Automatic driving trajectory prediction method, device, equipment and storage medium

Abstract

Embodiments of the present invention disclose a method, apparatus, device, and storage medium for predicting an autonomous driving trajectory. The method includes: obtaining first historical trajectory data of an autonomous driving vehicle, second historical trajectory data of surrounding obstacles, and current map data; preprocessing the first historical trajectory data, the second historical trajectory data, and the current map data to determine first historical trajectory feature information, second historical trajectory feature information, and current map feature information; downsampling lane nodes in the current map feature information based on lane type to obtain downsampled target map feature information; inputting the first historical trajectory feature information, the second historical trajectory feature information, and the target map feature information into a preset trajectory prediction model, and obtaining a target predicted trajectory of obstacles around the autonomous driving vehicle based on the output of the preset trajectory prediction model, thereby improving the trajectory prediction efficiency of obstacles around the autonomous driving vehicle.

Description

Technical Field

[0001] Embodiments of the present invention relate to autonomous driving technology, and in particular to an autonomous driving trajectory prediction method, apparatus, device, and storage medium. Background Art

[0002] To ensure that autonomous vehicles move safely and efficiently on the road, they not only need to perceive the current state of surrounding obstacles, but also need to actively predict their future movement trajectory in order to make reasonable and safe behavioral decisions.

[0003] Currently, autonomous vehicles can predict trajectories based on perceived obstacle motion trajectories and map environment information. However, due to the complexity of roads and obstacles, predictions take a long time and are difficult to meet the requirements of real-time predictions. Summary of the Invention

[0004] Embodiments of the present invention provide an autonomous driving trajectory prediction method, apparatus, device, and storage medium to improve the efficiency of trajectory prediction of obstacles around an autonomous driving vehicle.

[0005] According to one aspect of the present invention, a method for predicting an autonomous driving trajectory is provided, comprising:

[0006] Obtaining first historical trajectory data of the autonomous driving vehicle, second historical trajectory data of surrounding obstacles, and current map data;

[0007] preprocessing the first historical trajectory data, the second historical trajectory data, and the current map data to determine first historical trajectory feature information, second historical trajectory feature information, and current map feature information;

[0008] Based on the lane type, downsampling the lane nodes in the current map feature information to obtain downsampled target map feature information;

[0009] The first historical trajectory feature information, the second historical trajectory feature information, and the target map feature information are input into a preset trajectory prediction model, and based on the output of the preset trajectory prediction model, a target predicted trajectory of obstacles around the autonomous driving vehicle is obtained.

[0010] According to another aspect of the present invention, there is provided an automatic driving trajectory prediction device, comprising:

[0011] A data acquisition module, configured to acquire first historical trajectory data of the autonomous driving vehicle, second historical trajectory data of surrounding obstacles, and current map data;

[0012] a data processing module, configured to pre-process the first historical trajectory data, the second historical trajectory data, and the current map data to determine first historical trajectory feature information, second historical trajectory feature information, and current map feature information;

[0013] a downsampling module, configured to downsample lane nodes in the current map feature information based on lane type to obtain downsampled target map feature information;

[0014] A trajectory prediction module is used to input the first historical trajectory feature information, the second historical trajectory feature information, and the target map feature information into a preset trajectory prediction model, and obtain a target predicted trajectory of obstacles around the autonomous driving vehicle based on the output of the preset trajectory prediction model.

[0015] According to another aspect of the present invention, an electronic device is provided, comprising:

[0016] at least one processor; and

[0017] a memory communicatively connected to the at least one processor; wherein,

[0018] The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the autonomous driving trajectory prediction method described in any embodiment of the present invention.

[0019] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the autonomous driving trajectory prediction method described in any embodiment of the present invention when executed.

[0020] The technical solution of an embodiment of the present invention obtains first historical trajectory data of the autonomous vehicle, second historical trajectory data of surrounding obstacles, and current map data, and preprocesses the first historical trajectory data, second historical trajectory data, and current map data to respectively determine first historical trajectory feature information, second historical trajectory feature information, and current map feature information. Lane nodes in the current map feature information are downsampled based on the lane type in the current map feature information to obtain downsampled target map feature information, thereby reducing the number of map features, thereby reducing trajectory prediction time and improving trajectory prediction efficiency. Furthermore, the first historical trajectory feature information, second historical trajectory feature information, and target map feature information are input into a preset trajectory prediction model. The preset trajectory prediction model, while predicting the trajectory of the autonomous vehicle, can also predict the trajectories of all obstacles around the autonomous vehicle and output a target predicted trajectory of the obstacles around the autonomous vehicle. Based on the preset trajectory prediction model, the target predicted trajectory of the obstacles around the autonomous vehicle can be obtained more quickly, thereby significantly reducing the obstacle trajectory prediction time and improving trajectory prediction efficiency while ensuring trajectory prediction accuracy.

[0021] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present invention, nor is it intended to limit the scope of the present invention. Other features of the present invention will become readily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0023] Figure 1 This is a flowchart of an autonomous driving trajectory prediction method provided by the first embodiment of the present invention;

[0024] Figure 2 This is an example diagram of downsampling lane nodes involved in the first embodiment of the present invention;

[0025] Figure 3 This is a flowchart of another automatic driving trajectory prediction method provided by the second embodiment of the present invention;

[0026] Figure 4 This is a flowchart of an autonomous driving trajectory prediction model training process provided by an embodiment of the present invention;

[0027] Figure 5This is a flowchart of a process for using an autonomous driving trajectory prediction model provided by an embodiment of the present invention;

[0028] Figure 6 This is a structural diagram of an autonomous driving trajectory prediction device provided by Embodiment 3 of the present invention;

[0029] Figure 7 2 is a schematic diagram of the structure of an electronic device that implements the automatic driving trajectory prediction method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0030] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of the present invention.

[0031] It should be noted that the terms "first", "second", etc. in the description and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the numbers used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0032] Example 1

[0033] Figure 1 This is a flow chart of an autonomous driving trajectory prediction method provided in the first embodiment of the present invention. This embodiment is applicable to the case of predicting the motion trajectory of surrounding obstacles while an autonomous driving vehicle is driving. The method can be executed by an autonomous driving trajectory prediction device, which can be implemented in the form of hardware and / or software and can be configured in the electronic equipment of the vehicle. Figure 1 As shown, the method specifically includes the following steps:

[0034] S110: Obtain first historical trajectory data of the autonomous driving vehicle, second historical trajectory data of surrounding obstacles, and current map data.

[0035] Among them, the first historical trajectory data can be the historical trajectory data of the autonomous driving vehicle itself when it was driving before the current moment; the second historical trajectory data can be the historical trajectory data of each obstacle around the autonomous driving vehicle when it was driving before the current moment. Surrounding obstacles can refer to various obstacles that affect the driving of the vehicle within a preset range centered on the vehicle, that is, various obstacles whose motion trajectory needs to be predicted. The number of surrounding obstacles can be one or more. It is understandable that obstacles can refer to other vehicles on the road when the vehicle is driving, or pedestrians and non-motorized vehicles on the road. In the embodiment of the present invention, the surrounding obstacles can be one or more of them, and the present invention is not limited to this. The current map data can be map data within a preset range centered on the current position of the autonomous driving vehicle. The current map data may include data on various driving roads around the vehicle.

[0036] Specifically, the autonomous vehicle can obtain first historical trajectory data of its actual travel prior to the current moment, and can also obtain second historical trajectory data of each obstacle surrounding the vehicle prior to the current moment. This second historical trajectory data can be the obstacle's motion trajectory data predicted by a prediction device prior to the current moment, or the perceived actual motion trajectory of the obstacle. This embodiment can obtain the autonomous vehicle's current location information and obtain current map data within a preset range centered on the current location information.

[0037] S120 : Preprocess the first historical trajectory data, the second historical trajectory data, and the current map data to determine first historical trajectory feature information, second historical trajectory feature information, and current map feature information.

[0038] The first historical trajectory feature information may be each historical trajectory position point obtained after feature extraction of the first historical trajectory data. The second historical trajectory feature information may be each historical trajectory position point obtained after feature extraction of the second historical trajectory data. The current map feature information may be each map feature point after feature extraction of the current map data, i.e., each lane node. In this embodiment, each discrete point on the center line of each lane may be used as a lane node. Generally, the amount of lane node data in the current map feature information is large, reaching 10 5 Order of magnitude.

[0039] Specifically, the vehicle's first historical trajectory data can be preprocessed using existing feature extraction methods to obtain first historical trajectory feature information. The vehicle's second historical trajectory data can be preprocessed using existing feature extraction methods to obtain second historical trajectory feature information. The current map data can be preprocessed using feature extraction rules to obtain current map information, namely, each lane node on each road within a certain range around the vehicle.

[0040] S130 : Downsample the lane nodes in the current map feature information based on the lane type to obtain downsampled target map feature information.

[0041] The lane type may refer to the road type corresponding to the road being traveled. For example, lane types may include straight lanes and turning lanes. Downsampling may involve reducing the number of lane nodes. This embodiment of the present invention downsamples each lane node in the current map feature information based on the lane type, obtaining a target lane node for each lane after downsampling, i.e., the target map feature information.

[0042] Specifically, based on the lane type corresponding to each lane in the current map feature information, a downsampling method that matches the lane type is selected to downsample all lane nodes on each lane in the current map feature information, thereby obtaining the target map feature information after all lane nodes on each lane are downsampled. It can be understood that by classifying lanes into different lane types, the accuracy of vehicle trajectory prediction can be improved by using a more suitable lane type downsampling method. By downsampling the lane nodes in the current map feature information, this embodiment can reduce the number of lane nodes in the current map feature information from O(n) to O(k), where n>k, where n is the number of lane nodes before downsampling and k is the number of lane nodes after downsampling. By reducing the number of lane nodes, the time performance of the prediction can be improved while ensuring the accuracy of the prediction.

[0043] Optionally, S130 may include: obtaining each lane node on each lane in the current map feature information; if the lane type of the lane is a straight lane, then based on a preset distance sampling method, downsampling each lane node on the lane to obtain a target lane node after downsampling of the lane; if the lane type of the lane is a turning lane, then based on a preset angle sampling method, downsampling each lane node on the lane to obtain a target lane node after downsampling of the lane.

[0044] The preset distance sampling mode may be a pre-set mode for downsampling based on the distance between two lane nodes. The preset angle sampling mode may be a pre-set mode for downsampling based on the angle between two lane nodes.

[0045] Specifically, each lane node on each lane in the current map feature information is obtained, and the lane type of each lane is determined. If the lane type is a straight lane, then based on a preset distance sampling method, each lane node on the lane is downsampled based on the distance between two adjacent lane nodes to obtain the downsampled target lane node of the lane; if the lane type is a turning lane, then based on a preset angle sampling method, each lane node on the lane is downsampled based on the angle between two adjacent lane nodes, thereby reducing the number of map features while ensuring the accuracy of vehicle trajectory prediction.

[0046] Exemplarily, downsampling each lane node on the lane based on a preset distance sampling method to obtain the target lane node of the lane after downsampling may include: determining the first node distance between each group of two adjacent lane nodes on the lane, and accumulating the first node distances; downsampling once each time the currently accumulated total first node distance is greater than or equal to a preset distance threshold, to obtain the target lane node of the lane after downsampling.

[0047] The preset distance threshold may be a pre-set minimum distance between two adjacent target lane nodes after sampling. The target lane node may be a lane node retained after downsampling. Figure 2 An example of downsampling lane nodes is given, such as Figure 2 As shown, for each lane in the current map feature information, when the lane type is a straight lane, downsampling is performed using the preset distance sampling method with distance as a reference. The first node distance between each group of two adjacent lane nodes on the lane is calculated, and the first node distances are accumulated. When the accumulated total first node distance is greater than or equal to the preset distance threshold each time, downsampling is performed, retaining the first lane node and the last lane node when the current accumulated total first node distance is greater than or equal to the preset distance threshold each time; when the accumulated total first node distance is less than the preset distance threshold each time, the first node distance is continued to be accumulated until the target lane node after downsampling is finally obtained, so that sampling can be performed once at fixed intervals. The calculation formula for the first node distance between each group of two adjacent lane nodes is as follows:

[0048]

[0049] Among them, lane_nodes is the lane node set, x1 is the x-axis position coordinate of the first lane node, x2 is the x-axis position coordinate of the second lane node adjacent to the first lane node, y1 is the y-axis position coordinate of the first lane node, and y2 is the y-axis position coordinate of the second lane node adjacent to the first lane node.

[0050] Exemplarily, the preset distance threshold can be 50 meters. When all lane nodes on a certain lane are downsampled, the first downsampling process is: the cumulative total distance between the first lane node and the second lane node is calculated to be 20 meters. The cumulative total distance is less than the preset distance threshold, then the distance between the third lane node and the second lane node is continued to be accumulated, and the cumulative total distance obtained is 45 meters. The cumulative total distance is still less than the preset distance threshold, then the distance between the fourth lane node and the third lane node is continued to be accumulated, and the cumulative total distance obtained is 65 meters. The cumulative total distance is greater than the preset distance threshold, then the first lane node and the fourth lane node of the current cumulative total distance are retained, and the first lane node and the fourth lane node are used as target lane nodes; the second downsampling process: the cumulative total distance between the fourth lane node and the fifth lane node is 50 meters. If the cumulative total distance is equal to the preset distance threshold, then the fifth lane node of the current cumulative total distance is retained, and the fifth lane node is used as the target lane node, and so on.

[0051] Exemplarily, downsampling each lane node on the lane based on a preset angle sampling method to obtain the downsampled target lane node for the lane may include: determining the node angle between each pair of adjacent lane nodes on the lane and accumulating the node angles; and downsampling each time the currently accumulated total node distance is greater than or equal to a preset angle threshold to obtain the downsampled target lane node for the lane. Specifically, for a turning lane, similar to the preset distance threshold described above, downsampling each lane node on the turning lane may be performed by setting a preset angle threshold, thereby sampling each lane node at a fixed angle interval.

[0052] S140: Input the first historical trajectory feature information, the second historical trajectory feature information, and the target map feature information into a preset trajectory prediction model, and obtain a target predicted trajectory of obstacles around the autonomous driving vehicle based on an output of the preset trajectory prediction model.

[0053] The preset trajectory prediction model may be a pre-set prediction model for predicting the motion trajectory of the autonomous vehicle and surrounding obstacles. For example, the preset trajectory prediction model may be, but is not limited to, a LaneGCN (Graph Convolutional Network) model. The preset trajectory prediction model in this embodiment is pre-trained based on sample data and a preset loss function so that it can predict the trajectory of all surrounding obstacles while predicting the trajectory of the autonomous vehicle.

[0054] Specifically, the first historical trajectory feature information, the second historical trajectory feature information, and the target map feature information can be input into a trained preset trajectory prediction model. The preset trajectory prediction model can predict the motion trajectory of the autonomous vehicle for a period of time in the future and also predict the motion trajectory of each obstacle around the vehicle for a period of time in the future, and output each predicted motion trajectory. Since the prediction device in this embodiment only needs to predict the motion trajectory of the obstacles around the autonomous vehicle, the target predicted trajectory of all obstacles around the autonomous vehicle can be obtained at one time based on the output of the preset trajectory prediction model. Compared with the existing method, the number of predictions in this embodiment is reduced from O(M) to O(1), where M is the number of obstacles. Therefore, this embodiment can more quickly obtain the target predicted trajectory of all obstacles around the autonomous vehicle by using the preset trajectory prediction model, thereby greatly reducing the time consumption of obstacle trajectory prediction while ensuring the accuracy of trajectory prediction and improving trajectory prediction efficiency.

[0055] It should be noted that, in this embodiment, the target predicted trajectory of all obstacles around the autonomous driving vehicle predicted by the prediction device can be input into the planning device, so that the planning device plans the vehicle movement based on the input target predicted trajectory of the surrounding obstacles, thereby deciding the current control operation of the autonomous driving vehicle, and thus avoiding collisions with obstacles during the driving of the autonomous driving vehicle.

[0056] The technical solution of an embodiment of the present invention obtains first historical trajectory data of the autonomous vehicle, second historical trajectory data of surrounding obstacles, and current map data, and preprocesses the first historical trajectory data, second historical trajectory data, and current map data to respectively determine first historical trajectory feature information, second historical trajectory feature information, and current map feature information. Lane nodes in the current map feature information are downsampled based on the lane type in the current map feature information to obtain downsampled target map feature information, thereby reducing the number of map features, thereby reducing trajectory prediction time and improving trajectory prediction efficiency. Furthermore, the first historical trajectory feature information, second historical trajectory feature information, and target map feature information are input into a preset trajectory prediction model. The preset trajectory prediction model, while predicting the trajectory of the autonomous vehicle, can also predict the trajectories of all obstacles around the autonomous vehicle and output a target predicted trajectory of the obstacles around the autonomous vehicle. Based on the preset trajectory prediction model, the target predicted trajectory of the obstacles around the autonomous vehicle can be obtained more quickly, thereby significantly reducing the obstacle trajectory prediction time and improving trajectory prediction efficiency while ensuring trajectory prediction accuracy.

[0057] On the basis of the above technical solution, the downsampling of each lane node on the lane based on the preset angle sampling method to obtain the target lane node of the lane after downsampling can also include: determining the node angle between the first group of two adjacent lane nodes at each sampling; determining the current distance threshold based on the node angle; determining the second node distance between the first group of two adjacent lane nodes and the second node distance between each subsequent group of two adjacent lane nodes, and accumulating the second node distances; downsampling when the current accumulated total second node distance is greater than or equal to the current distance threshold to obtain the target lane node of the lane after downsampling.

[0058] like Figure 2 As shown, for the case where the lane type of the lane is a curved lane, at the beginning of each sampling, the node angle between the first two adjacent lane nodes to be sampled (i.e., the first lane node and the second lane node) can be determined first, and based on the node angle, the current distance threshold required for reference for the current sampling can be dynamically determined. The second node distance between each group of adjacent lane nodes for the current sampling is accumulated until the accumulated total distance of the second node is greater than or equal to the current distance threshold, and then downsampling is performed, that is, the last lane node is used as the target lane node after the current sampling. By sampling once at fixed intervals on the straight lane based on the preset distance threshold, and sampling once at current distances matching the node angle on the turning lane based on the current distance threshold dynamically determined based on the node angle, the sampling points on the straight lane can be sparse, and the sampling points on the turning lane can be dense, thereby retaining more detailed information on the turning lane. Therefore, this angle sampling method can further improve the accuracy of trajectory prediction.

[0059] Exemplarily, determining the current distance threshold based on the node angle may include: detecting whether the node angle is greater than a preset angle threshold; if so, determining the ratio between the preset angle threshold and the node angle, and determining the current distance threshold based on the ratio and the preset distance threshold; if not, using the preset distance threshold as the current distance threshold.

[0060] The preset angle threshold may be pre-set, and the minimum angle of the distance threshold needs to be re-determined. In the embodiment of the present invention, the node angle between the first two adjacent lane nodes in each sampling is calculated using the node angle calculation formula, and the node angle calculation formula is as follows:

[0061]

[0062] Where a is the vector formed by the first lane node in the first group of two adjacent lane nodes and the center of the curve, b is the vector formed by the second lane node in the first group of two adjacent lane nodes and the center of the curve, and θ is the node angle between the first group of two adjacent lane nodes. If the node angle between the first group of two adjacent lane nodes is greater than the preset angle threshold, the ratio between the preset angle threshold and the node angle is determined, and the product of the ratio and the preset distance threshold is determined as the current distance threshold, so that the current distance threshold is less than the preset distance threshold. This allows the lane node distance interval to be reduced during sampling when the node angle is large, thereby allowing more target lane nodes to be sampled. For example, the current distance threshold can be calculated based on the following formula:

[0063]

[0064] Among them, T dis is the current distance threshold, T θ is the preset angle threshold, and T is the preset distance threshold.

[0065] Example 2

[0066] Figure 3 This is a flowchart of another automatic driving trajectory prediction method provided by the second embodiment of the present invention. Based on the above embodiment, the embodiment adds a training process of a preset trajectory prediction model. The explanations of the terms that are the same or corresponding to the above embodiments are not repeated here. Figure 3 As shown, the training process of the preset trajectory prediction model provided in this embodiment specifically includes the following steps:

[0067] S210. Obtain training data corresponding to the autonomous driving vehicle, where the training data includes: first historical trajectory sample data and a corresponding first actual trajectory of the autonomous driving vehicle, second historical trajectory sample data and a corresponding second actual trajectory of surrounding obstacles, and map sample data around the autonomous driving vehicle.

[0068] The first historical trajectory sample data may be historical trajectory sample data of the autonomous vehicle itself prior to the current moment; the second historical trajectory sample data may be historical trajectory sample data of each obstacle surrounding the autonomous vehicle, excluding the vehicle itself; and the map sample data may be map sample data within a preset range centered on the autonomous vehicle's sample position. The first actual trajectory may be the actual trajectory of the autonomous vehicle itself prior to the current moment; and the second actual trajectory may be the actual trajectory of each obstacle surrounding the autonomous vehicle.

[0069] Specifically, the autonomous vehicle can obtain first historical trajectory sample data and the corresponding first actual trajectory for the vehicle's actual travel before the current moment, and can also obtain second historical trajectory sample data and the corresponding second actual trajectory for each obstacle around the vehicle before the current moment. Sample position information of the autonomous vehicle is obtained, and current map sample data within a preset range is obtained with the sample position information as the center.

[0070] S220 : Preprocess the first historical trajectory sample data, the second historical trajectory sample data, and the map sample data to determine first historical trajectory sample feature information, second historical trajectory sample feature information, and map sample feature information.

[0071] The first historical trajectory sample feature information may be a historical trajectory sample point after feature extraction, the second historical trajectory sample feature information may be a historical trajectory sample point after feature extraction of the second historical trajectory sample data, and the map feature sample information may be each lane node in each sample lane after feature extraction of the map feature sample data.

[0072] S230 : Downsample the lane nodes in the map sample feature information based on the lane type to obtain downsampled target map sample feature information.

[0073] Specifically, based on the downsampling method described in the above embodiment, the lane nodes in the map sample feature information may be downsampled to obtain the downsampled target map sample feature information.

[0074] S240: Input the first historical trajectory sample feature information, the second historical trajectory sample feature information, and the target map sample feature information into a preset trajectory prediction model to be trained to obtain a first sample predicted trajectory of the autonomous driving vehicle and a second sample predicted trajectory of surrounding obstacles.

[0075] Among them, the first sample predicted trajectory can be the predicted trajectory of the autonomous driving vehicle output by the preset trajectory prediction model; the second sample predicted trajectory can be the predicted trajectory of the surrounding obstacles output by the preset trajectory prediction model.

[0076] S250: Based on a preset loss function, determine a training error according to the first sample predicted trajectory, the second sample predicted trajectory, the first actual trajectory, and the second actual trajectory, backpropagate the training error to the preset trajectory prediction model to be trained, and adjust network parameters in the preset trajectory prediction model to be trained until the training ends when a preset convergence condition is reached.

[0077] The preset loss function may be a loss function pre-set based on the first sample prediction trajectory and the second sample prediction trajectory; the preset convergence condition may refer to a pre-set model training end condition, for example, the preset convergence condition may be that the training error is less than the preset error, the change in the training error tends to be stable, or the number of training iterations reaches a preset number. Optionally, the preset loss function may be as follows:

[0078]

[0079] Among them, L reg is the training error; p m,t Predict the sample position of the mth object at the tth second; is the actual position of the mth object at the tth second, where the mth object is the autonomous vehicle or the obstacle; reg() is a preset error function; M is the total number of autonomous vehicles and surrounding obstacles, and T is the total prediction time. The preset error function can be, but is not limited to, a mean square error function, a mean absolute value error function, or a cross entropy error function.

[0080] The technical solution of the embodiment of the present invention determines the training error based on a preset loss function through the first sample predicted trajectory, the second sample predicted trajectory, the first actual trajectory and the second actual trajectory, and backpropagates the training error obtained from the training to the preset trajectory prediction model to be trained, thereby adjusting the network parameters in the preset trajectory prediction model to be trained until the training is terminated when the preset convergence condition is reached. In this way, the trained preset trajectory prediction model can predict the trajectories of all surrounding obstacles while predicting the trajectory of the autonomous driving vehicle, thereby further improving the prediction efficiency of the obstacle trajectory.

[0081] For example, in another embodiment of the present invention, Figure 4 A flowchart of the autonomous driving trajectory prediction model training process is given, where Figure 4 The specific steps shown include:

[0082] S410: Acquire the trajectory data of the perceived vehicle and the surrounding map data, pre-process the data, and extract the vehicle features and map features. Execute step 420.

[0083] S420: Downsample the map features obtained by preprocessing according to certain rules to obtain a training sample set. Execute steps S430-S490 for each lane node on each road.

[0084] S430. Downsample the lane nodes of a certain road and determine whether the number of lane nodes is greater than 4. If so, execute step S440; otherwise, skip the downsampling step for this road, add all lane nodes to the target lane node set, and execute step S490.

[0085] Step S440: Determine whether the road is a straight lane or a turning lane to determine the downsampling rule. If it is a straight lane, execute step S470; if it is a turning lane, execute step S450.

[0086] Step S450: Calculate the node angle between the vectors forming the adjacent lane nodes. If it is greater than the preset angle threshold, execute step S460; otherwise, execute step S470.

[0087] Step S460: Calculate the current distance threshold using the calculation formula, and execute step S480.

[0088] Step S470: Set the current distance threshold as the preset distance threshold, and execute step S480.

[0089] Step S480: Sample lane nodes according to distance, calculate the distance between lane nodes, sample a lane node every preset distance threshold, and execute step S490.

[0090] Step S490: Add the target lane node obtained by downsampling to the target sampling set, and use the corresponding map features in the training sample set. Repeat steps S420-S480 for all roads until all roads are traversed, and then execute step S4100.

[0091] Step S4100: Input the training sample set into the pre-built node graph convolutional network LaneGCN model for training, and perform performance evaluation on the trained trajectory prediction model.

[0092] Optionally, in another embodiment of the present invention, Figure 5 A flowchart of the process of using the autonomous driving trajectory prediction model is given, where Figure 5 The specific steps shown include:

[0093] The prediction process of vehicle trajectory prediction specifically includes the following steps:

[0094] Step S510: pre-process the real-time trajectory data of the autonomous driving vehicle and the surrounding map data to extract vehicle features and map features.

[0095] Step S520: Downsample the map features obtained through preprocessing according to certain rules. Execute steps S530-S590 for each lane node on each road.

[0096] Step S530: Downsample the lane nodes of a certain road and determine whether the number of lane nodes is greater than 4. If so, execute step S540; otherwise, skip the downsampling step for this road, add all lane nodes to the target lane node set, and execute step S590.

[0097] Step S540: Determine whether the road is a straight lane or a turning lane to determine the downsampling rule. If it is a straight lane, execute step S570; if it is a turning lane, execute step S540.

[0098] Step S550: Calculate the node angle between the vectors of adjacent lane nodes. If the angle is greater than a preset angle threshold, execute step S560; otherwise, execute step S570.

[0099] Step S560: Calculate the current distance threshold, and execute step S580.

[0100] Step S570: Set the current distance threshold as the preset distance threshold, and execute step S580.

[0101] Step S580: Sample lane nodes according to distance, calculate the distance between lane nodes, sample a lane node every preset distance threshold, and execute step S590.

[0102] Step S590: Add the target lane node obtained by downsampling to the target sampling set, and use the corresponding map features in the training sample set. Repeat steps S520-S580 for all roads until all roads are traversed, and then execute step S5100.

[0103] Step S5100: Input the training sample set into the pre-built node graph convolutional network LaneGCN model for training, and perform performance evaluation on the trained trajectory prediction model.

[0104] Example 3

[0105] Figure 6 This is a schematic diagram of the structure of an automatic driving trajectory prediction device provided by the third embodiment of the present invention. Figure 6 As shown, the device specifically includes: a data acquisition module 610, a data processing module 620, a downsampling module 630 and a trajectory prediction module 640. Among them:

[0106] A data acquisition module 610 is configured to acquire first historical trajectory data of the autonomous driving vehicle, second historical trajectory data of surrounding obstacles, and current map data;

[0107] a data processing module 620 for preprocessing the first historical trajectory data, the second historical trajectory data, and the current map data to determine first historical trajectory feature information, second historical trajectory feature information, and current map feature information;

[0108] A downsampling module 630 is configured to downsample lane nodes in the current map feature information based on lane type to obtain downsampled target map feature information;

[0109] The trajectory prediction module 640 is used to input the first historical trajectory feature information, the second historical trajectory feature information, and the target map feature information into a preset trajectory prediction model, and obtain a target predicted trajectory of obstacles around the autonomous driving vehicle based on the output of the preset trajectory prediction model.

[0110] The technical solution of the embodiment of the present invention obtains first historical trajectory data of the autonomous driving vehicle, second historical trajectory data of the surrounding obstacles, and current map data, and then pre-processes the first historical trajectory data, the second historical trajectory data of the surrounding obstacles, and the current map data to generate first historical trajectory feature information, second historical trajectory feature information, and current map feature information. Lane nodes in the trajectory feature information are downsampled according to different lane types, and the first historical trajectory feature information, the second historical trajectory feature information, and the target map feature information are input into a preset trajectory prediction model. Based on the output of the preset trajectory prediction model, a target predicted trajectory of the obstacles around the autonomous driving vehicle is obtained. The present application uses the downsampled lane nodes to greatly reduce the number of sampling points for calculation, thereby greatly reducing the amount of calculation of the predicted trajectory and the prediction time of the trajectory prediction while ensuring the accuracy of the trajectory prediction. This solves the technical problem in the prior art that the complex conditions of roads and obstacles will cause the prediction to take a long time and it is difficult to meet the real-time requirements of the prediction, thereby achieving the technical effect of improving the efficiency of the prediction.

[0111] Optionally, the downsampling module 630 is specifically configured to:

[0112] Obtaining each lane node on each lane in the current map feature information;

[0113] If the lane type is a through lane, then based on the preset distance sampling method, each lane node on the lane is downsampled to obtain the downsampled target lane node of the lane;

[0114] If the lane type is a turning lane, each lane node on the lane is downsampled based on a preset angle sampling method to obtain a target lane node after downsampling of the lane.

[0115] Optionally, the downsampling module 630 is further configured to:

[0116] Determine the first node distance between each group of two adjacent lane nodes on the lane, and accumulate the first node distances;

[0117] Downsampling is performed each time the currently accumulated total distance of the first node is greater than or equal to a preset distance threshold to obtain the target lane node after downsampling of the lane.

[0118] Optionally, the downsampling module 630 is further configured to:

[0119] Determine the node angle between the first two adjacent lane nodes at each sampling time;

[0120] determining a current distance threshold based on the node angle;

[0121] Determining a second node distance between the first group of two adjacent lane nodes and a second node distance between each subsequent group of two adjacent lane nodes, and accumulating the second node distances;

[0122] When the currently accumulated total distance of the second node is greater than or equal to the current distance threshold, downsampling is performed to obtain the target lane node after downsampling of the lane.

[0123] Optionally, the downsampling module 630 is further configured to:

[0124] Detecting whether the node angle is greater than a preset angle threshold;

[0125] If so, determining a ratio between the preset angle threshold and the node angle, and determining a current distance threshold based on the ratio and the preset distance threshold;

[0126] If not, the preset distance threshold is used as the current distance threshold.

[0127] Optionally, the device is further used for:

[0128] Obtaining training data corresponding to the autonomous driving vehicle, the training data comprising: first historical trajectory sample data of the autonomous driving vehicle and a corresponding first actual trajectory, second historical trajectory sample data of surrounding obstacles and a corresponding second actual trajectory, and map sample data around the autonomous driving vehicle;

[0129] preprocessing the first historical trajectory sample data, the second historical trajectory sample data, and the map sample data to determine first historical trajectory sample feature information, second historical trajectory sample feature information, and map sample feature information;

[0130] Based on the lane type, downsampling the lane nodes in the map sample feature information to obtain downsampled target map sample feature information;

[0131] Inputting the first historical trajectory sample feature information, the second historical trajectory sample feature information, and the target map sample feature information into a preset trajectory prediction model to be trained to obtain a first sample predicted trajectory of the autonomous driving vehicle and a second sample predicted trajectory of surrounding obstacles;

[0132] Based on a preset loss function, a training error is determined according to the first sample predicted trajectory, the second sample predicted trajectory, the first actual trajectory, and the second actual trajectory, and the training error is back-propagated to the preset trajectory prediction model to be trained, and the network parameters in the preset trajectory prediction model to be trained are adjusted until the training ends when the preset convergence condition is reached.

[0133] Optionally, the apparatus is further configured to determine a training error based on the following preset loss function:

[0134]

[0135] Among them, L reg is the training error; p m,t Predict the sample position of the mth object at the tth second; is the actual position of the mth object at the tth second, where the mth object is the autonomous driving vehicle or the obstacle; reg() is a preset error function; M is the total number of autonomous driving vehicles and surrounding obstacles, and T is the total prediction time.

[0136] Optionally, the device is further used for:

[0137] The preset trajectory prediction model is the node graph convolutional network LaneGCN model.

[0138] The autonomous driving trajectory prediction device provided by the embodiment of the present invention can execute the autonomous driving trajectory prediction method provided by any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of executing the autonomous driving trajectory prediction method.

[0139] It is worth noting that in the embodiment of the above-mentioned autonomous driving trajectory prediction device, the various units and modules included are only divided according to functional logic, but are not limited to the above-mentioned division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention.

[0140] Example 4

[0141] Figure 7A schematic diagram of the structure of an electronic device 10 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or claimed herein.

[0142] like Figure 7 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., which is communicatively connected to the at least one processor 11. The memory stores a computer program that can be executed by the at least one processor. The processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 into the random access memory (RAM) 13. Various programs and data required for the operation of the electronic device 10 can also be stored in the RAM 13. The processor 11, ROM 12, and RAM 13 are connected to each other via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0143] Multiple components in the electronic device 10 are connected to the I / O interface 15, including an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a magnetic disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0144] The processor 11 can be any general-purpose and / or specialized processing component with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various specialized artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as the autonomous driving trajectory prediction method.

[0145] In some embodiments, the autonomous driving trajectory prediction method can be implemented as a computer program tangibly contained in a computer-readable storage medium, such as the storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the autonomous driving trajectory prediction method described above can be performed. Alternatively, in other embodiments, the processor 11 can be configured to execute the autonomous driving trajectory prediction method in any other appropriate manner (e.g., by means of firmware).

[0146] Various embodiments of the systems and techniques described herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), system-on-chip systems (SOCs), programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include being implemented in one or more computer programs that are executable and / or interpreted on a programmable system that includes at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0147] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when the computer program is executed by the processor, the functions / operations specified in the flowcharts and / or block diagrams are implemented. The computer program may be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0148] In the context of the present invention, computer-readable storage media can be tangible media that can contain or store a computer program for use with an instruction execution system, device or equipment or used in combination with an instruction execution system, device or equipment. Computer-readable storage media can include but are not limited to electronic, magnetic, optical, electromagnetic, infrared or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, computer-readable storage media can be machine-readable signal media. More specific examples of machine-readable storage media can include electrical connections based on one or more lines, portable computer disks, hard disks, random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memory), optical fibers, portable compact disk read-only memories (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0149] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0150] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0151] A computing system may include clients and servers. The clients and servers are typically remote from each other and typically interact via a communication network. This client-server relationship arises through computer programs running on the respective computers, creating a client-server relationship. The server may be a cloud server, also known as a cloud computing server or cloud host. This server is a hosting product within the cloud computing service ecosystem that addresses the management difficulties and limited scalability of traditional physical hosting and VPS services.

[0152] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in the present invention can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solution of the present invention can be achieved. This is not limited herein.

[0153] Example 5

[0154] This embodiment provides a computer-readable storage medium having a computer program stored thereon. When the program is executed by a processor, the program implements the steps of the autonomous driving trajectory prediction method provided in any embodiment of the present invention. The method includes:

[0155] Obtaining first historical trajectory data of the autonomous driving vehicle, second historical trajectory data of surrounding obstacles, and current map data;

[0156] preprocessing the first historical trajectory data, the second historical trajectory data, and the current map data to determine first historical trajectory feature information, second historical trajectory feature information, and current map feature information;

[0157] Based on the lane type, downsampling the lane nodes in the current map feature information to obtain downsampled target map feature information;

[0158] The first historical trajectory feature information, the second historical trajectory feature information, and the target map feature information are input into a preset trajectory prediction model, and based on the output of the preset trajectory prediction model, a target predicted trajectory of obstacles around the autonomous driving vehicle is obtained.

[0159] The computer storage medium of the embodiment of the present invention can adopt any combination of one or more computer-readable media. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium can be, for example, but not limited to: an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or component, or any combination of the above. More specific examples (non-exhaustive list) of computer-readable storage media include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this document, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device.

[0160] A computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, which carries computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transport a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0161] Program code embodied on a computer-readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0162] Computer program code for performing the operations of the present invention can be written in one or more programming languages, or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (e.g., through the Internet using an Internet service provider).

[0163] Those skilled in the art will appreciate that the modules or steps of the present invention described above can be implemented using a general-purpose computing device. They can be centralized on a single computing device or distributed across a network of multiple computing devices. Alternatively, they can be implemented using program code executable by a computer device, which can then be stored in a storage device and executed by the computing device. Alternatively, they can be fabricated into separate integrated circuit modules, or multiple modules or steps can be fabricated into a single integrated circuit module. Thus, the present invention is not limited to any specific combination of hardware and software.

[0164] Note that the above are only preferred embodiments of the present invention and the technical principles employed. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein, and that various obvious changes, readjustments, and substitutions can be made by those skilled in the art without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in detail through the above embodiments, the present invention is not limited to the above embodiments and may include many other equivalent embodiments without departing from the concept of the present invention. The scope of the present invention is determined by the scope of the appended claims.

[0165] Note that the above are only preferred embodiments of the present invention and the technical principles employed. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein, and that various obvious changes, readjustments, and substitutions can be made by those skilled in the art without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in detail through the above embodiments, the present invention is not limited to the above embodiments and may include many other equivalent embodiments without departing from the concept of the present invention. The scope of the present invention is determined by the scope of the appended claims.

Claims

1. A method for predicting an autonomous driving trajectory, characterized in that: include: Obtaining first historical trajectory data of the autonomous driving vehicle, second historical trajectory data of surrounding obstacles, and current map data; preprocessing the first historical trajectory data, the second historical trajectory data, and the current map data to determine first historical trajectory feature information, second historical trajectory feature information, and current map feature information; Based on the lane type, downsampling the lane nodes in the current map feature information to obtain downsampled target map feature information; Inputting the first historical trajectory feature information, the second historical trajectory feature information, and the target map feature information into a preset trajectory prediction model, and obtaining a target predicted trajectory of obstacles around the autonomous driving vehicle based on an output of the preset trajectory prediction model; The downsampling of lane nodes in the current map feature information based on the lane type to obtain downsampled target map feature information includes: Obtaining each lane node on each lane in the current map feature information; If the lane type is a through lane, then based on the preset distance sampling method, each lane node on the lane is downsampled to obtain the downsampled target lane node of the lane; If the lane type is a turning lane, then based on the preset angle sampling method, each lane node on the lane is downsampled to obtain the downsampled target lane node of the lane; The downsampling of each lane node on the lane based on the preset distance sampling method to obtain the downsampled target lane node of the lane includes: Determine the first node distance between each group of two adjacent lane nodes on the lane, and accumulate the first node distances; Downsampling is performed each time the currently accumulated total distance of the first node is greater than or equal to a preset distance threshold to obtain the target lane node after downsampling of the lane.

2. The method according to claim 1, characterized in that The downsampling of each lane node on the lane based on the preset angle sampling method to obtain the downsampled target lane node of the lane includes: Determine the node angle between the first two adjacent lane nodes at each sampling time; determining a current distance threshold based on the node angle; Determining a second node distance between the first group of two adjacent lane nodes and a second node distance between each subsequent group of two adjacent lane nodes, and accumulating the second node distances; When the currently accumulated total distance of the second node is greater than or equal to the current distance threshold, downsampling is performed to obtain the target lane node after downsampling of the lane.

3. The method according to claim 2, characterized in that The determining of the current distance threshold based on the node angle includes: Detecting whether the node angle is greater than a preset angle threshold; If so, determining a ratio between the preset angle threshold and the node angle, and determining a current distance threshold based on the ratio and the preset distance threshold; If not, the preset distance threshold is used as the current distance threshold.

4. The method according to claim 1, wherein Before using the preset trajectory prediction model, a training process of the preset trajectory prediction model is also included, specifically including: Obtaining training data corresponding to the autonomous driving vehicle, the training data comprising: first historical trajectory sample data of the autonomous driving vehicle and a corresponding first actual trajectory, second historical trajectory sample data of surrounding obstacles and a corresponding second actual trajectory, and map sample data around the autonomous driving vehicle; preprocessing the first historical trajectory sample data, the second historical trajectory sample data, and the map sample data to determine first historical trajectory sample feature information, second historical trajectory sample feature information, and map sample feature information; Based on the lane type, downsampling the lane nodes in the map sample feature information to obtain downsampled target map sample feature information; Inputting the first historical trajectory sample feature information, the second historical trajectory sample feature information, and the target map sample feature information into a preset trajectory prediction model to be trained to obtain a first sample predicted trajectory of the autonomous driving vehicle and a second sample predicted trajectory of surrounding obstacles; Based on a preset loss function, a training error is determined according to the first sample predicted trajectory, the second sample predicted trajectory, the first actual trajectory, and the second actual trajectory, and the training error is back-propagated to the preset trajectory prediction model to be trained, and the network parameters in the preset trajectory prediction model to be trained are adjusted until the training ends when the preset convergence condition is reached.

5. The method according to claim 4, characterized in that The training error is determined based on the following preset loss function: Among them, L reg is the training error; p m,t Predict the sample position of the mth object at the tth second; is the actual position of the mth object at the tth second, where the mth object is the autonomous driving vehicle or the obstacle; reg() is a preset error function; M is the total number of autonomous driving vehicles and surrounding obstacles, and T is the total prediction time.

6. The method according to any one of claims 1 to 5, characterized in that The preset trajectory prediction model is the node graph convolutional network LaneGCN model.

7. An automatic driving trajectory prediction device, characterized in that: include: A data acquisition module, configured to acquire first historical trajectory data of the autonomous driving vehicle, second historical trajectory data of surrounding obstacles, and current map data; a data processing module, configured to pre-process the first historical trajectory data, the second historical trajectory data, and the current map data to determine first historical trajectory feature information, second historical trajectory feature information, and current map feature information; a downsampling module, configured to downsample lane nodes in the current map feature information based on lane type to obtain downsampled target map feature information; a trajectory prediction module, configured to input the first historical trajectory feature information, the second historical trajectory feature information, and the target map feature information into a preset trajectory prediction model, and obtain a target predicted trajectory of obstacles around the autonomous driving vehicle based on an output of the preset trajectory prediction model; The downsampling module is specifically configured to: obtain each lane node on each lane in the current map feature information; if the lane type is a through lane, downsample each lane node on the lane based on a preset distance sampling method to obtain a target lane node after downsampling of the lane; If the lane type is a turning lane, then based on the preset angle sampling method, each lane node on the lane is downsampled to obtain the downsampled target lane node of the lane; The downsampling module is further specifically configured to: determine the first node distance between each group of two adjacent lane nodes on the lane, and accumulate the first node distances; perform downsampling each time the currently accumulated total first node distance is greater than or equal to a preset distance threshold, to obtain the target lane node for the lane after downsampling.

8. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the autonomous driving trajectory prediction method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the automatic driving trajectory prediction method according to any one of claims 1 to 6 when executed.

Images