Moving object path planning algorithm and equipment capable of positioning in real time

Abstract

The invention relates to the technical field of remote positioning algorithms, in particular to a moving object path planning algorithm and equipment capable of positioning in real time. The algorithm comprises the following steps: S1, acquiring a positioning signal; s2, abnormal value elimination; s3, signal filtering; s4, performing a canine path prediction algorithm; and S5, establishing an early warning mechanism. A plurality of positioning modules are arranged in the neck ring, the optimal positioning module is automatically switched to send position information, the stability of signal transmission is ensured, after the control center end receives positioning and motion data, abnormal values are subjected to impurity removal through deep learning, then the data are corrected through Kalman filtering, the accuracy of input signals is improved, and the accuracy of the input signals is improved. Through the corrected positioning and motion data, the control center end can automatically identify the current motion mode of the dogs and match a corresponding prediction algorithm, predict the motion path of the dogs in a certain time in the future and push early warning information to a mobile terminal interface of a user, and the accuracy and safety in the hunting process of the dogs are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of remote positioning algorithm, in particular to a mobile object path planning algorithm capable of real-time positioning and a device. BACKGROUND

[0002] Hunting dogs, also known as hunting dogs, are dogs trained to help hunters. Hunting dogs trained by special commands or gestures have developed olfactory, auditory and visual senses, which can help hunters find more prey and create more favorable shooting opportunities. Different hunting dogs have different roles, some are used for hunting, some are used as indicator dogs to disturb the prey, and some are used to pick up the prey.

[0003] The scene of hunting dogs chasing or indicating prey is generally wide open. In the case of obstruction by trees, shrubs and other obstructions, it is difficult for the owner to observe the specific position and action path of the hunting dog. Corresponding commands or gestures are also difficult to convey to the hunting dog. Correspondingly, the barking of the hunting dog when capturing the prey is also fuzzy when it is delivered to the owner. Therefore, a mobile object path planning algorithm capable of real-time positioning and a device are provided for real-time positioning and route planning of the current position of the hunting dog. SUMMARY

[0004] Other features and advantages of the present application will be set forth in the following specification, and in part will become apparent from the specification, or can be learned by practice of the present application. The objects and other advantages of the present application can be achieved and obtained by the structure specifically pointed out in the specification and other accompanying drawings.

[0005] The present application aims to overcome the above-mentioned deficiencies, and provides a mobile object path planning algorithm capable of real-time positioning and a device. By providing a plurality of positioning modules in the neck ring, the optimal positioning module is automatically switched to send position information, ensuring the stability of signal transmission. After receiving the positioning and motion data, the control center end removes the impurities through deep learning, and then corrects the data through Kalman filtering, improving the accuracy of the input signal. Through the corrected positioning and motion data, the control center end can automatically identify the current motion mode of the dog and match the corresponding prediction algorithm, predict the motion path of the dog within a certain time in the future, and push the warning information to the user's mobile terminal interface, improving the accuracy and safety of the hunting process of the dog.

[0006] The present application provides a mobile object path planning algorithm capable of real-time positioning, comprising: S1. Positioning signal acquisition: The collar is equipped with a UWB positioning module, a GPS positioning module, a dual-antenna positioning module, a judgment chip, and a three-axis accelerometer. The judgment chip collects multi-frequency signals and automatically switches to the optimal positioning module to transmit the current position data by judging the signal quality. The three-axis accelerometer synchronously sends the motion characteristics to the control center. S2. Outlier Removal: The control center establishes a deep learning model based on spatiotemporal feature fusion and adds a self-attention mechanism to remove outliers from the location data. S3. Signal Filtering: Perform Kalman filtering on the noise-removed position data to obtain corrected position data; S4. Canine Path Prediction Algorithm: Based on the corrected position data, store the n most recent position points, extract motion features, and predict the path using the current speed. Current direction angle and acceleration The system comprehensively judges the dog's current behavior pattern, selects the corresponding prediction algorithm based on the behavior pattern, and generates m prediction points for the future. The dog's current behavior pattern specifically includes running in a straight line, maintaining direction, making sharp turns, and chasing in an arc. S5. Establish an early warning mechanism: Based on m predicted points generated in the future, plan an alarm mechanism, push alarm information to the mobile terminal display interface, control the collar to issue corresponding instructions, and remind the user whether to intercept the password.

[0007] In some embodiments, the specific method for automatically switching to the optimal positioning module to transmit current location data by judging signal quality in step S1 includes: S11. Detect the connection status of the UWB positioning module, GPS positioning module and dual-antenna positioning module. For the unconnected positioning module, set its score to 0. S12. The signal quality of the UWB positioning module, GPS positioning module and dual-antenna positioning module is scored. If any positioning module scores 0.7 or higher, then the positioning module is selected as the optimal positioning module and the current location data is transmitted. S13. If at least two positioning modules have a score greater than or equal to 0.7, the positioning module with the highest score is selected to transmit the current location data; if all three positioning modules have a score less than 0.7, the positioning module with the highest score is still selected to transmit the current location data, and a weak positioning signal information is sent to the mobile terminal display interface.

[0008] In some embodiments, the specific formula for calculating the UWB positioning module score is as follows: in, The standard signal-to-noise ratio for UWB is... The standard value for the multi-path ratio of UWB. ; The specific formula for calculating the GPS positioning module score is as follows: in, The percentage of visible satellites after normalization. It is a piecewise function that maps the HDOP value of the GPS positioning module to 0-1. This is the standard signal-to-noise ratio value for GPS. ; The specific calculation formula for the dual-antenna positioning module score is as follows: in, It is a piecewise function that maps the fixed score of the dual-antenna positioning module to 0-1. This is another piecewise function that maps the HDOP value of the GPS positioning module to 0-1. .

[0009] In some embodiments, the specific steps for deep learning to remove outliers in step S2 are as follows: S21. Obtain the original trajectory data based on the current location data, preprocess the data, and extract spatiotemporal features; S22. Perform LSTM time series analysis on the extracted spatiotemporal features to extract time features; S23. Add an attention mechanism to the extracted spatiotemporal features to identify key points; S24. Perform feature fusion on the data obtained in steps S22 and S23 to obtain a comprehensive feature vector; S25. Input the comprehensive feature vector into the fully connected layer and calculate the anomaly probability using the sigmoid function.

[0010] In some embodiments, in step S25, differentiation is performed based on the numerical range of the anomaly probability P: If the abnormal probability If the value is missing, it is determined to be an outlier. The current value is removed and marked as missing. The data of the two points before and after the current point are averaged and then added to the missing position. If the abnormal probability If the value is 0, it is considered a normal value, and the original data is retained. If the abnormal probability If the current value is missing, wait for steps S3-S4 to be executed. After judging the canine behavior pattern in step S4, select the corresponding prediction algorithm, backtrack to derive the missing point data, and fill the missing position with the obtained backtracked value.

[0011] In some embodiments, in the S4 step, the correspondence between the canine behavior pattern and the prediction algorithm comprises: Straight running - constant straight line prediction: Where the current position is , the current speed is , the current direction angle is , the prediction time step is , , , and the output is m prediction point positions in the future for a certain time, ; Sharp turn - deceleration U-turn prediction: Where the current position is , the current speed is , the instantaneous speed is , the acceleration is , the prediction time step is , , , the U-turn direction angle is , , the current direction angle is , and the output is m prediction point positions in the future for a certain time, ; Arc chase - constant arc prediction: Where the current position is , the current speed is , the current direction angle is , and the centripetal acceleration is ; The turning radius calculation formula is , if is unknown, take the empirical value , and the angular velocity is ; The prediction time step is , , , and the output is m prediction point positions in the future for a certain time, ; Maintain direction - simplified straight line prediction: Where the current position is , the average speed of the last three points is , the position point is i, the current position point speed is , and the average angle of the direction of the last two displacements is , the position point is i, the current position displacement direction angle is ; the prediction time step , , , output the position of m prediction points in the future for a certain time, .

[0012] In some embodiments, the current behavior mode of the dog is determined by comprehensively judging the current speed , the current direction angle and the acceleration : The input features include the current speed , the calculated direction angle change and the acceleration .

[0013] In some embodiments, in the S5 step, according to the current behavior mode of the dog and the predicted path, the corresponding warning mechanism planning is carried out, and the warning mechanism specifically includes: straight running collision warning: set the collision warning threshold, when the predicted path intersects with the forbidden area or the dangerous area, according to the current speed , predict the time TTC to reach the area, judge whether a collision occurs, push the page to display the warning information, TTC = the distance from the dog to the nearest boundary of the dangerous area / current speed; when or deviates from the predicted path, the page is pushed to display green; when , the page is pushed to display yellow; when , the page is pushed to display red; arc pursuit out-of-control warning: set the minimum safe curvature radius and the collision warning threshold, when the predicted path will pass through the dangerous area and the sharp turn may be out of control, judge whether it is out of control, push the page to display the warning information; the path does not pass through the dangerous area and the curvature radius < the safety threshold, the page is pushed to display green; the path passes through the dangerous area but or the curvature radius reaches 1.2 times the safe curvature radius, the page is pushed to display yellow; the path passes through the dangerous area and or the curvature radius is less than the minimum safe curvature radius, the page is pushed to display red; sudden U-turn warning: set the distance threshold between the position after the U-turn and the dangerous area, judge whether the dangerous area is entered after the U-turn, push the page to display the warning information; The push page displays green color when the post-turn path has no collision risk; the push page displays yellow color when the post-turn path is 2-5 meters away from the dangerous area; and the push page displays red color when the post-turn path is less than 2 meters away from the dangerous area or the predicted collision time is less than or equal to 3 seconds.

[0014] A device for real-time positioning of a path of a mobile object, specifically comprising: A collar, in which a UWB positioning module, a GPS positioning module, a double-antenna positioning module, a judgment chip and a three-axis accelerometer are arranged; A control center, in which a deep learning model based on space-time feature fusion, a Kalman filter model and a canine path prediction algorithm are arranged; A push page, which is arranged on a mobile terminal and is electrically connected or wirelessly connected with the control center.

[0015] In some embodiments, a public address loudspeaker is further arranged on the collar, which is wirelessly connected with the control center and the mobile terminal, and when the user receives the early warning information of the push page on the mobile terminal, the password is transmitted to the public address loudspeaker of the collar to intercept the canine.

[0016] By adopting the above technical solutions, the device has the following advantages: The device has the following advantages: the multiple positioning modules are arranged in the collar, the optimal positioning module is automatically switched to send the position information, the stability of signal transmission is ensured, the control center receives the positioning and motion data, the abnormal values are removed through deep learning, the data is corrected through the Kalman filter, the accuracy of the input signal is improved, the current motion mode of the canine is automatically identified by the control center through the corrected positioning and motion data, the corresponding prediction algorithm is matched, the motion path of the canine in the future is predicted, and the early warning information is pushed to the mobile terminal interface of the user, thereby improving the accuracy and safety of the canine during the hunting process.

[0017] It should be understood that the above general description and the following detailed description are only exemplary and explanatory, and cannot limit the present disclosure.

[0018] It is obvious that such purposes and other purposes of the present application will become more apparent after the following description of the preferred embodiments with various drawings and diagrams.

[0019] In order to make the above and other purposes, features and advantages of the present application more apparent and easy to understand, one or more preferred embodiments are described below, and the drawings are shown as follows. BRIEF DESCRIPTION OF DRAWINGS

[0020] The drawings are used to provide a further understanding of the present application, and constitute a part of the specification. They are used together with the embodiments of the present application to explain the present application, and do not constitute a limitation of the present application.

[0021] In the drawings, the same components have the same reference numbers, and the drawings are schematic and are not necessarily to scale.

[0022] In order to more clearly illustrate the technical solutions of the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only one or some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative effort based on these drawings.

[0023] Figure 1 The schematic diagram of the overall flow of the path planning of the moving object in some embodiments of the present application; Figure 2 The schematic diagram of the automatic switching principle of the positioning module in some embodiments of the present application; Figure 3 The schematic diagram of the deep learning model structure in some embodiments of the present application; Figure 4 The schematic diagram of the differential processing of the abnormal probability in some embodiments of the present application; Figure 5 The schematic diagram of the corresponding relationship between the dog behavior pattern and the prediction algorithm in some embodiments of the present application; Figure 6 The schematic diagram of the type of the planning early warning mechanism in some embodiments of the present application; Figure 7 The schematic diagram of the fitting of the actual motion path and the predicted path of the dog in some embodiments of the present application. DETAILED DESCRIPTION

[0024] In order to make the purpose, technical solutions and advantages of the present application more clear, the present application will be further described in detail below in combination with specific embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application, but not to limit the present application.

[0025] In addition, in the description of the present application, it should be understood that the orientations or positional relationships indicated by the terms "center", "upper", "lower", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "axial", "radial", "circumferential" and the like are based on the orientations or positional relationships shown in the drawings, and are only for the convenience of describing the present application and simplifying the description, and therefore cannot be understood as indicating or implying that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as limiting the present application.

[0026] In the present application, unless otherwise explicitly specified and limited, the terms "mounting", "connection", "connecting", "fixing" and the like should be understood in a broad sense, for example, can be fixed connection, can also be detachable connection, or integral; can be directly connected, or indirectly connected through an intermediate medium, can be the internal communication of two elements or the interaction relationship of two elements. However, it is noted that direct connection means that the connection between the two main bodies does not form a connection relationship through a transition structure, but is connected only through a connection structure to form a whole. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances.

[0027] In the present application, unless otherwise explicitly specified and limited, the first feature is "on" or "under" the second feature, which can be direct contact between the first and second features, or indirect contact between the first and second features through an intermediate medium. In the description of the present specification, the description referring to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In the present specification, the illustrative description of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any appropriate manner in any one or more embodiments or examples.

[0028] Referring to Figures 1-2 , Figure 1 The figure is a schematic diagram of the overall flow of the mobile object path planning in some embodiments of the present application. Figure 2 The figure is a schematic diagram of the automatic switching principle of the positioning module in some embodiments of the present application.

[0029] According to some embodiments of the present application, the present application provides a mobile object path planning algorithm that can be positioned in real time, comprising: S1, positioning signal acquisition: the collar is provided with a UWB positioning module, a GPS positioning module, a double-antenna positioning module, a judgment chip and a three-axis accelerometer, the judgment chip collects multi-band signals, and automatically switches to the optimal positioning module to transmit current position data by judging the signal quality, and the three-axis accelerometer synchronously sends motion characteristics to the control center end; The characteristics of the UWB positioning module make it have good anti-interference performance on electromagnetic interference and obstacle penetration, so that it can realize real-time positioning and ensure the upper limit of the transmission of the positioning signal, but the disadvantage is that the coverage range is relatively small, especially in the case of shielding, the positioning accuracy will be affected, therefore, the GPS positioning module is considered to be added, the advantages of GPS positioning are wide application range, large number of satellites and easy observation, and the cost is low under the condition of ensuring the basic positioning ability, which can ensure the lower limit of the positioning ability of the necklace, and a double-antenna positioning module is also arranged in the necklace, the double-antenna positioning module realizes accurate calculation of the target position by analyzing the time difference and angle of arrival of the signals received by the two antennas, and has high-precision directional capability, but it may need to be combined with other positioning systems to provide absolute position, therefore, on the basis of the original two positioning modules, the double-antenna positioning module can be used as the third positioning source for auxiliary positioning; In the S1 step, the specific method of automatically switching to the optimal positioning module to transmit the current position data by judging the signal quality includes: S11, detecting the connection state of the UWB positioning module, the GPS positioning module and the double-antenna positioning module, and setting the score of the unconnected positioning module to 0; S12, scoring the signal quality of the UWB positioning module, the GPS positioning module and the double-antenna positioning module, if the score of a positioning module is greater than or equal to 0.7, selecting the positioning module as the optimal positioning module and transmitting the current position data; S13, if the scores of at least two positioning modules are greater than or equal to 0.7, selecting the positioning module with the highest score to transmit the current position data; if the scores of the three positioning modules are all less than 0.7, still selecting the positioning module with the highest score to transmit the current position data, and sending the information of weak positioning signal to the display interface of the mobile terminal; The specific calculation formula of the score of the UWB positioning module is: Among them, is the standard value of the signal-to-noise ratio of UWB, is the standard value of the multipath ratio of UWB, ; specifically, the signal-to-noise ratio reflects the signal quality, , is the signal-to-noise ratio of UWB, 40dB is the assumed maximum decibel of UWB signal, and the standardized value is normalized to 0-1; the multipath ratio reflects the anti-interference ability, , 10dB is the assumed optimal multipath ratio, and the standardized value is normalized to 0-1, and finally the score of the UWB positioning module is obtained by assigning weights; The specific calculation formula of the score of the GPS positioning module is: wherein, is the normalized visible satellite ratio, is a piecewise function mapping the HDOP value of the GPS positioning module to 0-1, is the standard value of the signal-to-noise ratio of GPS, ; Specifically, the total number of positioning satellites of the GPS positioning is 12, and the minimum number of visible satellites in every corner of the earth is 4. The ratio of the current number of visible satellites to the total number of satellites is normalized and mapped to 0-1. For the piecewise function mapping the HDOP value of the GPS positioning module to 0-1 , different segment values are obtained according to different obtained HDOP values. HDOP is the horizontal dilution of precision: The signal-to-noise ratio of GPS is similar to that of UWB. The GPS signal-to-noise ratio is limited to -20 dB to 20 dB, and the normalized value is normalized to 0-1. Finally, the GPS positioning module score is obtained by assigning weights. The specific calculation formula of the dual-antenna positioning module score is: wherein, is a piecewise function mapping the fixed score of the dual-antenna positioning module to 0-1, is another piecewise function mapping the HDOP value of the GPS positioning module to 0-1, ; For the piecewise function mapping the fixed score of the dual-antenna positioning module to 0-1 , different segment values are obtained according to the solution of the fixed state fix: The specific process of how the dual-antenna positioning module obtains the fixed state score can refer to the prior art, which will not be described in detail here. For another piecewise function mapping the HDOP value of the GPS positioning module to 0-1 , different segment values are obtained according to different obtained HDOP values. HDOP is the horizontal dilution of precision: The reason for providing two different segmented functions for the HDOP horizontal precision factor also obtained from the GPS positioning module is that the weights of the HDOP are distributed differently in the GPS positioning score and the dual-antenna positioning score. In actual use, the smaller the HDOP value is, the better. In the GPS positioning score, the maximum tolerance is given to 9, and for the dual-antenna positioning score, the maximum tolerance of HDOP is only given to 5, because the weight term in the GPS positioning score is more, and the corresponding influence of the HDOP value is reduced.

[0030] Referring to Figures 3-4 , Figure 3 is a deep learning model structure schematic diagram in some embodiments of the application; Figure 4 is an abnormal probability differential processing schematic diagram in some embodiments of the application.

[0031] S2, abnormal value exclusion: the control center end establishes a deep learning model based on spatiotemporal feature fusion, adds a self-attention mechanism, and removes impurities from abnormal values in position data; In the S2 step, the specific steps of the deep learning to remove abnormal values are: S21, obtaining original trajectory data according to current position data, pre-processing the data, and extracting spatiotemporal features; Receiving original trajectory point data from a UWB, GPS or dual-antenna positioning module, each point containing a timestamp, an x coordinate and a y coordinate, resampling the trajectory points at a fixed time interval, taking the last T points as model input, and normalizing the coordinate values to the 0-1 interval; The deep learning model based on spatiotemporal feature fusion specifically includes an input layer, an LSTM layer, a self-attention layer, a fully connected layer and an output layer, wherein the size of the input layer is T*3; S22, performing LSTM time series analysis on the extracted spatiotemporal features to extract time features; S23, adding an attention mechanism to the extracted spatiotemporal features to identify key points; S24, performing feature fusion on the data obtained in the S22 step and the S23 step to obtain a comprehensive feature vector; S25, inputting the comprehensive feature vector into the fully connected layer, calculating the abnormal probability through the sigmoid function; the output layer finally outputs the abnormal probability, which is between 0 and 1; According to the numerical range of the abnormal probability P, differential processing is performed: If the abnormal probability P is less than 0.5, it is determined that the value is an abnormal value, the current value is removed and marked as missing, and the data of the two points before and after the current point are averaged and filled into the missing position; If the abnormal probability P is greater than 0.5, it is determined that the value is a normal value, and the original data is retained; If the abnormal probability P is greater than 0.5, it is determined that the value is a normal value, and the original data is retained; If the abnormal probability P is greater than 0.5, it is determined that the value is a normal value, and the original data is retained; If the abnormal probability is , the current value is marked as missing, and the S3-S4 steps are executed. After judging the behavior mode of the dog in the S4 step, the corresponding prediction algorithm is selected, the missing point data is derived back, and the obtained backtracking value is supplemented to the missing position in reverse.

[0032] When the abnormal probability of the value is in the interval of 0.7-0.9, the usual processing method is to perform speed constraint correction, that is, the current position speed is calculated through historical average data, and the fitted data is supplemented to the missing position. However, considering the uncertainty of the dog's movement, there may be a large difference between the current movement state and the historical movement state, so the subsequent corresponding prediction algorithm is selected by sequentially executing the behavior mode of the dog, and the missing point data is derived back through the prediction algorithm. After the experiment, it is found that the accuracy of the position data derived by the backtracking method is improved by 32.8% compared with the data obtained by the speed constraint correction method, so the backtracking method is used to supplement the missing value.

[0033] S3, signal filtering: performing Kalman filtering on the position data after impurity removal to obtain corrected position data; the operation of performing Kalman filtering on the positioning signal is a common operation in the navigation field to improve the accuracy and robustness of the signal. In the specific use process, the user connects the positioning signal to the Kalman filtering model, and the output data is the corrected positioning data. Preferably, the Kalman filtering selects a two-dimensional uniform Kalman filtering model.

[0034] Referring to Figure 5 and Figure 7 , Figure 5 is a corresponding relationship diagram between the behavior mode of the dog in some embodiments of the present application and the prediction algorithm; Figure 7 is a fitting diagram of the actual movement path of the dog and the predicted path in some embodiments of the present application.

[0035] S4, dog path prediction algorithm: according to the corrected position data, the latest n position points are stored, the motion characteristics are extracted, the current speed , the current direction angle and the acceleration are comprehensively judged to determine the current behavior mode of the dog, the corresponding prediction algorithm is selected according to the behavior mode, and the future m predicted points are generated; the current behavior mode of the dog specifically includes straight running, keeping direction, sharp turning and arc chasing; The current speed , the current direction angle and the acceleration are obtained through a three-axis accelerometer, and the instantaneous speed required for subsequent calculation can also be obtained through calculation , the direction angle change , the turning angle , and the centripetal acceleration and angular velocity Preferably, the motion features output by the triaxial accelerometer are also filtered by Kalman filter to improve the accuracy and reliability of the data. To improve the accuracy of determining the current behavior pattern of the dog, the motion features of common dogs are introduced herein, which are shown in Table 1: Table 1: Preset values of motion features of dogs By the preset values of the motion features of common dogs described above, the current behavior pattern of the dog can be comprehensively determined in combination with the decision tree model, so as to facilitate the selection of the corresponding prediction algorithm according to the behavior pattern subsequently: It can be understood that the data in Table 1 is only general data of the motion of common dogs, and a larger range is given herein, and the producer can make adaptive modifications according to the actual test data and different dog breeds, and the data in the table are only for reference. The corresponding relationship between the behavior pattern of the dog and the prediction algorithm includes: Straight running - uniform straight line prediction: wherein the current position is , the current speed is , the current direction angle is , the prediction time step is , , , and the output of the m prediction point positions in a certain time in the future is ; Sharp turn - deceleration U-turn prediction: wherein the current position is , the current speed is , the instantaneous speed is , the acceleration is , the prediction time step is , , , the U-turn direction angle is , , the current direction angle is , the U-turn angle is, and the output of the m prediction point positions in a certain time in the future is ; Arc chase - uniform arc prediction: wherein the current position is , the current speed is , the current direction angle is , centripetal acceleration is ; turning radius calculation formula , if unknown, take the empirical value , angular velocity ; Predictive time step , , , output m predicted point positions in a certain time in the future, ; Keep direction - Simplified straight line prediction: Wherein, the current position is , the average speed of the last three points , the position point is i, and the current position point speed is ; The average angle of the displacement direction of the last two times is , the position point is i, and the current position displacement direction angle is ; Predictive time step , , , output m predicted point positions in a certain time in the future, ; Compared with the common patrol robot car movement mode, the dog movement mode has certain similarity in straight line movement and arc movement, but the difference is also obvious. First of all, the freedom of dog movement is high, so a high-precision positioning module is required, and the time interval of the positioning signal capture is also shorter. The scheme of mutual cooperation of UWB positioning module, GPS positioning module and double antenna positioning module can well solve this point. The positioning signal sending frequency of the robot car can be long time interval. In the process of robot car exploration, laser radar and vision camera are necessary components for detecting obstacles and target tracking. Dogs have visual function and basic obstacle avoidance ability, so laser radar is not a necessary component. Dogs can realize positioning and obstacle avoidance function by wearing a collar. In the specific prediction algorithm, the robot car and the chassis are rigidly connected, so it is difficult to make 180° turn and similar actions in a short time, so the prediction algorithm is relatively weak. The dog prediction algorithm is convenient, and the randomness of dog movement needs to be considered. The arc pursuit, sharp turn and other actions are included in the prediction algorithm, which improves the accuracy of the predicted route.

[0036] Referring to Figure 6 , Figure 6 is the schematic diagram of the planning warning mechanism type in some embodiments of the application.

[0037] S5, establishing an early warning mechanism: according to the future generated m prediction points, planning the alarm mechanism, pushing the alarm information into the mobile terminal display interface, the control ring sends corresponding instructions, reminding the user whether to perform password interception; In the S5 step, according to the current behavior pattern and the predicted path of the dog, the corresponding early warning mechanism is planned, and the early warning mechanism specifically includes: Straight running collision warning: Set the collision warning threshold, when the predicted path intersects with the forbidden area or the dangerous area, according to the current speed Predict the time TTC to reach the area, judge whether a collision occurs, push the page to display the warning information, TTC = dog's distance to the nearest boundary of the dangerous area / current speed; When Or deviate from the predicted path, push the page to display green; when , push the page to display yellow; when , push the page to display red; Arc chase out of control warning: Set the minimum safe curvature radius and the collision warning threshold, when the predicted path will pass through the dangerous area and the sharp turn may be out of control, judge whether it is out of control, push the page to display the warning information; The path does not pass through the dangerous area and the curvature radius is less than the safety threshold, push the page to display green; the path passes through the dangerous area but Or the curvature radius reaches 1.2 times the safety curvature radius, push the page to display yellow; the path passes through the dangerous area and Or the curvature radius is less than the minimum safety curvature radius, push the page to display red; Sudden U-turn warning: Set the distance threshold between the position after the U-turn and the dangerous area, judge whether the dangerous area is entered after the U-turn, push the page to display the warning information; The path after the U-turn has no collision risk, push the page to display green; the path after the U-turn is 2-5 meters away from the dangerous area, push the page to display yellow; the path after the U-turn is less than 2 meters away from the dangerous area or the predicted collision time is less than or equal to 3s, push the page to display red.

[0038] The application also provides a device for real-time positioning of a moving object path, which specifically comprises: The collar is internally provided with a UWB positioning module, a GPS positioning module, a double-antenna positioning module, a judgment chip and a three-axis accelerometer; The control center end is internally provided with a deep learning model based on space-time feature fusion, a Kalman filter model and a dog path prediction algorithm; The push page is arranged on the mobile terminal and is electrically connected or wirelessly connected with the control center end; The broadcast speaker is wirelessly connected with the control center terminal and the mobile terminal. When the user receives the early warning information of the push page on the mobile terminal, the password is transmitted to the broadcast speaker of the necklace, and the dog is intercepted by the password.

[0039] It should be understood that the embodiments disclosed herein are not limited to particular process steps or materials disclosed herein, but rather, the same or equivalent processes can be used. It should also be understood that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting.

[0040] Reference throughout this specification to "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment of the application. Thus, appearances of the phrase "in one embodiment" or "in an embodiment" in various places throughout this specification are not necessarily all referring to the same embodiment.

[0041] Furthermore, the described features, characteristics, or aspects can be combined in any suitable manner in one or more embodiments. In the above description, numerous specific details are provided, such as specific material types, dimensions, etc., to provide a thorough understanding of embodiments of the application. It will be apparent, however, to one skilled in the art that the application can be practiced without one or more of the specific details.

Claims

1. A real-time locatable mobile object path planning algorithm, characterized in that, Comprising S1, positioning signal acquisition: the necklace is provided with a UWB positioning module, a GPS positioning module, a double-antenna positioning module, a judgment chip and a three-axis accelerometer. The judgment chip collects multi-band signals, automatically switches to the optimal positioning module to transmit the current position data by judging the signal quality, and the three-axis accelerometer synchronously sends the motion characteristics to the control center end; S2, abnormal value exclusion: the control center end establishes a deep learning model based on space-time feature fusion, adds a self-attention mechanism, and removes the abnormal values in the position data; S3, signal filtering: Kalman filtering is performed on the position data after the impurities are removed to obtain the corrected position data; S4, canine path prediction algorithm: according to the corrected position data, store the latest n position points, extract motion features, and comprehensively judge the current behavior mode of the canine through the current speed , current direction angle and acceleration ; according to the behavior mode, select the corresponding prediction algorithm to generate m future prediction points; the current behavior mode of the canine specifically includes straight running, keeping direction, sharp turning and arc chasing; S5, establishing an early warning mechanism: according to the m prediction points generated in the future, the alarm mechanism is planned, the alarm information is pushed into the mobile terminal display interface, the necklace sends corresponding instructions, and the user is reminded whether to perform password interception.

2. The real-time locatable mobile object path planning algorithm of claim 1, wherein, In S1, the specific method of automatically switching to the optimal positioning module to transmit the current position data by judging the signal quality includes: S11, detecting the connection state of the UWB positioning module, the GPS positioning module and the double-antenna positioning module, and setting the score of the unconnected positioning module to 0; S12, scoring the signal quality of the UWB positioning module, the GPS positioning module and the double-antenna positioning module, if the score of a positioning module is greater than or equal to 0.7, selecting the positioning module as the optimal positioning module and transmitting the current position data; S13, if the scores of at least two positioning modules are greater than or equal to 0.7, selecting the positioning module with the highest score to transmit the current position data, if the scores of the three positioning modules are all less than 0.7, still selecting the positioning module with the highest score to transmit the current position data, and sending the information that the positioning signal is weak to the mobile terminal display interface.

3. The real-time positioning mobile object path planning algorithm according to claim 2, characterized in that, The specific calculation formula of the UWB positioning module score is: wherein is a standard value of a signal-to-noise ratio of the UWB, is a standard value of a multipath ratio of the UWB, ; The specific calculation formula of the GPS positioning module score is: wherein, is the normalized visible satellite ratio, is a piecewise function mapping the HDOP value of the GPS positioning module to 0-1, is the signal-to-noise ratio standard value of the GPS, ; The specific calculation formula of the double-antenna positioning module score is: wherein, is a piecewise function mapping the fixed score of the dual antenna positioning module to 0-1, is another piecewise function mapping the HDOP value of the GPS positioning module to 0-1, .

4. The real-time locatable mobile object path planning algorithm of claim 1, wherein, In S2, the specific steps of deep learning to remove abnormal values are: S21, obtaining original trajectory data according to the current position data, preprocessing the data, and extracting space-time features; S22, performing LSTM time series analysis on the extracted space-time features to extract time features; S23, adding an attention mechanism to the extracted space-time features to identify key points; S24, performing feature fusion on the data obtained in steps S22 and S23 to obtain a comprehensive feature vector; S25, inputting the comprehensive feature vector into a fully connected layer and calculating the abnormal probability by a sigmoid function.

5. The real-time locatable mobile object path planning algorithm of claim 4, wherein, In S25, the abnormal probability P is differentiated according to the numerical range: If the abnormal probability is greater than 0.05, it is determined that the value is an abnormal value, the current value is removed and marked as missing, and the data of the two points before and after the current point are averaged and filled into the missing position. If the abnormal probability is , the normal value is determined, and the original data is retained. If the abnormal probability , the current value is marked as missing, and the S3-S4 steps are executed. After judging the behavior pattern of the dog in the S4 step, the corresponding prediction algorithm is selected, the missing point data is derived back, and the obtained backtracking value is supplemented to the missing position.

6. The real-time locatable mobile object path planning algorithm of claim 1, wherein, In S4, the corresponding relationship between the canine behavior mode and the prediction algorithm includes: Straight running - uniform straight line prediction: Wherein the current position is , the current speed is , the current direction angle is , the prediction time step is , , , the output future position of m prediction points within a certain time, ; Sharp turn - deceleration U-turn prediction: Wherein the current position is , the current speed is , the instantaneous speed is , the acceleration is , the prediction time step is , , , the turn-back direction angle is , , the current direction angle is , the turn-back angle is, and the output is m prediction point positions in the future for a certain time ; Arc chase - uniform arc line prediction: Wherein, the current position is , the current speed is , the current direction angle is , and the centripetal acceleration is ; Turning radius calculation formula , if unknown, take the empirical value , angular velocity ; Predicted time step , , , output the position of m prediction points in the future for a certain time, ; Keep direction - simplified straight line prediction: Wherein, the current position is , the average speed of the last three points , the position point is i, and the current position point speed is ; the average angle of the last two displacement directions is , the position point is i, and the current position displacement direction angle is ; the prediction time step is , , , the output future position of m prediction points within a certain time, .

7. The real-time locatable mobile object path planning algorithm of claim 6, wherein, by the current speed , the current direction angle and the acceleration The decision tree model is adopted to comprehensively judge the current behavior mode of the dog according to the cases The input features include current speed , a direction angle change amount , and acceleration that are calculated after.

8. The real-time locatable mobile object path planning algorithm of claim 1, wherein, In S5, according to the current behavior mode of the dog and the prediction path, the corresponding early warning mechanism is planned, and the early warning mechanism specifically includes: Straight running collision warning: Collision warning threshold is set, when the predicted path intersects with the forbidden area or dangerous area, according to the current speed Predict the time TTC to reach the area, judge whether to produce collision, push the page to display warning information, TTC = the nearest boundary distance of the dangerous area reached by the dog / current speed; When the predicted path is followed, the push page displays green; when the predicted path is followed with some deviation, the push page displays yellow; and when the predicted path is not followed, the push page displays red. Arc chasing out-of-control early warning: Set the minimum safe curvature radius and collision warning threshold, when the predicted path will pass through the dangerous area and the sharp turn may be out of control, judge whether it is out of control, push the page to display the warning information; The push page displays green if the path does not pass through a dangerous area and the radius of curvature is < a safety threshold; the push page displays yellow if the path passes through a dangerous area but the radius of curvature reaches 1.2 times the safety radius of curvature; the push page displays red if the path passes through a dangerous area and the radius of curvature is less than the minimum safety radius of curvature. Sudden turn-back early warning: Set the distance threshold between the position after turn-back and the dangerous area, judge whether the dangerous area is entered after turn-back, push the page to display the warning information; The path after turn-back has no collision risk, the page is pushed to display green; The path after turn-back is 2-5 meters away from the dangerous area, the page is pushed to display yellow; The path after turn-back is less than 2 meters away from the dangerous area or the predicted collision time is less than 3 seconds, the page is pushed to display red.

9. A device for real-time localization of a path of a mobile object, characterized in that The application of the real-time positioning mobile object path planning algorithm in any one of claims 1-8, the device specifically comprises: A collar, which is internally provided with a UWB positioning module, a GPS positioning module, a double-antenna positioning module, a judgment chip and a three-axis accelerometer; A control center end, which is internally provided with a deep learning model based on space-time feature fusion, a Kalman filter model and a canine path prediction algorithm; A push page, which is provided on a mobile end and is electrically connected or wirelessly connected with the control center end.

10. The real-time locatable mobile object path device of claim 9, wherein, The collar is also provided with a public address loudspeaker, which is wirelessly connected with the control center end and the mobile end, when the user receives the warning information of the push page on the mobile end, the password is transmitted to the public address loudspeaker of the collar to intercept the password of the dog.

Images