Trajectory prediction method and device, terminal equipment and readable storage medium

Abstract

The application is suitable for the technical field of intelligent control, and provides a trajectory prediction method and device, terminal equipment and a readable storage medium. The trajectory prediction method specifically comprises: obtaining environment data of an environment in which a target object is located, the environment comprising at least one avoidance target related to the target object; determining at least one motion trajectory of each avoidance target from a current time to a future time and a trajectory confidence of each motion trajectory according to the environment data; adjusting the trajectory confidence of each motion trajectory according to map information of a current region to obtain an adjusted trajectory confidence, the current region containing a region passed through by each motion trajectory, and the adjusted trajectory confidence being used for obstacle avoidance of the target object to each avoidance target. The embodiment of the application can improve the safety of the target object when driving automatically.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of intelligent control, and particularly relates to a trajectory prediction method and device, a terminal device and a readable storage medium. BACKGROUND

[0002] The trajectory prediction algorithm plays a very important role in automatic driving technology. An excellent trajectory prediction algorithm can reduce the pressure of the motion planning module and make the driving process more stable and smooth. In some specific environments, the motion intention of the avoidance target is relatively subjective. For example, the unmanned vehicle delivery in a closed park usually drives on a small road with many pedestrians and bicycles. The motion intention of the pedestrians and bicycles has strong subjectivity, and the vehicles in the closed park also do not usually drive according to the predetermined rules. The current trajectory prediction method cannot well cope with the suddenness of the avoidance target, resulting in misjudgment when the avoidance target does not drive normally, and further causing traffic accidents. SUMMARY

[0003] The embodiments of the present application provide a trajectory prediction method, device, terminal device and readable storage medium, which can improve the safety of automatic driving.

[0004] The first aspect of the embodiments of the present application provides a trajectory prediction method, comprising: obtaining environment data of an environment in which a target object is located, the environment comprising at least one avoidance target related to the target object; determining at least one motion trajectory of each avoidance target from a current time to a future time and a trajectory confidence of each motion trajectory according to the environment data; adjusting the trajectory confidence of each motion trajectory according to map information of a current area to obtain an adjusted trajectory confidence, the current area containing an area passed through by each motion trajectory, and the adjusted trajectory confidence being used for obstacle avoidance of the target object to each avoidance target.

[0005] In some embodiments of the present application, the determining at least one motion trajectory of each avoidance target from a current time to a future time according to the environment data comprises: determining a to-be-matched position of the avoidance target in a plurality of motion times within a preset time length according to the environment data, the preset time length being a time period from the current time to the future time; obtaining motion data of each avoidance target; matching the to-be-matched position for each avoidance target according to the motion data to obtain a predicted position of each avoidance target at each motion time, wherein the predicted position at each motion time is one or more; and connecting the predicted positions of each avoidance target in time sequence to obtain at least one motion trajectory within the preset time length.

[0006] In some embodiments of the present application, the matching of the to-be-matched position for each of the avoidance targets according to the motion data to obtain the predicted position of the avoidance target at each of the motion time points comprises: determining a position offset of each of the avoidance targets at each of the motion time points relative to a previous motion time point according to the motion data; and determining, among the plurality of to-be-matched positions, a position that satisfies a position offset relative to a second time point between the first position at a first time point as the predicted position at the second time point, wherein the second time point is any one of the motion time points within the preset time length, and the first time point is a previous motion time point of the second time point.

[0007] In some embodiments of the present application, the determining of the to-be-matched positions of the avoidance targets at the plurality of motion time points within the preset time length according to the environment data comprises: determining, according to the motion data, the plurality of motion time points within the preset time length, the to-be-matched positions of the avoidance targets at each of the motion time points, and the reference type of the avoidance target at each of the to-be-matched positions; and determining, among the plurality of to-be-matched positions, a to-be-matched position that satisfies a position offset relative to the second time point in a distance from the first position as the predicted position at the second time point, wherein the second time point is any one of the motion time points within the preset time length, and the first time point is a previous motion time point of the second time point.

[0008] In some embodiments of the present application, the determining of the trajectory confidence of each of the motion trajectories comprises: determining a position confidence of the avoidance target at each of the predicted positions; and determining the trajectory confidence corresponding to the motion trajectory according to the position confidence of the predicted position on the same motion trajectory.

[0009] In some embodiments of the present application, the map information records an impassable area of the avoidance target; and the adjusting of the trajectory confidence of each of the motion trajectories according to the map information of the current area to obtain an adjusted trajectory confidence comprises: adjusting the position confidence of the predicted position located in the impassable area; and determining the adjusted trajectory confidence corresponding to the motion trajectory according to the adjusted position confidence.

[0010] In some embodiments of the present application, the determining, according to the environment data, at least one motion trajectory of each of the avoidance targets from a current time to a future time and a trajectory confidence of each of the motion trajectories comprises: inputting the environment data into a target neural network to obtain at least one motion trajectory of each of the avoidance targets from the current time to the future time and the trajectory confidence of each of the motion trajectories output by the target neural network, wherein the target neural network is trained by sample environment data containing multiple types of the avoidance targets, and sizes of at least part of the types of the avoidance targets are different.

[0011] The second aspect of the embodiments of the present application provides a trajectory prediction device, comprising: an acquisition unit configured to acquire environment data of an environment in which a target object is located, the environment comprising at least one avoidance target related to the target object; a determination unit configured to determine, according to the environment data, at least one motion trajectory of each of the avoidance targets from a current time to a future time and a trajectory confidence of each of the motion trajectories; and an adjustment unit configured to adjust the trajectory confidence of each of the motion trajectories according to map information of a current region to obtain an adjusted trajectory confidence, the current region containing a region passed through by each of the motion trajectories, and the adjusted trajectory confidence being used by the target object to avoid obstacles of each of the avoidance targets.

[0012] The third aspect of the embodiments of the present application provides a terminal device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the steps of the trajectory prediction method when executing the computer program.

[0013] The fourth aspect of the embodiments of the present application provides a computer readable storage medium, the computer readable storage medium stores a computer program, and the computer program is executed by a processor to implement the steps of the trajectory prediction method.

[0014] The fifth aspect of the embodiments of the present application provides a computer program product, when the computer program product is executed on a terminal device, the terminal device executes the trajectory prediction method of any one of the first aspect.

[0015] In the embodiments of the present application, the environment data of the environment where the target object is located is acquired, and according to the environment data, at least one motion trajectory of each avoidance target from the current time to the future time in the environment is determined, and the trajectory confidence of each motion trajectory is determined. Then, according to the map information of the current area, the trajectory confidence of each motion trajectory is adjusted. Based on the adjusted trajectory confidence, the target object can avoid obstacles for each avoidance target, so that the target object effectively refers to each possible trajectory and the probability of occurrence of each motion trajectory when avoiding obstacles, rather than directly determining a unique driving trajectory of each avoidance target from the map. This avoids the occurrence of traffic accidents due to the omission of the motion trajectory of the avoidance target when driving abnormally, and improves the safety of automatic driving. BRIEF DESCRIPTION OF DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the embodiments or prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0017] Figure 1 is a schematic diagram of an implementation process of a trajectory prediction method provided by an embodiment of the present application;

[0018] Figure 2 is a schematic diagram of a specific implementation process of trajectory prediction provided by an embodiment of the present application;

[0019] Figure 3 is a schematic diagram of the structure of a trajectory prediction device provided by an embodiment of the present application;

[0020] Figure 4 is a schematic diagram of the structure of a terminal device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0021] 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 the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application, and are not used to limit the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0022] In order to illustrate the technical solutions of the present application, the following will be described through specific embodiments.

[0023] Figure 1An implementation flowchart of a trajectory prediction method provided by an embodiment of the present application is shown, which can be applied to a terminal device and can be suitable for a situation where the safety of automatic driving needs to be improved. The terminal device can be a computer, a mobile phone, an automatic driving vehicle, a robot, or other electronic devices.

[0024] Specifically, the trajectory prediction method can include the following steps S101-S103.

[0025] In step S101, environment data of an environment where a target object is located is obtained.

[0026] In an embodiment of the present application, the environment where the target object is located can include at least one evasive target related to the target object. The target object can be a robot, a vehicle, an unmanned vehicle, or other devices with automatic driving capability. The evasive target is an obstacle that the target object needs to avoid during driving, including but not limited to pedestrians, vehicles, pets, robots, etc.

[0027] In order to realize trajectory prediction, the terminal device needs to obtain the environment data of the environment where the target object is located. The environment data can include environment images and / or environment point cloud data. In some embodiments, one or more sensors can be arranged on the target object, and the terminal device can obtain the environment data collected by the sensors of the target object. In other embodiments, if the target object is driving in a preset area, the terminal device can obtain the environment data collected by the sensors arranged in the preset area. For example, if the target object is driving in a closed park, the terminal device can obtain the environment images taken by the cameras in the closed park. In other embodiments, the terminal device can obtain the motion information reported by the evasive target and the target object to obtain the environment data. Specifically, the evasive target and the target object can be provided with a motion monitoring module, which can be used to detect the motion information of itself, such as collecting the driving speed and driving acceleration of itself through a speed sensor and an acceleration sensor, and obtaining the current location through Beidou positioning or GPS positioning, etc.

[0028] It should be noted that the environment data can include current time and historical time environment data. The current time is the starting time of trajectory prediction, and the historical time refers to a sampling time before the current time, for example, the previous sampling time of the current time.

[0029] In step S102, at least one motion trajectory of each evasive target from the current time to a future time is determined according to the environment data, and a trajectory confidence of each motion trajectory is determined.

[0030] In the embodiments of the present application, through analysis of the environmental data, the motion information of the avoidance target and the road information around the position where the avoidance target is located can be obtained. The road information includes but is not limited to lane line identification, barrier position, step position, etc. According to the motion information and the road information around the position where the avoidance target is located, the terminal device can perform trajectory prediction on each avoidance target, determine at least one motion trajectory of each avoidance target from the current time to the future time, and a trajectory confidence of each motion trajectory. The trajectory confidence is the probability of the avoidance target driving according to the motion trajectory.

[0031] In step S103, the trajectory confidence of each motion trajectory is adjusted according to the map information of the current area to obtain an adjusted trajectory confidence.

[0032] The current area contains the area passed by each motion trajectory, and correspondingly, the map information can record the passable area and the impassable area of the avoidance target in the current area.

[0033] It should be understood that the passable area and the impassable area of different types of avoidance targets are different. For example, for a vehicle, a sidewalk should be an impassable area. At the same time, if the driving direction allowed by the lane is different from the driving direction of the vehicle (i.e., the vehicle is driving in reverse), the lane also belongs to the impassable area. However, in some special scenarios, such as in a closed park, the vehicle often does not drive according to the established rules, for example, it can drive to the sidewalk to park, or the vehicle drives in reverse. The conventional trajectory prediction generally only determines a unique path in the passable area, and the ability to respond to these sudden behaviors is poor. In the embodiments of the present application, each possible motion trajectory is retained, and the trajectory confidence of each motion trajectory is adjusted according to the map information of the current area, that is, the probability of occurrence of each trajectory is adjusted.

[0034] In the embodiments of the present application, the adjusted trajectory confidence can be used for obstacle avoidance of the target object, so that the target object can avoid all avoidance targets when automatically driving.

[0035] Specifically, based on the adjusted trajectory confidence, the terminal device can sort all motion trajectories of the same avoidance target to obtain a trajectory sorting result corresponding to each avoidance target, and then use the trajectory sorting result to make an obstacle avoidance decision.

[0036] In some embodiments, the terminal device can refer to the trajectory sorting result of each evasive target, and take the top N trajectories with the highest trajectory confidence of each evasive target as the reference for obstacle avoidance. N is a positive integer greater than or equal to 1, and the specific value can be adjusted according to the total number of evasive targets. If the total number of evasive targets is less than a preset number value, it means that there are fewer traffic participants, and in this case, N can be larger, so as to refer to more occurring trajectories. If the total number of evasive targets is greater than or equal to the preset number value, it means that there are more traffic participants, and in this case, N can be smaller to avoid too low decision efficiency. In some other embodiments, the terminal device can also retain at least one trajectory with a trajectory confidence greater than or equal to a confidence threshold as the reference for obstacle avoidance for each obstacle object. The preset number value and the confidence threshold can be adjusted according to actual conditions.

[0037] In the embodiments of the present application, the environment data of the environment where the target object is located is obtained, and according to the environment data, at least one trajectory of each evasive target from the current time to the future time is determined, as well as the trajectory confidence of each trajectory. Then, according to the map information of the current area, the trajectory confidence of each trajectory is adjusted. Based on the adjusted trajectory confidence, the target object can avoid obstacles for each evasive target, so that the target object effectively refers to each possible trajectory and the probability of occurrence of each trajectory when avoiding obstacles, rather than directly determining a unique driving trajectory for each evasive target from the map, avoiding the occurrence of traffic accidents due to the omission of the motion trajectory of the evasive target when driving abnormally, and improving the safety of autonomous driving.

[0038] The process of trajectory prediction will be described in detail below.

[0039] Please refer to Figure 2 In some embodiments of the present application, the terminal device can determine at least one trajectory by the following steps S201 to S204.

[0040] Step S201: According to the environment data, the positions of the evasive targets at multiple motion times within a preset time period are determined.

[0041] The preset time period is the time period in which the trajectory occurs, that is, a time period with the current time as the starting point and the future time as the ending point. The specific value of the time period can be selected according to actual conditions, for example, it can be 3s, 5s, etc. The preset time period can include multiple motion times, for example, every 1s is a motion time. By predicting the position of each motion time within the preset time period, the trajectory of the evasive target within the preset time period can be obtained.

[0042] Specifically, the terminal device can use the environment information of the current moment and the historical moment as the input of the neural network, and the output of the neural network is the position (i.e. the to-be-matched position) of the possible avoidance target at multiple motion moments within a preset time length, and the reference type corresponding to the to-be-matched position. Wherein, the number of to-be-matched positions at each moment can be equal or unequal, and for all the to-be-matched positions, the present application will be reserved, so that the to-be-matched positions can include positions located in the impassable area of the avoidance target.

[0043] Step S202, obtaining the motion data of each avoidance target.

[0044] Step S203, according to the motion data, matching the to-be-matched position for each avoidance target, and obtaining the predicted position of the avoidance target at each motion moment.

[0045] In the embodiments of the present application, the to-be-matched position is the position of the appearing avoidance target, but it is not clear which avoidance target each to-be-matched position belongs to, therefore, the motion data of each avoidance target needs to be obtained, and the to-be-matched position is matched for each avoidance target according to the motion data.

[0046] Wherein, the motion data can be the speed, position and other data of the historical moment and the current moment reported by the avoidance target, or the motion data determined through the environment data.

[0047] In the embodiments of the present application, according to the motion data, the terminal device can determine the position offset relative to the previous moment at each motion moment. Specifically, the position offset between the position of each avoidance target in each motion moment and the position of the avoidance target in the previous moment can be predicted in the neural network described in step S201. In order to simplify the prediction, the position offset can be the offset of the two positions in the Bird's Eye View (bev) perspective, that is, only the offset of the two positions in the x and y of the target coordinate system needs to be predicted, and the offset of z does not need to be predicted. The target coordinate system can refer to the radar coordinate system or the camera coordinate system, and the z axis of the target coordinate system is generally perpendicular to the ground.

[0048] Correspondingly, in the plurality of to-be-matched positions, a position that satisfies the position offset amount related to the second time moment between the first position at the first time moment can be determined as the predicted position at the second time moment. The second time moment is any one movement time moment within the preset time length, and the first time moment is the previous movement time moment of the second time moment. When the second time moment is the current time moment, the first time moment is the historical time moment, and the first position is the historical position at the historical time moment. When the second time moment is the future time moment, the first time moment is the previous movement time moment of the second time moment within the preset time length, and the first position is the predicted position at the previous movement time moment. That is, if the to-be-matched position (the second position) at the t+1 time moment is A, and the position offset amount is (x1, y1), and the distance between the first position B at which the target is just evaded at the t time moment and the to-be-matched position A satisfies the position offset amount (x1, y1), it is considered that the two positions are matched, that is, the to-be-matched position A is the position reached by the evading target at the t+1 time moment after reaching the first position B at the t time moment. Then, starting from the historical time moment, the to-be-matched positions at each movement time moment within the preset time length can be matched in turn, and the matched position is the predicted position.

[0049] It should be noted that, in the matching, the same to-be-matched position can have a competitive relationship, that is, the same to-be-matched position can be matched by different evading targets at the same time, and meanwhile, the same evading target can have a plurality of matched to-be-matched positions at the same time moment. In this regard, the present application can use a greedy matching algorithm to retain all possibilities, and finally, the predicted position at each movement time moment can be one or more.

[0050] In addition, the foregoing neural network can output a reference type corresponding to each to-be-matched position at each movement time moment, that is, the evading target appearing in the to-be-matched position belongs to a pedestrian, a bicycle, a car, or other types. In order to ensure the reliability of the movement trajectory, the terminal device can take the to-be-matched position that satisfies the position offset amount related to the second time moment and has the same type as the evading target as the predicted position at the second time moment.

[0051] In step S204, the predicted positions of each evading target are connected in time sequence to obtain at least one movement trajectory within the preset time length.

[0052] After matching the positions, the terminal device can obtain the predicted positions of each avoidance target at each movement time, and then connect the predicted positions in time sequence to obtain the movement trajectory of each avoidance target within a preset time length. It should be understood that, since the predicted position at each movement time can be one or more, for a certain avoidance target, if the predicted position at each movement time is one, the obtained movement trajectory after connection is unique; if there are multiple predicted positions at any movement time, the obtained movement trajectory after connection can be multiple.

[0053] Correspondingly, for the aforementioned trajectory confidence, the terminal device can determine the position confidence of the avoidance target at each predicted position, and determine the trajectory confidence of the corresponding movement trajectory according to the position confidence of the predicted positions on the same movement trajectory.

[0054] For example, the movement trajectory C is connected by the predicted positions D, E and F in sequence, and the trajectory confidence of the movement trajectory C can be determined by the position confidences of the predicted positions D, E and F, for example, can be obtained by adding or weighted adding the position confidences of the predicted positions D, E and F. The confidence of each predicted position can be a preset confidence value, or can be output by the aforementioned neural network. The aforementioned neural network can refer to the environmental information to assign a position confidence to each predicted position.

[0055] Correspondingly, the terminal device can adjust the position confidence of the predicted position located in the impassable area, to determine the adjusted trajectory confidence of the corresponding movement trajectory according to the adjusted position confidence. For example, a preset adjustment parameter p can be obtained, and for each predicted position, if it is located in the impassable area, the position confidence can be subtracted by p, and the trajectory confidence of the movement trajectory containing the predicted position is recalculated. Wherein, 0<p<1, the specific value can be adjusted according to the actual situation.

[0056] Therefore, the predicted movement trajectory is optimized by using the map information, and the corresponding movement trajectory can be retained for some abnormal vehicles, and the corresponding trajectory confidence is output. Then, for some seemingly impossible movement trajectories (such as driving out of the lane, reversing and other abnormal driving behaviors), the movement trajectories can be retained for the subsequent decision module, and thus the occurrence of traffic accidents can be avoided.

[0057] In actual applications, considering the diversity of types of avoidance targets, the environment data can be input into neural networks corresponding to different types of avoidance targets, that is, the neural networks correspond one-to-one to the types of avoidance targets, but the detection efficiency is low. Therefore, in some embodiments of the present application, the environment data can be input into a target neural network to obtain at least one motion trajectory of each avoidance target from the current time to the future time output by the target neural network, and the trajectory confidence of each motion trajectory. The target neural network is trained by sample environment data containing multiple types of avoidance targets, wherein at least part of the types of avoidance targets have different sizes. That is, the target neural network can simultaneously detect avoidance targets of different types and different sizes, improving the detection efficiency.

[0058] Specifically, the target neural network used in the present application can be a Bi-FPN network structure. The Bi-FPN structure considers the fusion of high semantic information and low semantic information, allowing the target neural network to combine high and low semantic features during operation and output reliable results. Specifically, after extracting the high semantic features, the low semantic features are connected through deconvolution, thereby retaining more detailed information and reducing the miss detection rate of small targets such as pedestrians and bicycles.

[0059] Since the common perception module process is target detection, target tracking, and trajectory prediction in sequence, the results of trajectory prediction are calculated based on the results of target detection and tracking. If the target detector and the target tracker cannot achieve qualified results, the performance of the trajectory prediction module will be severely affected. The fusion of high semantic information and low semantic information used in the present application can output the predicted positions at the current time and the future time in parallel, thereby avoiding the interference of the obstacle trajectory prediction task by the preceding tasks. At the same time, the target neural network can capture more original environment data, making the prediction of the motion trajectory more accurate.

[0060] It should be noted that, for the foregoing method embodiments, in order to simply describe, they are all expressed as a series of action combinations, but those skilled in the art should know that the present application is not limited by the action sequence described, because according to the present application, certain steps can be performed in other sequences.

[0061] As Figure 3 Fig. 1 shows a structure schematic diagram of a trajectory prediction device 300 provided by an embodiment of the present application, which is configured on a terminal device.

[0062] Specifically, the trajectory prediction device 300 can include:

[0063] The acquisition unit 301 is configured to acquire environment data of an environment in which a target object is located, the environment including at least one avoidance target related to the target object;

[0064] The determining unit 302 is configured to determine at least one motion trajectory of each of the avoidance targets from the current time to a future time and a trajectory confidence of each of the motion trajectories according to the environment data.

[0065] The adjusting unit 303 is configured to adjust the trajectory confidence of each of the motion trajectories according to map information of a current region to obtain an adjusted trajectory confidence, the current region including a region passed through by each of the motion trajectories, and the adjusted trajectory confidence being used for the target object to avoid obstacles for each of the avoidance targets.

[0066] In some embodiments of the present application, the determining unit 302 can be specifically configured to: determine a to-be-matched position of the avoidance target at a plurality of motion times within a preset time length according to the environment data, the preset time length being a time period from the current time to the future time; obtain motion data of each of the avoidance targets; perform matching of the to-be-matched position for each of the avoidance targets according to the motion data to obtain a predicted position of each of the avoidance targets at each of the motion times, wherein the predicted position at each of the motion times is one or more; and connect the predicted positions of each of the avoidance targets in time sequence to obtain at least one motion trajectory within the preset time length.

[0067] In some embodiments of the present application, the determining unit 302 can be specifically configured to: determine a position offset of each of the avoidance targets at each of the motion times relative to a previous motion time according to the motion data; and determine, among a plurality of the to-be-matched positions, a position with an offset between a first position at a first time satisfying a position offset related to a second time as the predicted position at the second time, wherein the second time is any one of the motion times within the preset time length, and the first time is a previous motion time of the second time.

[0068] In some embodiments of the present application, the determining unit 302 can be specifically configured to: determine a plurality of motion times within a preset time length and a to-be-matched position of the avoidance target at each of the motion times, and a reference type of the avoidance target at each of the to-be-matched positions according to the motion data; and determine, as the predicted position at the second time, the to-be-matched position with a distance from the first position satisfying a position offset related to the second time and the reference type being the same as a type of the avoidance target.

[0069] In some embodiments of this application, the determining unit 302 may be specifically used to: determine the position confidence of the avoidance target at each of the predicted positions; and determine the trajectory confidence of the corresponding motion trajectory based on the position confidence of the predicted positions on the same motion trajectory.

[0070] In some embodiments of this application, the map information records impassable areas for avoiding targets; the adjustment unit 303 can be specifically used to: adjust the position confidence of the predicted position located in the impassable area; and determine the adjusted trajectory confidence corresponding to the motion trajectory based on the adjusted position confidence.

[0071] In some embodiments of this application, the determining unit 302 described above may be specifically used to: input the environmental data into the target neural network, obtain at least one motion trajectory of each avoidance target output by the target neural network from the current time to a future time, and the trajectory confidence of each motion trajectory, wherein the target neural network is trained by sample environmental data containing multiple types of avoidance targets, and at least some types of avoidance targets have different sizes.

[0072] It should be noted that, for the sake of convenience and brevity, the specific working process of the trajectory prediction device 300 described above can be found in the following reference: Figures 1 to 2 The corresponding process of the method will not be described in detail here.

[0073] like Figure 4 The diagram shown is a schematic of a terminal device provided in an embodiment of this application. The terminal device 4 may include: a processor 40, a memory 41, and a computer program 42 stored in the memory 41 and executable on the processor 40, such as a trajectory prediction program. When the processor 40 executes the computer program 42, it implements the steps in the various trajectory prediction method embodiments described above, for example... Figure 1 The steps S101 to S103 are shown. Alternatively, when the processor 40 executes the computer program 42, it implements the functions of each module / unit in the above-described device embodiments, for example... Figure 3 The shown units are: acquisition unit 301, determination unit 302, and adjustment unit 303.

[0074] The computer program can be divided into one or more modules / units, which are stored in the memory 41 and executed by the processor 40 to complete this application. The one or more modules / units can be a series of computer program instruction segments capable of performing specific functions, which describe the execution process of the computer program in the terminal device.

[0075] For example, the computer program can be divided into: an acquisition unit, a determination unit, and an adjustment unit.

[0076] The specific functions of each unit are as follows: the acquisition unit is configured to acquire environment data of an environment in which a target object is located, the environment including at least one avoidance target related to the target object; the determination unit is configured to determine, according to the environment data, at least one motion trajectory of each avoidance target from a current time to a future time, and a trajectory confidence of each motion trajectory; and the adjustment unit is configured to adjust the trajectory confidence of each motion trajectory according to map information of a current region, to obtain an adjusted trajectory confidence, the current region containing a region through which each motion trajectory passes, and the adjusted trajectory confidence being used for the target object to perform obstacle avoidance on each avoidance target.

[0077] The terminal device can include, but is not limited to, a processor 40 and a memory 41. Those skilled in the art can understand that, Figure 4 The terminal device is merely an example and does not constitute a limitation on the terminal device, and can include more or fewer components than those shown, or combine certain components, or different components, for example, the terminal device can also include an input / output device, a network access device, a bus, and the like.

[0078] The processor 40 can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gates or transistor logic components, discrete hardware components, or the like. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor.

[0079] The memory 41 can be an internal storage unit of the terminal device, such as a hard disk or a memory of the terminal device. The memory 41 can also be an external storage device of the terminal device, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the terminal device. Further, the memory 41 can also include both the internal storage unit and the external storage device of the terminal device. The memory 41 is used to store the computer program and other programs and data required by the terminal device. The memory 41 can also be used to temporarily store data that has been output or will be output.

[0080] It should be noted that, for the convenience and brevity of description, the structure of the terminal device can also refer to the specific description of the structure in the method embodiments, which will not be repeated here.

[0081] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the division of the above functional units and modules is exemplified, and in actual application, the above functions can be completed by different functional units and modules according to needs, that is, the internal structure of the device is divided into different functional units or modules to complete all or part of the functions described above. Each functional unit and module in the embodiment can be integrated in one processing unit, or each unit can exist physically, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of software functional unit. In addition, the specific name of each functional unit and module is only for easy distinction, and does not limit the protection scope of the present application. The specific working process of the unit and module in the system can refer to the corresponding process in the foregoing method embodiments, which will not be repeated here.

[0082] In the above embodiments, the description of each embodiment has its own emphasis, and the parts not described or recorded in detail in a certain embodiment can be referred to the related description of other embodiments.

[0083] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be realized in electronic hardware or a combination of computer software and electronic hardware. Whether the functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0084] In the embodiments of the present application, it should be understood that the disclosed apparatus / terminal device and method can be implemented in other manners. For example, the embodiments of the apparatus / terminal device described above are merely schematic, and the division of the modules or units is merely logical function division, and there can be another division manner in actual implementation. For example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections between the units can be indirect couplings or communication connections through some interfaces, devices or units, and can be electrical, mechanical or in other forms.

[0085] The units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, i.e., can be located in one place, or can be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the embodiments.

[0086] In addition, each functional unit in the various embodiments of the present application can be integrated in one processing unit, or each unit can be physically present separately, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of a software functional unit.

[0087] The integrated module / unit, if realized in the form of a software functional unit and sold or used as an independent product, can be stored in a computer readable storage medium. Based on such understanding, all or part of the flow of the above-mentioned embodiment methods can also be completed by a computer program instructing related hardware, and the computer program can be stored in a computer readable storage medium. When the processor executes the computer program, the steps of each method embodiment described above can be implemented. The computer program includes computer program code, which can be in the form of source code, object code, executable file or some intermediate form. The computer readable medium can include any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal and software distribution medium, etc. It should be noted that the computer readable medium can include appropriate contents according to the requirements of legislation and patent practice in the jurisdiction, for example, in some jurisdictions, according to legislation and patent practice, the computer readable medium does not include electrical carrier signals and telecommunication signals.

[0088] The above-described embodiments are only used to illustrate the technical solutions of the present application, but not limit the present application; although the present application is described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application.

Claims

1. A trajectory prediction method characterized by, The method comprises: acquiring environment data of an environment in which a target object is located, the environment including at least one avoidance target related to the target object; determining, according to the environment data, at least one motion trajectory of each avoidance target from a current time to a future time, and a trajectory confidence of each motion trajectory; wherein the determination of the at least one motion trajectory of each avoidance target from the current time to the future time according to the environment data comprises: determining, according to the environment data, a plurality of matching positions of the avoidance target in a preset time period, the preset time period being a time period from the current time to the future time; acquiring motion data of each avoidance target; matching, according to the motion data, each matching position of each avoidance target to obtain a predicted position of each avoidance target at each motion time, wherein the predicted position at each motion time is one or more; and connecting the predicted positions of each avoidance target in time sequence to obtain at least one motion trajectory in the preset time period; adjusting, according to map information of a current region, the trajectory confidence of each motion trajectory to obtain an adjusted trajectory confidence, the current region containing a region passed through by each motion trajectory, and the adjusted trajectory confidence being used for obstacle avoidance of the target object to each avoidance target.

2. The trajectory prediction method of claim 1, wherein, The matching, according to the motion data, of each matching position of each avoidance target to obtain a predicted position of each avoidance target at each motion time comprises: determining, according to the motion data, a position offset of each avoidance target at each motion time relative to a previous motion time; in a plurality of matching positions, determining a position that satisfies a position offset related to a second time between a first position at a first time as a predicted position at the second time, wherein the second time is any motion time in the preset time period, and the first time is a previous motion time of the second time.

3. The trajectory prediction method of claim 2, wherein, The determination, according to the motion data, of a plurality of motion times in the preset time period and a matching position of the avoidance target at each motion time, and a reference type of the avoidance target at each matching position comprises: determining, according to the motion data, a plurality of motion times in the preset time period, a matching position of the avoidance target at each motion time, and a reference type of the avoidance target at each matching position; The determination, in a plurality of matching positions, of a position that satisfies a position offset related to a second time between a first position at a first time as a predicted position at the second time comprises: determining, as the predicted position at the second time, a matching position that satisfies the position offset related to the second time from the first position and has the same type as the avoidance target.

4. The trajectory prediction method of claim 1, wherein, The determination of the trajectory confidence of each motion trajectory comprises: determining a position confidence of the avoidance target at each predicted position; According to a position confidence of the predicted position on the same motion trajectory, a trajectory confidence corresponding to the motion trajectory is determined.

5. The trajectory prediction method of claim 4, wherein, The map information records an impassable area of the avoidance target; The adjusting of the trajectory confidence of each motion trajectory according to the map information of the current area to obtain an adjusted trajectory confidence comprises: adjusting the position confidence of the predicted position located in the impassable area; According to the adjusted position confidence, an adjusted trajectory confidence corresponding to the motion trajectory is determined.

6. The trajectory prediction method according to any one of claims 1 to 5, characterized in that, The determining of at least one motion trajectory of each avoidance target from the current time to the future time and the trajectory confidence of each motion trajectory according to the environment data comprises: The environment data is input into a target neural network to obtain at least one motion trajectory of each avoidance target from the current time to the future time and the trajectory confidence of each motion trajectory output by the target neural network, wherein the target neural network is trained by sample environment data containing multiple types of avoidance targets, and the sizes of at least part of the types of avoidance targets are different.

7. A trajectory prediction device characterized by comprising: comprise: The obtaining unit is configured to obtain environment data of an environment in which a target object is located, the environment comprising at least one avoidance target related to the target object; The determining unit is configured to determine at least one motion trajectory of each avoidance target from a current time to a future time and a trajectory confidence of each motion trajectory according to the environment data; wherein the determining of at least one motion trajectory of each avoidance target from the current time to the future time according to the environment data comprises: determining, according to the environment data, a plurality of motion time points within a preset time period, at which all the avoidance targets appear at to-be-matched positions, the preset time period being a time period from the current time to the future time; obtaining motion data of each avoidance target; matching, according to the motion data, the to-be-matched positions for each avoidance target to obtain predicted positions of the avoidance target at each motion time point, wherein each predicted position at each motion time point is one or more; and connecting the predicted positions of each avoidance target in time sequence to obtain at least one motion trajectory within the preset time period; The adjusting unit is configured to adjust the trajectory confidence of each motion trajectory according to map information of a current area to obtain an adjusted trajectory confidence, the current area comprising an area through which each motion trajectory passes, and the adjusted trajectory confidence being used for the target object to avoid obstacles of each avoidance target.

8. A terminal device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, The processor executes the computer program to implement the steps of the trajectory prediction method according to any one of claims 1 to 6.

9. A computer-readable storage medium storing a computer program, the computer program comprising instructions that, when executed by a computer, cause the computer to perform the method of any one of claims 1 to 8. The computer program is executed by the processor to implement the steps of the trajectory prediction method according to any one of claims 1 to 6.

Images