoiding trajectory deviation from the actual motion ability due to excessive correction.

[0056] When determining the reference trajectory segment, the historical motion data stored in the historical behavior pattern library includes the motion rhythm map and its corresponding trajectory sequence. The motion rhythm map records the periodic motion characteristics of the target in the form of a time and frequency matrix, including limb swing frequency, amplitude, and phase information. During the matching process, the dynamic time warping (DTW) algorithm is used to calculate the time series alignment error between the current motion rhythm map and the historical rhythm maps in the library, which is specifically achieved by minimizing the cumulative distance between the two sequences on the time axis. A preset similarity threshold (e.g., 0.8) is set as the screening condition, and only the candidate trajectory segments with alignment errors lower than the threshold are retained.

[0057] To improve the matching efficiency, an index structure is established for the frequency, amplitude, and phase characteristics of the historical rhythm maps through multi-dimensional indexing techniques (such as R-trees or hash tables), and the entries that are closest to the current map in the key dimensions are retrieved preferentially. Finally, through similarity sorting and threshold filtering, the reference trajectory segment with the highest matching degree to the current motion rhythm map is determined to ensure the rationality of the selected segment in terms of motion rhythm characteristics.

[0058] Based on the trajectory segment correction intensity and the reference trajectory segment for the perturbation compensation trajectory When performing spatio-temporal superposition, a weighted interpolation algorithm is used to fuse the compensation trajectory and the reference trajectory to generate the target trajectory prediction result. Specifically, the trajectory segment correction intensity is determined by the comparison result between the energy consumption rate and the preset energy threshold, and its value range is from 0 to 1. When the trajectory segment correction intensity approaches 0, it indicates that the current trajectory segment needs to significantly refer to the historical behavior pattern. At this time, the weight coefficient of the reference trajectory segment is , which dominates the trajectory superposition; when the trajectory segment correction intensity approaches 1, it indicates that the current trajectory segment is more dependent on the environmental perturbation compensation result, and the weight coefficient of the perturbation compensation trajectory is , and its main structure is retained.

[0059] During the spatio-temporal superposition process, the position, velocity, and acceleration parameters of each trajectory point are expressed as: where, is the perturbation compensation trajectory generated by dynamically correcting the initial trajectory through the perturbation compensation model, which includes the compensation results of environmental disturbances such as air resistance, terrain undulation, and air pressure gradient; is the reference trajectory segment in the historical behavior pattern library with a similarity greater than the preset threshold to the current motion rhythm map; is the time step corresponding to the trajectory segment correction intensity, which is dynamically adjusted by the real-time comparison result between the energy consumption rate and the preset energy threshold. For a high correction intensity For the trajectory segment, the curvature and velocity parameters of the reference trajectory segment cover and compensate the corresponding parameters of the trajectory through an interpolation algorithm, such as directly replacing the radius of curvature or adjusting the velocity direction; for the trajectory segment with low correction intensity the physical rationality of the compensated trajectory is retained, and only local path points are corrected for smooth transition.

[0060] The superimposed trajectory point sequence is globally smoothed through a cubic spline interpolation or Kalman filtering algorithm to eliminate the problems of sudden curvature changes or velocity discontinuities caused by segmental correction. The finally generated target trajectory prediction result is a time-series set of three-dimensional coordinates, containing position, velocity, and acceleration information at each time step. This result not only integrates the physical motion laws of environmental disturbance compensation but also embeds the physiological constraints of historical behavior patterns, achieving a dual optimization from physical dynamics to biological behavior, and improving the trajectory prediction accuracy and adaptability in complex scenarios. In UAV navigation, it is necessary to adjust the flight path in real time to ensure that at each time step avoid dynamic risk areas (such as sudden airflow change areas or terrain obstacles).

[0061] In some instances, it also includes: Based on the target trajectory prediction result, a trajectory visualization map is generated, where the trajectory visualization map contains dynamic risk area markings; Based on the dynamic risk area markings, obstacle avoidance instructions or path optimization suggestions corresponding to the moving target are determined.

[0062] Exemplarily, when generating a trajectory visualization map based on the target trajectory prediction result, the three-dimensional coordinate sequence of the predicted trajectory is spatially mapped with the dynamic risk area parameters. The dynamic risk area markings are determined through real-time analysis of environmental disturbance parameters (such as airflow velocity, terrain undulation, and air pressure gradient) and biological behavior characteristic parameters (such as energy consumption rate). Specifically, the area where the airflow velocity exceeds the preset safety threshold is marked as the "high airflow risk area", the area where the terrain undulation causes the slope angle to be greater than the preset angle is marked as the "terrain obstacle area", and the trajectory segment where the energy consumption rate exceeds the physiological tolerance threshold is marked as the "high energy consumption risk segment". The risk level is rendered in a three-dimensional geographic information system through color coding (e.g., red represents high risk, yellow represents medium risk, and green represents the safe area), forming a visualization map superimposed on the predicted trajectory. The construction of the map relies on a multi-source data fusion algorithm to spatially register the data collected by millimeter-wave radars, distributed barometric sensor arrays, and infrared thermal imagers with the DEM to ensure that the risk markings are strictly aligned with the true geographic coordinates, providing intuitive visual support for path decision-making.

[0063] When determining obstacle avoidance instructions or path optimization suggestions based on dynamic risk area marking, the path of the high-risk area is replanned through a trajectory optimization algorithm. For the "high air flow risk area", the obstacle avoidance instruction generation module calculates a detour path based on the air flow direction and intensity, preferentially selects adjacent areas where the air flow speed is lower than the safety threshold, and verifies the feasibility of the detour path through a kinematic model; for the "terrain obstacle area", the path optimization suggestion module combines the terrain elevation change rate and the target's movement capabilities (such as the maximum climbing angle, minimum turning radius) to generate a circuitous path adapted to the slope; for the "high energy consumption risk section", based on the prediction of the energy consumption rate and the remaining energy, deceleration or pause strategies are recommended to reduce the metabolic load. All instructions and suggestions are comprehensively evaluated through a decision tree algorithm to ensure that the obstacle avoidance strategy simultaneously meets environmental safety, movement efficiency, and physiological constraints, and finally outputs a structured control instruction containing coordinate offsets, speed adjustment values, and direction correction angles to drive the movement target to execute trajectory adjustment in real time.

[0064] In some examples, it also includes: Based on the target trajectory prediction result, the deviation degree between the actual movement path of the movement target and the target trajectory prediction result is monitored in real time; When the deviation degree is greater than the first preset threshold, a re-acquisition instruction for multi-source dynamic perception data is triggered; Based on the re-acquisition instruction, multi-source dynamic perception data is re-obtained.

[0065] Exemplarily, when monitoring the deviation degree between the actual movement path of the movement target and the target trajectory prediction result in real time, the actual movement coordinates of the target are continuously collected through a positioning device (such as a GPS or a visual positioning system) and aligned in time and space with the coordinate sequence of the predicted trajectory. The calculation of the deviation degree uses the dynamic time warping algorithm, and the coordinate differences between the actual path and the predicted path are compared point by point to cumulatively calculate the overall deviation value. The specific process is as follows: First, the timestamps of the actual path and the predicted path are aligned through an interpolation algorithm to ensure that there are corresponding actual and predicted coordinates at each time point; then the Euclidean distance between each aligned point is calculated, and all distances are accumulated and normalized to obtain the comprehensive deviation degree. This deviation degree reflects the consistency between the actual movement and the predicted trajectory, and the larger the value, the more significant the deviation.

[0066] When the deviation exceeds the first preset threshold, a re - acquisition instruction for multi - source dynamic perception data is triggered. The first preset threshold is set through historical data analysis according to the application scenario. For example, it is set to 2 meters in the UAV navigation scenario and 5 meters in the wildlife tracking scenario. The triggering logic is as follows: The deviation monitored in real - time is compared with the preset threshold. If it exceeds the threshold for multiple consecutive time steps (such as 3 sampling periods), it is determined as a persistent deviation, and the data re - acquisition process is immediately started. The triggering instruction is synchronously sent to the millimeter - wave radar, the distributed barometric pressure sensor array, and the infrared thermal imager, forcing these sensors to re - acquire the target motion state parameters (speed, acceleration, direction), environmental disturbance parameters (airflow speed, barometric pressure gradient, terrain undulation), and biological behavior characteristic parameters (body surface temperature distribution, limb swing frequency) within a preset time window (such as 5 seconds before and after the current moment) to ensure coverage of the time interval of environmental mutation or abnormal target behavior.

[0067] After re - obtaining the multi - source dynamic perception data, data update and trajectory correction are performed, including: The millimeter - wave radar, by analyzing the Doppler shift and angle of arrival of the reflected signal, recalculates the real - time three - dimensional speed, acceleration, and motion direction of the target, and constructs an updated motion state matrix; The barometric pressure sensor array synchronously collects the barometric pressure data of each node, reconstructs the airflow velocity field through a spatial interpolation algorithm, and combines with the laser ranging module to update the terrain undulation, generating corrected environmental disturbance parameters; The infrared thermal imager captures the latest body surface thermal imaging sequence, extracts the limb swing frequency through feature point tracking, and inversely calculates the energy consumption rate based on a preset physiological model to update the biological behavior characteristic parameters.

[0068] After the above data is calibrated in space - time and filtered to remove noise, it is input into the trajectory prediction model to regenerate the initial trajectory, disturbance compensation trajectory, and final target trajectory. This closed - loop feedback mechanism effectively responds to environmental dynamic changes or target behavior mutations through real - time data update and trajectory correction, improving the prediction accuracy and system stability in complex scenarios.

[0069] Please refer to Figure 2 , which is a schematic structural diagram of a moving target trajectory prediction device provided by an embodiment of this application, including: The multi - source data acquisition unit 21 is used to acquire multi - source dynamic perception data of a moving target, where the multi - source dynamic perception data includes target motion state parameters, environmental disturbance parameters, and biological behavior characteristic parameters; The initial trajectory generation unit 22 generates an initial trajectory prediction result based on the target motion state parameters; The compensation trajectory determination unit 23 dynamically corrects the initial trajectory prediction result based on the environmental disturbance parameters to obtain a disturbance compensation trajectory; The target trajectory prediction unit 24 performs behavior pattern matching on the disturbance compensation trajectory based on the biological behavior characteristic parameters, and generates a target trajectory prediction result.

[0070] Please refer to Figure 3 , this embodiment of the present application further provides an electronic device 300, including a memory 310, a processor 320, and a computer program 311 stored on the memory 310 and executable on the processor. When the processor 320 executes the computer program 311, it implements the steps of any method for predicting the trajectory of a moving target.

[0071] Since the electronic device introduced in this embodiment is the device adopted for implementing a moving target trajectory prediction device in the embodiments of the present application, based on the methods introduced in the embodiments of the present application, those skilled in the art can understand the specific implementation manners and various variations of the electronic device in this embodiment. Therefore, the specific implementation of how this electronic device implements the methods in the embodiments of the present application will not be described in detail here. As long as the device adopted by those skilled in the art to implement the methods in the embodiments of the present application belongs to the scope protected by the present application.

[0072] In the specific implementation process, when the computer program 311 is executed by the processor, it can implement any implementation manner in the corresponding embodiment of the first aspect.

[0073] It should be noted that in the above embodiments, the descriptions of each embodiment have their own emphases. For the parts not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0074] Those skilled in the art should understand that the embodiments of the present application can provide methods, systems, or computer program products. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-readable storage media (including but not limited to disk memories, CD-ROMs, optical memories, etc.) containing computer-readable program codes.

[0075] The present application is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams can be implemented by computer program instructions, and the combination of the flows and / or blocks in the flowcharts and / or block diagrams can also be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded computer, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate for implementing in the process Figure 1 one process or multiple processes and / or blocks Figure 1a device for the functions specified in one or more boxes.

[0076] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory produce a manufacture including an instruction device that implements the functions specified in one Figure 1 process or more processes and / or boxes Figure 1 a box or more boxes.

[0077] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are performed on the computer or other programmable device to produce a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in one Figure 1 process or more processes and / or boxes Figure 1 a box or more boxes.

[0078] An embodiment of this application also provides a computer program product, which includes computer software instructions that, when running on a processing device, cause the processing device to execute Figure 1 the process of a motion target trajectory prediction method in the corresponding embodiment.

[0079] The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the processes or functions according to the embodiments of this application are fully or partially generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another, for example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center in a wired or wireless manner. The computer-readable storage medium can be any available medium that a computer can store or a data storage device such as a server or data center that includes one or more integrated available media. The available medium can be a magnetic medium, an optical medium, or a semiconductor medium, etc.

[0080] Those skilled in the art can clearly understand that for the convenience and conciseness of description, the specific working processes of the above-described systems, devices, and units can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0081] In several embodiments provided by this application, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical function division. In actual implementation, there can be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces. The indirect couplings or communication connections of devices or units can be in electrical, mechanical, or other forms.

[0082] The units described as separate components may or may not be physically separated. The components shown as units may or may not be physical units, that is, they can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0083] In addition, in each embodiment of this application, each functional unit can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit. The above-mentioned integrated units can be implemented in the form of hardware and / or software functional units.

[0084] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to enable a computer device to execute all or part of the steps of the methods in each embodiment of this application. The aforementioned storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories, magnetic disks, or optical discs that can store program codes.

[0085] The above embodiments are only used to illustrate the technical solutions of this application, rather than to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of each embodiment of this application.

[0086] Although the preferred embodiments of the present specification have been described, additional changes and modifications can be made to these embodiments by those skilled in the art once they learn the basic creative concepts. Therefore, the appended claims are intended to be construed to include the preferred embodiments as well as all changes and modifications that fall within the scope of this specification.

[0087] Obviously, those skilled in the art can make various changes and deformations to this specification without departing from the spirit and scope of this specification. Thus, if these modifications and deformations of this specification fall within the scope of the claims of this specification and their equivalent technologies, this specification is also intended to include these modifications and deformations.

Claims

1. A moving target trajectory prediction method, characterized in that: include: Acquiring multi-source dynamic perception data of a moving target, wherein the multi-source dynamic perception data includes target motion state parameters, environmental disturbance parameters, and biological behavior characteristic parameters; Based on the target motion state parameters, generating an initial trajectory prediction result; Based on the environmental disturbance parameter, dynamically correct the initial trajectory prediction result to obtain a disturbance compensation trajectory; Based on the biological behavior characteristic parameters, behavior pattern matching is performed on the disturbance compensation trajectory to generate a target trajectory prediction result.

2. The method according to claim 1, characterized in that The acquiring of multi-source dynamic perception data of the moving target includes: The real-time speed, acceleration and moving direction of the moving target are obtained by millimeter wave radar as the target motion state parameters; The air flow velocity, air pressure gradient and terrain undulation around the moving target are collected as the environmental disturbance parameters through a distributed air pressure sensor array; The body surface temperature distribution and limb swing frequency of the moving target are extracted by an infrared thermal imager as the biological behavior characteristic parameters.

3. The method according to claim 2, characterized in that The generating an initial trajectory prediction result based on the target motion state parameter includes: Calculating the instantaneous trajectory vector of the moving target based on the real-time speed, the acceleration and the moving direction; Constructing a target motion state matrix according to the instantaneous trajectory vector and a preset kinematic model; The target motion state matrix is ​​input into a trajectory prediction neural network to generate the initial trajectory prediction result, wherein the trajectory prediction neural network is obtained by training historical motion data and is used to output a trajectory sequence matching the current motion state.

4. The method according to claim 2, characterized in that: The dynamically correcting the initial trajectory prediction result based on the environmental disturbance parameter to obtain a disturbance compensation trajectory includes: Determining a fluid resistance coefficient of the moving target based on the airflow velocity and the moving direction; Based on the terrain relief, determining a terrain elevation change rate; Determining a disturbance compensation model based on the terrain elevation change rate and the air pressure gradient; Based on the fluid resistance coefficient and the disturbance compensation model, the initial trajectory prediction result is dynamically corrected to determine a disturbance compensation trajectory.

5. The method according to claim 2, characterized in that: The performing behavior pattern matching on the disturbance compensation trajectory based on the biological behavior characteristic parameters to generate a target trajectory prediction result includes: Determining the energy consumption rate of the sports target based on the body surface temperature distribution; Based on the limb swing frequency, constructing a motion rhythm map of the motion target; Determining a trajectory segment correction intensity based on a first comparison result between the energy consumption rate and a preset energy threshold; Based on the motion rhythm map, matching a reference trajectory segment in a historical behavior pattern library, wherein the similarity between the reference trajectory segment and the motion rhythm map is greater than a preset similarity threshold; Based on the trajectory segment correction strength and the reference trajectory segment, the disturbance compensation trajectory is temporally and spatially superimposed to generate a target trajectory prediction result.

6. The method according to claim 1, characterized in that Also includes: Based on the target trajectory prediction result, generating a trajectory visualization map, wherein the trajectory visualization map includes a dynamic risk area mark; Based on the dynamic risk area mark, an obstacle avoidance instruction or a path optimization suggestion corresponding to the moving target is determined.

7. The method according to claim 1, characterized in that Also includes: Based on the target trajectory prediction result, real-time monitoring of the deviation between the actual motion path of the moving target and the target trajectory prediction result; When the deviation is greater than a first preset threshold, triggering a re-collection instruction of the multi-source dynamic sensing data; Based on the re-acquisition instruction, the multi-source dynamic sensing data is reacquired.

8. A moving target trajectory prediction device, characterized in that: include: A multi-source data acquisition unit, used to acquire multi-source dynamic perception data of a moving target, wherein the multi-source dynamic perception data includes target motion state parameters, environmental disturbance parameters, and biological behavior characteristic parameters; An initial trajectory generating unit, which generates an initial trajectory prediction result based on the target motion state parameter; A compensation trajectory determination unit dynamically corrects the initial trajectory prediction result based on the environmental disturbance parameter to obtain a disturbance compensation trajectory; The target trajectory prediction unit performs behavior pattern matching on the disturbance compensation trajectory based on the biological behavior characteristic parameters to generate a target trajectory prediction result.

9. An electronic device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor is used to implement the steps of the moving target trajectory prediction method as described in any one of claims 1 to 7 when executing the computer program stored in the memory.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the moving target trajectory prediction method according to any one of claims 1 to 7 is implemented.

Images