Obstacle trajectory prediction method, device, equipment and storage medium

Abstract

The present application discloses a method, device, equipment and storage medium for predicting the motion trajectory of an obstacle, which relates to the field of autonomous driving technology and can improve the prediction accuracy of the motion trajectory of an obstacle based on traditional electronic prediction. The specific scheme includes: receiving the obstacle types and historical trajectories of each obstacle included in a preset range around the vehicle, receiving the vehicle planned path sent by the planning module, and receiving the vehicle position sent by the positioning module; obtaining a regional map of the preset range from the map module based on the vehicle position, the map module including a high-precision map and an electronic map; if the regional electronic map of the preset range is obtained but the regional high-precision map of the preset range is not obtained, then obtaining an image of the preset range from the perception module, and generating a high-precision map of the class area of ​​the preset range based on the image and the regional electronic map, and determining the predicted trajectory of the obstacle based on the high-precision map of the class area, the obstacle type, the historical trajectory of the obstacle and the vehicle planned path.

Description

Technical Field

[0001] The present application relates to the field of autonomous driving technology, and in particular to a method, device, equipment, and storage medium for predicting obstacle motion trajectories. Background Art

[0002] With advancements in artificial intelligence, electronic information, automatic control, and intelligent manufacturing, autonomous driving technology is developing rapidly. This includes predicting the trajectory of obstacles around the vehicle, including pedestrians, bicycles, and other motor vehicles.

[0003] Currently, obstacle trajectory prediction is typically based on road network information provided by high-precision maps and perceived obstacle motion. However, in some areas where high-precision maps are not yet available, obstacle trajectory prediction can only be based on traditional electronic maps and obstacle motion information. Because traditional electronic maps lack precise lane information, prediction accuracy is low. Summary of the Invention

[0004] The present application provides a method, apparatus, device and storage medium for predicting obstacle motion trajectories, which can improve the prediction accuracy of obstacle motion trajectories based on traditional electronic prediction.

[0005] To achieve the above objectives, this application adopts the following technical solutions:

[0006] In a first aspect of an embodiment of the present application, a method for predicting the motion trajectory of an obstacle is provided, which is applied to a prediction module in an autonomous driving system of a vehicle, the system also including a perception module, a planning module, a positioning module, and a mapping module. The method includes:

[0007] Receive the types of obstacles and historical trajectories of each obstacle within a preset range around the vehicle from the perception module, receive the planned vehicle path from the planning module, and receive the vehicle position from the positioning module;

[0008] Obtain a map of a preset area from a map module based on the vehicle's location. The map module includes a high-precision map and an electronic map.

[0009] If an electronic map of the area within the preset range is obtained, but a high-precision map of the area within the preset range is not obtained, an image of the preset range is obtained from the perception module, and a high-precision map of the quasi-area of ​​the preset range is generated based on the image and the electronic map of the area, and the predicted trajectory of the obstacle is determined based on the high-precision map of the quasi-area, the obstacle type, the historical trajectory of the obstacle, and the planned path of the vehicle.

[0010] In one embodiment, generating a high-precision map of a quasi-region of a preset range based on an image and a regional electronic map includes:

[0011] Input the image into a preset object recognition model based on the attention mechanism to obtain environmental information within a preset range, including recognized road information within the preset range;

[0012] The identified road information is matched with the road information in the regional electronic map to obtain a high-precision map of the pre-set area.

[0013] In one embodiment, determining a predicted obstacle trajectory based on a high-precision map of the class area, the obstacle type, the obstacle's historical trajectory, and the vehicle's planned path includes:

[0014] The high-precision map of the class area, obstacle type, historical trajectory of the obstacle and vehicle planned path are input into the preset trajectory prediction model based on the graph network to obtain the movement intention of the obstacle and the predicted trajectory of the obstacle.

[0015] In one embodiment, after determining the predicted trajectory of the obstacle based on the high-precision map of the class area, the obstacle type, the historical trajectory of the obstacle, and the planned path of the vehicle, the method further includes:

[0016] Obtaining position information of the predicted trajectory, and if the position information indicates that the predicted trajectory of the obstacle is located in the road area, generating a target trajectory of the obstacle based on the lane direction of the lane to which the predicted trajectory belongs;

[0017] If the position information indicates that the predicted trajectory of the obstacle is located in the intersection area, a target trajectory of the obstacle is generated according to the movement intention of the obstacle and the predicted trajectory of the obstacle.

[0018] In one embodiment, after obtaining a map of an area within a preset range from a map module according to the vehicle location, the method further includes:

[0019] If a high-precision map of the area within the preset range is obtained, the predicted trajectory of the obstacle is determined based on the high-precision map of the area, the obstacle type, the historical trajectory of the obstacle, and the planned path of the vehicle.

[0020] In one embodiment, after determining the predicted trajectory of the obstacle based on the high-precision map of the area, the obstacle type, the historical trajectory of the obstacle, and the planned path of the vehicle, the method further includes:

[0021] Determine the target lane to which the predicted trajectory belongs in the high-precision map, and generate the target trajectory of the obstacle based on the lane direction of the target lane.

[0022] In one embodiment, obtaining a map of an area within a preset range from a map module according to the vehicle position includes:

[0023] Sending a data acquisition request for a regional map to a map module, so that the map module calls a first search engine corresponding to the high-precision map to determine a high-precision map of the region, or, if the first search engine fails to determine a high-precision map of the region, calls a second search engine corresponding to the electronic map to determine an electronic map of the region;

[0024] Receive the regional high-precision map or regional electronic map sent by the map module.

[0025] In a second aspect of an embodiment of the present application, a device for predicting obstacle motion trajectories is provided. The device is located in an autonomous driving system of a vehicle. The system further includes a planning module, a perception module, a positioning module, and a mapping module. The device includes:

[0026] A receiving module is configured to receive information on obstacle types and historical trajectories of obstacles within a preset range around the vehicle from the perception module, receive the planned vehicle path from the planning module, and receive the vehicle position from the positioning module;

[0027] An acquisition module is used to obtain a map of a preset area from a map module based on the vehicle's location. The map module includes a high-precision map and an electronic map.

[0028] The processing module is configured to obtain an image of the preset range from the perception module if an electronic map of the area within the preset range is obtained but a high-precision map of the area within the preset range is not obtained, and to generate a high-precision map of the quasi-area of ​​the preset range based on the image and the electronic map of the area, and to determine a predicted trajectory of the obstacle based on the high-precision map of the quasi-area, the obstacle type, the historical trajectory of the obstacle, and the planned path of the vehicle.

[0029] According to a third aspect of an embodiment of the present application, an electronic device is provided, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the method for predicting the obstacle motion trajectory according to the first aspect of the embodiment of the present application is implemented.

[0030] In a fourth aspect of an embodiment of the present application, a computer-readable storage medium is provided, characterized in that a computer program is stored thereon, and when the computer program is executed by a processor, the method for predicting the obstacle motion trajectory in the first aspect of the embodiment of the present application is implemented.

[0031] The beneficial effects of the technical solutions provided in the embodiments of the present application include at least:

[0032] The obstacle trajectory prediction method provided in embodiments of the present application receives obstacle types and historical trajectories within a preset range around the vehicle from a perception module, receives the vehicle's planned path from a planning module, and receives the vehicle's position from a positioning module. Based on the vehicle's position, the method then obtains a map of the area within the preset range from a mapping module, where the map module includes both a high-precision map and an electronic map. If an electronic map of the area within the preset range is obtained but an high-precision map of the area within the preset range is not, the method obtains an image of the preset range from the perception module and generates a high-precision map of the area within the preset range based on the image and the electronic map. Finally, the predicted obstacle trajectory is determined based on the high-precision map of the area, the obstacle type, the obstacle's historical trajectory, and the vehicle's planned path. The obstacle trajectory prediction method provided in embodiments of the present application utilizes the electronic map and the image of the area to generate map information for the area where only an electronic map is available. Since the image can reflect the actual environmental information of the area, a high-precision map similar to the area can be generated from the image and the electronic map, thereby improving the accuracy of obstacle prediction. Furthermore, this method does not require trajectory prediction based on high-precision map data, thus requiring less computing power and deployment equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 A schematic diagram of the internal structure of a vehicle-mounted terminal provided in an embodiment of the present application;

[0034] Figure 2 A schematic diagram of the software architecture of a method for predicting obstacle motion trajectories provided in an embodiment of the present application;

[0035] Figure 3 A flowchart of a method for predicting the motion trajectory of an obstacle provided in an embodiment of the present application;

[0036] Figure 4 This is a structural diagram of a device for predicting the motion trajectory of an obstacle provided in an embodiment of the present application. DETAILED DESCRIPTION

[0037] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0038] In the following, the terms "first" and "second" are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of the technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of the embodiments of the present disclosure, unless otherwise specified, "plurality" means two or more.

[0039] Additionally, the use of “based on” or “according to” is intended to be open and inclusive, in that a process, step, calculation, or other action “based on” or “according to” one or more conditions or values ​​may, in practice, be based on additional conditions or beyond values.

[0040] With advancements in artificial intelligence, electronic information, automatic control, and intelligent manufacturing, autonomous driving technology is developing rapidly. This includes predicting the trajectory of obstacles around the vehicle, including pedestrians, bicycles, and other motor vehicles.

[0041] Currently, vehicle prediction modules in autonomous driving systems are primarily based on high-precision maps (HD maps), or they perform only coarse-grained vehicle trajectory prediction without a map. Current HD map-based vehicle trajectory prediction, whether AI-based or road network rule-based, strongly relies on HD maps. The prediction module requires querying and calculating road network information around obstacles. Non-map-based prediction methods generally only provide coarse-grained dynamic predictions of target obstacles, or directly generate coarse-grained, short-term vehicle trajectory predictions based on perception methods. Current technical solutions have significant limitations. While HD map-based solutions can produce highly accurate results, their scope of application is limited. Currently, most areas lack HD map coverage, making these solutions inapplicable. Furthermore, most current vehicle prediction methods based on deep learning and HD maps require high computing power, making them difficult to deploy on low-computing platforms and limiting their hardware applicability. Non-map-based solutions have a wide range of applications, but their algorithmic accuracy is relatively low, placing significant pressure on downstream modules within autonomous driving systems and failing to meet the requirements of high-level autonomous driving systems.

[0042] To address the aforementioned issues, embodiments of the present application provide a method for predicting obstacle trajectories. The method receives obstacle types and historical trajectories within a preset range around a vehicle from a perception module, receives the vehicle's planned path from a planning module, and receives the vehicle's position from a positioning module. Based on the vehicle's position, the method then obtains a map of the area within the preset range from a mapping module, where the map module includes a high-precision map and an electronic map. If an electronic map of the area within the preset range is obtained but an high-precision map of the area within the preset range is not, the method obtains an image of the preset range from the perception module and generates a high-precision map of the area within the preset range based on the image and the electronic map. Finally, the predicted obstacle trajectory is determined based on the high-precision map of the area, the obstacle type, the obstacle's historical trajectories, and the vehicle's planned path. The method for predicting obstacle trajectories provided in embodiments of the present application utilizes the electronic map and the image of the area to generate map information for an area where only an electronic map is available. Because the image can reflect the actual environmental information of the area, a high-precision map similar to the area can be generated using the image and the electronic map, thereby improving the accuracy of the map data for the area and, in turn, the accuracy of obstacle prediction. At the same time, this method does not require trajectory prediction based on high-precision map data, so it has low requirements on computing power and deployment equipment.

[0043] The execution subject of the obstacle motion trajectory prediction method provided in the embodiment of the present application can be an electronic device, which can be a computer device, a terminal device, or a server. The terminal device can be a vehicle-mounted terminal, various personal computers, laptops, smart phones, tablet computers and portable wearable devices, etc., and this application does not make specific limitations.

[0044] Taking the electronic device as a vehicle terminal as an example, Figure 1 This is a schematic diagram of the internal structure of a vehicle-mounted terminal provided in an embodiment of the present application. Figure 1 As shown, the vehicle-mounted terminal includes a processor and memory connected via a system bus. The processor is used to provide computing and control capabilities. The memory may include a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and a computer program. The computer program can be executed by the processor to implement the steps of the obstacle trajectory prediction method provided in each of the above embodiments. The internal memory provides a high-speed cache operating environment for the operating system and computer program stored in the non-volatile storage medium.

[0045] Those skilled in the art will understand that Figure 1The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the electronic device to which the solution of the present application is applied. The specific electronic device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0046] like Figure 2 As shown, an embodiment of the present application provides a software architecture diagram of a method for predicting obstacle motion trajectories, wherein the autonomous driving system is deployed in an electronic device. The autonomous driving system includes a perception module, a planning module, a positioning module, a mapping module, and a prediction module.

[0047] The perception module is used to obtain information about obstacles and the surrounding environment of the ego vehicle, the planning module is used to generate a planned path for the ego vehicle, the positioning module is used to provide the ego vehicle's location information, the mapping module is used to provide high-precision map information or electronic map information, and the prediction module is used to predict the movement trajectory of obstacles. The obstacle movement trajectory prediction method provided in the embodiments of the present application is applied to the prediction module of the autonomous driving system.

[0048] Based on the above execution subject and the above software architecture, the embodiment of the present application provides a method for predicting the movement trajectory of an obstacle. Figure 3 As shown, the method includes the following steps:

[0049] Step 301: Receive the types of obstacles and historical trajectories of obstacles within a preset range around the vehicle from the perception module, receive the planned vehicle path from the planning module, and receive the vehicle position from the positioning module.

[0050] Obstacle types include other motor vehicles, bicycles, and pedestrians around the vehicle. The obstacle's historical trajectory is the obstacle's historical movement trajectory. The vehicle's planned path is the vehicle's local planned path.

[0051] It should be noted that before the method is executed, each module in the autonomous driving system will be initialized and the subscription nodes before each module will be configured. After the method starts executing, data will be sent and received according to the subscription nodes between the configured modules.

[0052] Step 302: Obtain a map of a preset area from a map module according to the vehicle location.

[0053] The map module includes high-precision maps and electronic maps. The main difference between HD maps and traditional electronic maps is that HD maps are used by autonomous driving systems, while traditional electronic maps are used by human drivers. Traditional electronic maps depict roads, some with lane distinctions. HD maps not only depict the roads but also accurately depict the number of lanes on a road, faithfully reflecting the actual road layout. Traditional electronic maps do not fully display the details of the road shape, but HD maps accurately and precisely display the road shape, allowing autonomous driving systems to better identify traffic conditions and plan driving plans in advance. The details of the road shape are precisely displayed, including where it widens and narrows, exactly as it would on a real road.

[0054] Optionally, the process of obtaining the regional map may include: sending a data acquisition request for the regional map to the map module; upon receiving the data acquisition request, the map module calls a first search engine corresponding to the high-precision map to determine a high-precision map of the region within a preset range; if a high-precision map of the region within the predicted range is determined, the high-precision map of the region is sent to the prediction module. If, upon receiving the data acquisition request, the map module calls the first search engine corresponding to the high-precision map but fails to determine a high-precision map of the region within the preset range, the map module calls a second search engine corresponding to the electronic map to determine an electronic map of the region within the preset range, and sends the electronic map to the prediction module.

[0055] It is understandable that the implementation of autonomous driving requires high-precision maps. However, since some areas are currently not covered by high-precision maps and only have traditional electronic maps, if no corresponding high-precision map is detected for a certain area, the electronic map of that area is sent to the prediction module for predicting the movement trajectory of obstacles.

[0056] Step 303: If an electronic map of the area within the preset range is obtained but a high-precision map of the area within the preset range is not obtained, an image of the preset range is obtained from the perception module, and a high-precision map of the quasi-area of ​​the preset range is generated based on the image and the electronic map of the area. The predicted trajectory of the obstacle is determined based on the high-precision map of the quasi-area, the obstacle type, the historical trajectory of the obstacle, and the planned path of the vehicle.

[0057] The regional electronic map is an electronic map of a preset range, and the regional high-precision map is a high-precision map of a preset range. The image is image information of the surrounding environment captured by the image acquisition device on the vehicle. The predicted obstacle trajectory is the predicted short-term movement trajectory of the obstacle, for example, the obstacle's movement trajectory in the next 2 to 4 seconds.

[0058] Optionally, there are two ways to generate the predicted trajectory of the obstacle.

[0059] In one implementation, if a high-precision map is deployed in the preset range, the predicted trajectory of the obstacle is determined directly based on the high-precision map of the area in the preset range, the obstacle type, the historical trajectory of the obstacle, and the planned path of the vehicle.

[0060] In another implementation, if a high-precision map is not deployed for the preset area and only an electronic map of the preset area is available, the electronic map of the preset area and the image of the preset area are used to generate map information for the preset area, namely a high-precision map of the similar area. Finally, the predicted obstacle trajectory is determined based on the high-precision map of the similar area, the obstacle type, the obstacle's historical trajectory, and the vehicle's planned path.

[0061] The obstacle trajectory prediction method provided in embodiments of the present application receives obstacle types and historical trajectories within a preset range around the vehicle from a perception module, receives the vehicle's planned path from a planning module, and receives the vehicle's position from a positioning module. Based on the vehicle's position, the method then obtains a map of the area within the preset range from a mapping module, where the mapping module includes a high-precision map and an electronic map. If an electronic map of the area within the preset range is obtained but an high-precision map of the area within the preset range is not, the method obtains an image of the preset range from the perception module and generates a high-precision map of the area within the preset range based on the image and the electronic map. Finally, the predicted obstacle trajectory is determined based on the high-precision map of the area, the obstacle type, the obstacle's historical trajectory, and the vehicle's planned path. The obstacle trajectory prediction method provided in embodiments of the present application generates map information for areas where only an electronic map is available, using the electronic map and the image of the area. Because the image can reflect the actual environmental information of the area, a high-precision map similar to the area can be generated using the image and the electronic map of the area, thereby improving the accuracy of obstacle prediction.

[0062] Optionally, the process of generating a high-precision map of a quasi-region of a preset range based on the image and the regional electronic map in step 303 may be:

[0063] The image is input into a preset target recognition model based on the attention mechanism to obtain environmental information within the preset range, including the identified road information within the preset range; the identified road information is matched with the road information in the regional electronic map to obtain a high-precision map of the quasi-region within the preset range.

[0064] The recognized road information refers to the road information around the vehicle extracted from the image. Since images can reflect the actual environment of the area, extracting this road information from the image and matching it with the road information in the electronic map can generate a high-precision map of the area, improving the accuracy of obstacle prediction.

[0065] Optionally, the process of determining the predicted trajectory of the obstacle based on the high-precision map of the class area, the obstacle type, the historical trajectory of the obstacle, and the planned path of the vehicle in step 303 may be:

[0066] The high-precision map of the class area, obstacle type, historical trajectory of the obstacle and vehicle planned path are input into the preset trajectory prediction model based on the graph network to obtain the movement intention of the obstacle and the predicted trajectory of the obstacle.

[0067] The movement intentions of obstacles include: going straight, changing lanes, and turning around.

[0068] Among them, the graph network is a set of functions organized according to the graph structure in the topological space for relational reasoning. In deep learning theory, it is a generalization of graph neural networks and probabilistic graph models. Compared with convolutional neural networks and recurrent neural networks, graph networks can better combine local and global information, so that the algorithm can better consider the relationship between surrounding obstacles, and at the same time has the characteristics of simple structure. These characteristics make the deep learning algorithm used in the present invention have the characteristics of high precision, high speed and low computing power, making it possible to deploy L4 level autonomous driving prediction modules under various platform computing powers.

[0069] In actual implementation, after obtaining a high-precision map of the area, obstacle type, obstacle history, and the vehicle's planned path, this information is uniformly encoded and input into a graph network-based trajectory prediction model. This allows prediction of the obstacle's intention and future trajectory. Because the behavior of an obstacle significantly influences the future driving intent of surrounding obstacles, this encoding method, similar to human driving habits, not only considers the interaction between obstacles but also their environment and history, providing better future predictions.

[0070] Optionally, after determining the predicted trajectory of the obstacle based on the high-precision map of the class area, the obstacle type, the historical trajectory of the obstacle, and the planned path of the vehicle in step 303, the predicted trajectory needs to be extended to obtain the target trajectory of the obstacle to generate a trajectory that meets the trajectory requirements of the downstream decision-making and planning module.

[0071] The target predicted trajectory can be the obstacle's trajectory in the next 7 to 9 seconds. That is, after obtaining the predicted trajectory of the obstacle, it is necessary to extend the predicted trajectory to obtain the target trajectory of the obstacle.

[0072] There are two ways to generate target trajectories: trajectory extension based on high-precision maps and trajectory extension based on electronic maps.

[0073] In one possible implementation, if a high-precision map of the area within a preset range is not obtained, and only an electronic map of the area within the preset range is obtained, then when obtaining the position information of the predicted trajectory, if the position information indicates that the predicted trajectory of the obstacle is located in the road area, the target trajectory of the obstacle is generated according to the lane direction of the lane to which the predicted trajectory belongs; if the position information indicates that the predicted trajectory of the obstacle is located in the intersection area, the target trajectory of the obstacle is generated according to the movement intention of the obstacle and the predicted trajectory of the obstacle.

[0074] In another possible implementation, if a high-precision map of a region within a preset range is obtained, the target lane to which the predicted trajectory belongs in the high-precision map is determined, and a target trajectory of the obstacle is generated according to the lane direction of the target lane.

[0075] In the actual implementation process, Figure 2 The software architecture diagram provided is as follows. The prediction module includes a cache container submodule, an attention mechanism submodule, and an intent submodule. Figure 2 In the container representation, the attention mechanism submodule is Figure 2 In the expression of attention, the intention submodule is Figure 2 Intention is used in Chinese.

[0076] The cache container submodule primarily caches data received from other modules and calculates the data required by the attention mechanism submodule. The attention mechanism submodule generates a high-precision map of the pre-set region based on the image and regional electronic map. It also determines the predicted trajectory of the obstacle based on the high-precision map, obstacle type, historical obstacle trajectory, and the vehicle's planned path. The intention submodule then extends the predicted trajectory to obtain the target trajectory of the obstacle.

[0077] like Figure 4 As shown, an embodiment of the present application provides a device for predicting the motion trajectory of an obstacle, characterized in that it is located in a vehicle automatic driving system, the system also includes a planning module, a perception module, a positioning module and a map module, and the device includes:

[0078] The receiving module 11 is used to receive the types of obstacles and historical trajectories of each obstacle within a preset range around the vehicle sent by the perception module, receive the planned vehicle path sent by the planning module, and receive the vehicle position sent by the positioning module;

[0079] An acquisition module 12 is configured to acquire a map of a preset area from a map module according to the vehicle's location. The map module includes a high-precision map and an electronic map.

[0080] The processing module 13 is configured to obtain an image of the preset range from the perception module if an electronic map of the area within the preset range is obtained but a high-precision map of the area within the preset range is not obtained, generate a high-precision map of the quasi-area within the preset range based on the image and the electronic map of the area, and determine a predicted trajectory of the obstacle based on the high-precision map of the quasi-area, the obstacle type, the historical trajectory of the obstacle, and the planned path of the vehicle.

[0081] In one embodiment, the processing module 13 is specifically configured to:

[0082] Input the image into a preset object recognition model based on the attention mechanism to obtain environmental information within a preset range, including recognized road information within the preset range;

[0083] The identified road information is matched with the road information in the regional electronic map to obtain a high-precision map of the pre-set area.

[0084] In one embodiment, the processing module 13 is specifically configured to:

[0085] The high-precision map of the class area, obstacle type, historical trajectory of the obstacle and vehicle planned path are input into the preset trajectory prediction model based on the graph network to obtain the movement intention of the obstacle and the predicted trajectory of the obstacle.

[0086] In one embodiment, the apparatus further includes a generating module 14, the generating module 14 being configured to:

[0087] Obtaining position information of the predicted trajectory, and if the position information indicates that the predicted trajectory of the obstacle is located in the road area, generating a target trajectory of the obstacle based on the lane direction of the lane to which the predicted trajectory belongs;

[0088] If the position information indicates that the predicted trajectory of the obstacle is located in the intersection area, a target trajectory of the obstacle is generated according to the movement intention of the obstacle and the predicted trajectory of the obstacle.

[0089] In one embodiment, the processing module 13 is further configured to:

[0090] If a high-precision map of the area within the preset range is obtained, the predicted trajectory of the obstacle is determined based on the high-precision map of the area, the obstacle type, the historical trajectory of the obstacle, and the planned path of the vehicle.

[0091] In one embodiment, the generating module 14 is further configured to:

[0092] Determine the target lane to which the predicted trajectory belongs in the high-precision map, and generate the target trajectory of the obstacle based on the lane direction of the target lane.

[0093] In one embodiment, the acquisition module 12 is specifically configured to:

[0094] Sending a data acquisition request for a regional map to a map module, so that the map module calls a first search engine corresponding to the high-precision map to determine a high-precision map of the region, or, if the first search engine fails to determine a high-precision map of the region, calls a second search engine corresponding to the electronic map to determine an electronic map of the region;

[0095] Receive the regional high-precision map or regional electronic map sent by the map module.

[0096] The obstacle motion trajectory prediction device provided in this embodiment can execute the above method embodiment. Its implementation principle and technical effects are similar and will not be elaborated here.

[0097] The specific definitions of the obstacle trajectory prediction device can be found in the definitions of the obstacle trajectory prediction method above and will not be repeated here. The various modules in the above-mentioned obstacle trajectory prediction device can be implemented in whole or in part through software, hardware, or a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the server in hardware form, or can be stored in the memory of the server in software form, so that the processor can call and execute the corresponding operations of each of the above modules.

[0098] In another embodiment of the present application, a computer device is provided, including a memory and a processor. The memory stores a computer program, and when the computer program is executed by the processor, the steps of the obstacle motion trajectory prediction method in the embodiment of the present application are implemented.

[0099] In another embodiment of the present application, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method for predicting the obstacle motion trajectory in the embodiment of the present application are implemented.

[0100] In another embodiment of the present application, a computer program product is further provided, which includes computer instructions. When the computer instructions are executed on a device for predicting an obstacle's motion trajectory, the device for predicting an obstacle's motion trajectory executes each step of the method for predicting an obstacle's motion trajectory in the method flow shown in the above-mentioned method embodiment.

[0101] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using a software program, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer execution instructions are loaded and executed on a computer, the process or function according to the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from a website, computer, server or data center to another website, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains one or more media that can be integrated. The available medium may be a magnetic medium (eg, a floppy disk, a hard disk, a magnetic tape), an optical medium (eg, a DVD), or a semiconductor medium (eg, a solid state disk (SSD)).

[0102] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0103] The above embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art could make various modifications and improvements without departing from the spirit of the present application, all of which fall within the scope of protection of the present application. Therefore, the scope of protection of the present patent application shall be determined by the appended claims.

Claims

1. A method for predicting obstacle motion trajectory, characterized in that: A prediction module is applied to an autonomous driving system for a vehicle, wherein the system further includes a perception module, a planning module, a positioning module, and a mapping module. The method includes: Receiving the types of obstacles and historical trajectories of obstacles within a preset range around the vehicle from the perception module, receiving the planned vehicle path from the planning module, and receiving the vehicle position from the positioning module; Acquire a map of the area within the preset range from the map module according to the vehicle position, wherein the map module includes a high-precision map and an electronic map; If an electronic map of the area within the preset range is obtained but a high-precision map of the area within the preset range is not obtained, an image of the preset range is obtained from the perception module, and a high-precision map of the quasi-area of ​​the preset range is generated based on the image and the electronic map of the area, and a predicted trajectory of the obstacle is determined based on the high-precision map of the quasi-area, the obstacle type, the historical trajectory of the obstacle, and the planned path of the vehicle.

2. The method according to claim 1, characterized in that Generating a high-precision map of the quasi-region of the preset range based on the image and the regional electronic map includes: Inputting the image into a preset object recognition model based on an attention mechanism to obtain environmental information within the preset range, wherein the environmental information includes recognized road information within the preset range; The identified road information is matched with the road information in the regional electronic map to obtain a high-precision map of the class area within the preset range.

3. The method according to claim 1, characterized in that The determining the predicted trajectory of the obstacle based on the high-precision map of the class area, the obstacle type, the historical trajectory of the obstacle, and the planned path of the vehicle includes: The high-precision map of the class area, the obstacle type, the historical trajectory of the obstacle and the planned path of the vehicle are input into a preset trajectory prediction model based on a graph network to obtain the movement intention of the obstacle and the predicted trajectory of the obstacle.

4. The method according to claim 3, characterized in that After determining the predicted trajectory of the obstacle based on the high-precision map of the class area, the obstacle type, the historical trajectory of the obstacle, and the planned path of the vehicle, the method further includes: obtaining position information of the predicted trajectory, and if the position information indicates that the predicted trajectory of the obstacle is located in a road area, generating a target trajectory of the obstacle according to a lane direction of a lane to which the predicted trajectory belongs; If the position information indicates that the predicted trajectory of the obstacle is located in an intersection area, a target trajectory of the obstacle is generated according to the movement intention of the obstacle and the predicted trajectory of the obstacle.

5. The method according to claim 1, wherein After obtaining the area map of the preset range from the map module according to the vehicle position, the method further includes: If a high-precision map of the area within the preset range is obtained, the predicted trajectory of the obstacle is determined based on the high-precision map of the area, the obstacle type, the historical trajectory of the obstacle, and the planned path of the vehicle.

6. The method according to claim 5, characterized in that After determining the predicted trajectory of the obstacle based on the high-precision map of the area, the obstacle type, the historical trajectory of the obstacle, and the planned path of the vehicle, the method further includes: Determine the target lane to which the predicted trajectory belongs in the high-precision map, and generate a target trajectory of the obstacle based on the lane direction of the target lane.

7. The method according to claim 1, characterized in that The obtaining of the area map of the preset range from the map module according to the vehicle position includes: Sending a data acquisition request for the regional map to the map module, so that the map module calls a first search engine corresponding to the high-precision map to determine the high-precision map of the region, or, if the first search engine fails to determine the high-precision map of the region, calls a second search engine corresponding to the electronic map to determine the electronic map of the region; Receive the high-precision map of the area or the electronic map of the area sent by the map module.

8. A device for predicting the trajectory of an obstacle, characterized in that: Located in a vehicle autonomous driving system, the system further includes a planning module, a perception module, a positioning module, and a map module, and the device includes: a receiving module, configured to receive the types of obstacles and historical trajectories of obstacles within a preset range around the vehicle sent by the perception module, receive the planned vehicle path sent by the planning module, and receive the vehicle position sent by the positioning module; an acquisition module, configured to acquire a map of the area within the preset range from the map module according to the vehicle position, wherein the map module includes a high-precision map and an electronic map; a processing module configured to obtain an image of the preset range from the perception module if an electronic map of the area within the preset range is obtained but a high-precision map of the area within the preset range is not obtained, generate a high-precision map of a class area of ​​the preset range based on the image and the electronic map of the area, and determine a predicted trajectory of the obstacle based on the high-precision map of the class area, the obstacle type, the historical trajectory of the obstacle, and the planned path of the vehicle.

9. An electronic device, characterized in that: The method comprises a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the method for predicting the obstacle motion trajectory according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that A computer program is stored thereon, and when the computer program is executed by a processor, the method for predicting the obstacle motion trajectory according to any one of claims 1 to 7 is implemented.

Images