Action trajectory prediction method and device

Abstract

The application relates to the technical field of action track, in particular to an action track prediction method and device. The action track prediction method comprises the following steps: obtaining first environment characteristic information of a to-be-monitored target in a first environment, inputting the first environment characteristic information into a pre-trained action track prediction model, predicting at least one predicted action track and corresponding predicted environment characteristic information of each predicted action track in a surrounding environment according to the first environment characteristic information in a digital twin virtual environment, then matching second environment characteristic information with the predicted environment characteristic information, and determining at least one predicted action track as a target running track according to a matching degree. When the action track of the monitoring target is predicted, the change of the surrounding environment characteristic of the monitoring target is fully considered, the target running track is determined according to the change of the environment characteristic and the matching degree, and the prediction precision of the action track of the monitoring target is improved.

Description

Technical Field

[0001] This invention relates to the field of motion trajectory technology, and in particular to a method and apparatus for predicting motion trajectories. Background Technology

[0002] In recent years, with the development of science and technology and the continuous improvement of people's quality of life, more and more cities are committed to building smart cities in order to further improve the living environment of residents. Among them, intelligent transportation has become an indispensable part of smart cities.

[0003] The realization of intelligent transportation relies on the prediction of the movement trajectories of objects in the environment. Current methods for predicting object trajectories primarily abstract the object as a movable point, then predict its next position based on its historical location, and finally plot the object's trajectory at that next moment. However, it is well known that the trajectory of an object changes constantly with environmental variations. Predicting trajectories solely based on historical location information ignores the environmental characteristics of the object, reducing the valuable information available in the prediction process and resulting in low accuracy of the predicted trajectory. Summary of the Invention

[0004] This invention provides a method and apparatus for predicting movement trajectories, which addresses the technical deficiency of low accuracy in predicting the movement trajectory of a target object in existing technologies.

[0005] On one hand, the present invention provides a method for predicting movement trajectories, comprising:

[0006] Acquire the first environmental characteristic information of the target to be monitored in the first environment;

[0007] The first environmental feature information is input into a pre-trained action trajectory prediction model. The action trajectory prediction model is used to select the corresponding digital twin virtual environment based on the first environmental feature information, and predict at least one predicted action trajectory and the predicted environmental feature information corresponding to each predicted action trajectory in the surrounding environment based on the first environmental feature information.

[0008] Obtain the second environmental characteristic information of the target to be monitored in the second environment;

[0009] The second environmental feature information is matched with the predicted environmental feature information, and at least one predicted action trajectory is determined as the target running trajectory based on the degree of matching.

[0010] According to the method for predicting movement trajectories provided by the present invention, the movement trajectory prediction model is trained and obtained through the following method:

[0011] Acquire surveillance videos corresponding to multiple moving targets in the environment to be monitored;

[0012] Feature extraction is performed on the surveillance video to obtain the third environmental feature information corresponding to each moving target when it starts moving, which is used as the training input dataset;

[0013] Feature extraction is performed on the surveillance video to obtain the movement trajectory of each moving target and the fourth environmental feature information around the movement trajectory; the movement trajectory of each moving target and the fourth environmental feature information around the movement trajectory are used as the training output dataset.

[0014] The initial trajectory prediction model is trained using the training input dataset and the training output dataset to obtain the trained trajectory prediction model.

[0015] According to the movement trajectory prediction method provided by the present invention, the first environmental feature information, the second environmental feature information, the third environmental feature information and the fourth environmental feature information each include one or more of the following: location information, landmark information, road sign information and target turning information.

[0016] According to a method for predicting movement trajectories provided by the present invention, the step of matching the second environmental feature information with the predicted environmental feature information and determining at least one predicted movement trajectory as the target movement trajectory based on the degree of matching includes:

[0017] Calculate the matching degree between the second environmental feature information and the predicted environmental feature information, and determine one or more predicted action trajectories with the highest matching degree as the target running trajectory.

[0018] According to the method for predicting movement trajectories provided by the present invention, before acquiring the first environmental feature information of the target to be monitored in a first environment, the method further includes:

[0019] Retrieve surveillance video of the target to be monitored from the monitoring blockchain;

[0020] Feature extraction is performed on the surveillance video to obtain the first environmental feature information of the target to be monitored in the first environment.

[0021] On the other hand, the present invention also provides a motion trajectory prediction device, comprising:

[0022] The first feature extraction module is used to obtain the first environmental feature information of the target to be monitored in the first environment;

[0023] The motion trajectory prediction module is used to input the first environmental feature information into a pre-trained motion trajectory prediction model. The motion trajectory prediction model is used to select the corresponding digital twin virtual environment based on the first environmental feature information, and predict at least one predicted motion trajectory and the predicted environmental feature information corresponding to each predicted motion trajectory in the surrounding environment based on the first environmental feature information.

[0024] The second feature extraction module is used to obtain the second environmental feature information of the target to be monitored in the second environment;

[0025] The target trajectory determination module is used to match the second environmental feature information with the predicted environmental feature information, and determine at least one predicted action trajectory as the target trajectory based on the degree of matching.

[0026] According to the present invention, a movement trajectory prediction device is provided, wherein the movement trajectory prediction model is trained and obtained through the following method:

[0027] Acquire surveillance videos corresponding to multiple moving targets in the environment to be monitored;

[0028] Feature extraction is performed on the surveillance video to obtain the third environmental feature information corresponding to each moving target when it starts moving, which is used as the training input dataset;

[0029] Feature extraction is performed on the surveillance video to obtain the movement trajectory of each moving target and the fourth environmental feature information around the movement trajectory; the movement trajectory of each moving target and the fourth environmental feature information around the movement trajectory are used as the training output dataset.

[0030] The initial trajectory prediction model is trained using the training input dataset and the training output dataset to obtain the trained trajectory prediction model.

[0031] On the other hand, the present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the motion trajectory prediction method as described above.

[0032] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the motion trajectory prediction method as described above.

[0033] On the other hand, the present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the motion trajectory prediction method as described above.

[0034] The trajectory prediction method provided by this invention acquires first environmental feature information of the target in a first environment, inputs this first environmental feature information into a pre-trained trajectory prediction model, and predicts at least one predicted trajectory and corresponding predicted environmental feature information in the surrounding environment based on the first environmental feature information in a digital twin virtual environment. Then, it matches second environmental feature information with the predicted environmental feature information, and determines at least one predicted trajectory as the target's running trajectory based on the degree of matching. This invention fully considers the changes in the surrounding environmental features when predicting the target's trajectory, and determines the target's running trajectory based on these changes and the degree of matching, thus improving the accuracy of the target's trajectory prediction. Attached Figure Description

[0035] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0036] Figure 1 This is a flowchart illustrating the motion trajectory prediction method provided by the present invention;

[0037] Figure 2 This is a flowchart illustrating the training method for the motion trajectory prediction model provided by the present invention;

[0038] Figure 3 This is a schematic diagram of the movement trajectory prediction device provided by the present invention;

[0039] Figure 4 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0040] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0041] The following is combined with Figures 1-3 The technical solution of the present invention is described.

[0042] This invention is based on digital twin technology and uses the surrounding environmental features of a moving target as an important basis for prediction. By inputting the first environmental feature information of the previous moment into a trained motion trajectory prediction model, at least one predicted motion trajectory and the predicted environmental feature information corresponding to each predicted motion trajectory are predicted based on the first environmental feature information. Furthermore, the second environmental feature information of the current moment is obtained and matched with the predicted environmental feature information. Based on the degree of matching, at least one predicted motion trajectory can be determined as the target's running trajectory.

[0043] Example 1:

[0044] This embodiment provides a method for predicting movement trajectories, such as... Figure 1 The method for predicting movement trajectories includes:

[0045] Step 101: Obtain the first environmental feature information of the target to be monitored in the first environment.

[0046] It is understandable that as the target being monitored moves, its surrounding environment constantly changes, and therefore the corresponding environmental characteristics also change continuously. The "first environment" can be understood as the surrounding environment at the first moment. Typically, since the target is constantly moving, the environmental characteristics surrounding it change continuously at different times.

[0047] In this embodiment, the monitoring video of the target to be monitored is obtained from the monitoring blockchain. After the target to be monitored is determined in the monitoring video, feature extraction is performed on the monitoring video to obtain the first environmental feature information of the target to be monitored in the first environment.

[0048] Step 102: Input the first environmental feature information into the pre-trained action trajectory prediction model. The action trajectory prediction model is used to select the corresponding digital twin virtual environment based on the first environmental feature information, and predict at least one predicted action trajectory and the predicted environmental feature information of each predicted action trajectory in the surrounding environment based on the first environmental feature information.

[0049] The movement trajectory prediction model, which pre-acquires a large amount of monitoring data, can generate digital twin virtual environments corresponding to different areas. Therefore, it can determine which digital twin virtual environment the input first environmental feature information appears in, thus determining which digital twin virtual environment to select. The movement trajectory prediction model is trained on a large amount of data, and can predict movement trajectories related to the currently input environmental feature information. For example, if the currently input first environmental feature information includes "the entrance of the first primary school," the movement trajectory prediction model can predict several possible movement trajectories passing through the entrance of the first primary school based on "the entrance of the first primary school." Simultaneously, based on historical training data, it can acquire the predicted environmental feature information corresponding to the surrounding environment of each predicted movement trajectory. This predicted environmental feature information is the environmental feature information on the predicted movement trajectory; that is, as the movement trajectory changes, the environmental feature information in the surrounding environment also changes continuously. Therefore, it simultaneously acquires the constantly changing environmental feature information on each predicted movement trajectory.

[0050] Step 103: Obtain the second environmental characteristic information of the target to be monitored in the second environment.

[0051] Here, the first environment can be understood as the environment surrounding the target to be monitored at the previous moment, and the corresponding first environmental feature information is the environmental feature information surrounding the target to be monitored at the previous moment. The second environment can be understood as the environment surrounding the target to be monitored at the current moment, and the corresponding second environmental feature information is the environmental feature information surrounding the target to be monitored at the current moment.

[0052] Step 104: Match the second environmental feature information with the predicted environmental feature information, and determine at least one predicted action trajectory as the target running trajectory based on the degree of matching.

[0053] In this embodiment, a feature matching method is used to calculate the matching degree between the second environmental feature information and the predicted environmental feature information, and to determine one or more predicted action trajectories with the highest matching degree as the target trajectory. For example, the action trajectory prediction model outputs five predicted action trajectories and the corresponding predicted environmental feature information for each predicted action trajectory. After feature matching between the second environmental feature information and the predicted environmental feature information, the predicted action trajectory with the highest matching degree is obtained as the target trajectory. If feature matching between the second environmental feature information and the predicted environmental feature information yields two predicted action trajectories with similar matching degrees, both of these predicted action trajectories are used as target trajectories. At the next moment, environmental feature information around the target is collected and monitored, and further feature matching is performed with these two target trajectories to determine the predicted action trajectory with the higher matching degree as the final target trajectory.

[0054] The trajectory prediction method in this embodiment fully considers the impact of environmental feature information on prediction accuracy. It also utilizes knowledge of digital twins and neural networks. Based on digital twin technology, a virtual environment is generated. Deep learning technology within the neural network is used to predict possible trajectory movements within this virtual environment. Finally, through one or more environmental feature matches, the predicted trajectory with the highest matching degree is obtained as the final target trajectory. Testing has shown that the trajectory prediction method in this embodiment achieves high prediction accuracy.

[0055] In this embodiment, the first environmental feature information, the second environmental feature information, the third environmental feature information, and the fourth environmental feature information refer to the surrounding environmental feature information under different environments. The main features considered in the environmental feature information of this embodiment include one or more of the following: location information, landmark information, road sign information, and target turning information.

[0056] Location information refers to the position of the target to be monitored within the current larger environment. For example, if the environmental information of a city is virtualized into a 3D map using digital twin technology, then the target to be monitored represents its coordinates within the virtual 3D map. Landmark building information refers to buildings that can serve as road signs. Sometimes, a building itself occupies a large area, such as when it is adjacent to two roads. In this case, landmark building information can also be understood as information about a specific side of the building or its entrance, etc. As long as the current road information or orientation information can be determined based on the building's characteristic information, it is sufficient. Road sign information is an important feature, as it clearly indicates the name and direction of the current road, and therefore requires focused collection. Simultaneously, target turning information should also be collected. For example, if the target to be monitored is a vehicle, its turning information can be used; if the target to be monitored is a pedestrian, the turning of the pedestrian's head or body can be used.

[0057] In this embodiment, when acquiring environmental feature information of the target to be monitored in different environments, the monitoring video of the target is first obtained from the monitoring blockchain. Generally, each region has an independent monitoring blockchain, and video data collected by cameras is uploaded to this blockchain. Feature extraction is performed on the monitoring video to obtain the environmental feature information of the target in the current environment. Then, the environmental feature information is stored according to the location information of the path to facilitate subsequent environmental feature matching.

[0058] The movement trajectory prediction model can be trained for a single region or a single city. Furthermore, the choice of training dataset varies depending on the target being monitored. For example, to predict the movement trajectory of a specific type of worker, only the historical movement trajectories of that type of worker can be used as training data. Similarly, to predict the movement trajectory of sanitation workers, the historical movement trajectories of multiple sanitation workers within the current area, along with surrounding environmental features, can be collected as training data. If the movement trajectory of any individual within a region needs to be predicted, the historical movement trajectories of multiple pedestrians within that region over a given period, along with their corresponding environmental features, can be collected as the training dataset to train the movement trajectory prediction model. In one embodiment, when the target is a vehicle, the historical movement trajectories of multiple vehicles can also be used as training data to train the movement trajectory prediction model.

[0059] The trajectory prediction model in this embodiment is trained using the following method:

[0060] Step 201: Obtain monitoring videos of multiple moving targets in the environment to be monitored.

[0061] Step 202: Extract features from the surveillance video to obtain the third environmental feature information corresponding to each moving target when it starts moving, as the training input dataset.

[0062] It is understandable that, since each moving target is in a different position and at a different time when it begins to move, the corresponding third environmental feature information is also different.

[0063] Step 203: Extract features from the surveillance video to obtain the movement trajectory of each moving target and the fourth environmental feature information around the movement trajectory; use the movement trajectory of each moving target and the fourth environmental feature information around the movement trajectory as the training output dataset.

[0064] The trajectory of each moving target can be fitted by fitting the coordinates of the moving target during the movement.

[0065] Step 204: Train the initial trajectory prediction model using the training input dataset and the training output dataset to obtain the trained trajectory prediction model.

[0066] In this embodiment, a large amount of training input dataset is input into the initial movement trajectory prediction model. The movement trajectory corresponding to each moving target and the corresponding fourth environmental feature information are used as the training output dataset to train the movement trajectory prediction model. The training is completed when the prediction accuracy of the movement trajectory prediction model is verified and it is determined that the movement trajectory prediction model has reached the required level.

[0067] Example 2:

[0068] The movement trajectory prediction device provided by the present invention is described below. The movement trajectory prediction device described below can be referred to in correspondence with the movement trajectory prediction method described above.

[0069] A motion trajectory prediction device, such as Figure 3 The trajectory prediction device includes: a first feature extraction module 301, a trajectory prediction determination module 302, a second feature extraction module 303, and a target trajectory determination module 304.

[0070] The system comprises the following modules: a first feature extraction module 301, which acquires first environmental feature information of the target in a first environment; a predicted trajectory determination module 302, which inputs the first environmental feature information into a pre-trained trajectory prediction model, which selects a corresponding digital twin virtual environment based on the first environmental feature information and predicts at least one predicted trajectory and the corresponding predicted environmental feature information of each predicted trajectory in the surrounding environment; a second feature extraction module 303, which acquires second environmental feature information of the target in a second environment; and a target trajectory determination module 304, which matches the second environmental feature information with the predicted environmental feature information and determines at least one predicted trajectory as the target trajectory based on the degree of matching.

[0071] The training method of the motion trajectory prediction model in this embodiment is the same as that in Embodiment 1 above. In addition, the implementation methods of each functional module in this embodiment are the same as those in Embodiment 1 above, and will not be repeated here.

[0072] Figure 4 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 4As shown, the electronic device may include: a processor 410, a communication interface 420, a memory 430, and a communication bus 440, wherein the processor 410, the communication interface 420, and the memory 430 communicate with each other through the communication bus 440. The processor 410 can call logical instructions in the memory 430 to execute the motion trajectory prediction method provided by the above methods. The method includes: acquiring first environmental feature information of the target to be monitored in a first environment; inputting the first environmental feature information into a pre-trained motion trajectory prediction model, the motion trajectory prediction model being used to select a corresponding digital twin virtual environment based on the first environmental feature information, and predicting at least one predicted motion trajectory and the predicted environmental feature information corresponding to each predicted motion trajectory in the surrounding environment based on the first environmental feature information; acquiring second environmental feature information of the target to be monitored in a second environment; matching the second environmental feature information with the predicted environmental feature information, and determining at least one predicted motion trajectory as the target's running trajectory based on the degree of matching.

[0073] Furthermore, the logical instructions in the aforementioned memory 430 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, essentially, or the part that contributes to the prior art, or a part of the 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 cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0074] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the motion trajectory prediction method provided by the above methods. The method includes: acquiring first environmental feature information of the target to be monitored in a first environment; inputting the first environmental feature information into a pre-trained motion trajectory prediction model, wherein the motion trajectory prediction model is used to select a corresponding digital twin virtual environment based on the first environmental feature information, and predict at least one predicted motion trajectory and the predicted environmental feature information corresponding to each predicted motion trajectory in the surrounding environment based on the first environmental feature information; acquiring second environmental feature information of the target to be monitored in a second environment; matching the second environmental feature information with the predicted environmental feature information, and determining at least one predicted motion trajectory as the target's running trajectory based on the degree of matching.

[0075] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the motion trajectory prediction method provided by the above methods. The method includes: acquiring first environmental feature information of the target to be monitored in a first environment; inputting the first environmental feature information into a pre-trained motion trajectory prediction model, wherein the motion trajectory prediction model is used to select a corresponding digital twin virtual environment based on the first environmental feature information, and predict at least one predicted motion trajectory and the predicted environmental feature information corresponding to each predicted motion trajectory in the surrounding environment based on the first environmental feature information; acquiring second environmental feature information of the target to be monitored in a second environment; matching the second environmental feature information with the predicted environmental feature information, and determining at least one predicted motion trajectory as the target's running trajectory based on the degree of matching.

[0076] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0077] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0078] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for predicting movement trajectories, characterized in that, The method comprises the following steps: obtaining first environment feature information of a target to be monitored in a first environment; inputting the first environment feature information into a pre-trained action trajectory prediction model, the action trajectory prediction model being configured to select a corresponding digital twin virtual environment according to the first environment feature information and to predict at least one predicted action trajectory in the digital twin virtual environment according to the first environment feature information and corresponding predicted environment feature information of each predicted action trajectory in the surrounding environment; obtaining second environment feature information of the target to be monitored in a second environment; matching the second environment feature information with the predicted environment feature information and determining at least one predicted action trajectory as a target running trajectory according to the matching degree; the action trajectory prediction model is trained by the following method: obtaining monitoring videos corresponding to multiple mobile targets in a monitoring environment; extracting features from the monitoring videos to obtain third environment feature information corresponding to each mobile target when the mobile target starts moving as training input data set; extracting features from the monitoring videos to obtain an action trajectory corresponding to each mobile target and fourth environment feature information around the action trajectory; using the action trajectory corresponding to each mobile target and the fourth environment feature information around the action trajectory as training output data set; training an initial action trajectory prediction model using the training input data set and the training output data set to obtain the trained action trajectory prediction model.

2. The action trajectory prediction method of claim 1, wherein, The first environment feature information, the second environment feature information, the third environment feature information and the fourth environment feature information each comprise one or more of position information, landmark building information, road sign information and target turning information. 3.The action trajectory prediction method of claim 1, wherein, The matching of the second environment feature information with the predicted environment feature information and the determination of at least one predicted action trajectory as a target running trajectory according to the matching degree comprise: calculating the matching degree between the second environment feature information and the predicted environment feature information and determining one or more predicted action trajectories with high ranking matching degree as target running trajectories.

4. The action trajectory prediction method of claim 1, wherein, Before obtaining the first environment feature information of the target to be monitored in the first environment, the method further comprises: obtaining a monitoring video of the target to be monitored from a monitoring blockchain; extracting features from the monitoring video to obtain the first environment feature information of the target to be monitored in the first environment.

5. An action trajectory prediction apparatus characterized by comprising: The method comprises the following steps: a first feature extraction module for obtaining first environment feature information of a target to be monitored in a first environment; a predicted action trajectory determination module for inputting the first environment feature information into a pre-trained action trajectory prediction model, the action trajectory prediction model being configured to select a corresponding digital twin virtual environment according to the first environment feature information and to predict at least one predicted action trajectory in the digital twin virtual environment according to the first environment feature information and corresponding predicted environment feature information of each predicted action trajectory in the surrounding environment; a second feature extraction module for obtaining second environment feature information of the target to be monitored in a second environment; The target running track determination module is configured to match the second environment feature information with the predicted environment feature information, and determine at least one predicted action track as a target running track according to a matching degree. The action track prediction model is obtained by training through the following method: Obtain monitoring videos corresponding to a plurality of moving targets in a to-be-monitored environment; Extract features of the monitoring videos to obtain third environment feature information corresponding to each moving target when the moving target starts to move as training input data sets; Extract features of the monitoring videos to obtain an action track corresponding to each moving target and fourth environment feature information around the action track; Take the action track corresponding to each moving target and the fourth environment feature information around the action track as training output data sets; Train an initial action track prediction model using the training input data sets and the training output data sets to obtain the trained action track prediction model.

6. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor implements the action track prediction method according to any one of claims 1 to 4 when executing the program.

7. A non-transitory computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program implements the action track prediction method according to any one of claims 1 to 4 when executed by the processor.

8. A computer program product comprising a computer program, characterized in that, The computer program implements the action track prediction method according to any one of claims 1 to 4 when executed by the processor. The computer program implements the action track prediction method according to any one of claims 1 to 4 when executed by the processor.

Images