Autonomous driving scenario obstacle prediction

Abstract

The present disclosure relates to autonomous driving scenario obstacle prediction, which is implemented by means of a trained prediction model. The trained prediction model comprises a global encoder, a first local encoder, and a first decoder. The method comprises: acquiring state information and environment information of each obstacle in a target scenario from the perspective of a vehicle; inputting the state information and the environment information of each obstacle in the target scenario into the global encoder to obtain first feature information; acquiring prompt information of a target obstacle; inputting the prompt information of the target obstacle into the first local encoder to obtain second feature information; and taking the first feature information as input information of the first decoder, taking the second feature information as prompt information of the first decoder, and inputting the input information and the prompt information into the first decoder to obtain prediction information of the target obstacle.

Description

Obstacle prediction in autonomous driving scenarios Technical Field

[0001] The present disclosure relates to artificial intelligence technology, and more specifically, to obstacle prediction in autonomous driving scenarios. Background Art

[0002] In the prior art, the encoder used for trajectory prediction is often an encoder centered around the predicted target obstacle. The encoder models the obstacle information and scene information around each predicted target obstacle.

[0003] Summary of the Invention

[0004] One objective of the present disclosure is to provide a new technical solution for obstacle prediction in autonomous driving scenarios.

[0005] According to a first aspect of the present disclosure, there is provided an autonomous driving scenario obstacle prediction method, which is implemented by a trained prediction model, wherein the trained prediction model includes a global encoder, a first local encoder, and a first decoder. The method includes:

[0006] Obtaining state information and environmental information of each obstacle in a target scene from the perspective of the autonomous driving vehicle; wherein the target obstacle is at least one obstacle in the target scene;

[0007] Inputting state information and environmental information of each obstacle in the target scene into the global encoder to obtain first feature information;

[0008] Get prompt information of target obstacles;

[0009] Inputting the prompt information of the target obstacle into the first local encoder to obtain second feature information;

[0010] The first feature information is used as input information of the first decoder, and the second feature information is used as prompt information of the first decoder, and input into the first decoder to obtain prediction information of the target obstacle.

[0011] Optionally, the global encoder is used to construct feature information based on the state information and environmental information of each obstacle in the target scene from the perspective of the autonomous driving vehicle, with the autonomous driving vehicle as the center, and the first local encoder is used to construct feature information based on the prompt information of the target obstacle obtained from the perspective of the autonomous driving vehicle.

[0012] Optionally, the state information of each obstacle includes position information and motion information of each obstacle, and the environmental information includes lane information and traffic light information.

[0013] Optionally, the prompt information of the target obstacle includes prediction task type information of the target obstacle and category information of the target obstacle.

[0014] Optionally, inputting the state information and environment information of each obstacle in the target scene into the global encoder to obtain the first feature information includes:

[0015] Inputting the state information and environmental information of each obstacle in the target scene into the global encoder to obtain a global feature vector and a position vector; wherein the position vector includes the position information corresponding to each eigenvalue in the global feature vector;

[0016] The global feature vector and the position vector are added to obtain the first feature information.

[0017] Optionally, the trained prediction model further includes a second local encoder and a second decoder; wherein,

[0018] Obtain prompt information of a second prediction task; wherein the first prediction task is a prediction task in the obstacle prediction, and the first prediction task and the second prediction task are different prediction tasks;

[0019] Inputting the prompt information of the second prediction task into the second local encoder to obtain third feature information;

[0020] The first feature information is used as input information of the second decoder, and the third feature information is used as prompt information of the second decoder, and input into the second decoder to obtain prediction result information of the second prediction task.

[0021] Optionally, the first prediction task and the second prediction task are any one of a trajectory prediction task, an intention understanding task, an intelligent simulation task, and a path planning task.

[0022] Optionally, the embodiment further includes:

[0023] When the prediction model is trained using the first training sample to obtain a trained global encoder, a trained first local encoder, and a trained first decoder, freezing the trained global encoder;

[0024] Only the second local encoder and the second decoder are trained using the second training sample to obtain a trained second local encoder and a trained second decoder.

[0025] According to a second aspect of the present disclosure, an electronic device is provided, comprising a memory and a processor, wherein the memory stores computer instructions, and when the computer instructions are executed by the processor, the method for predicting obstacles in an autonomous driving scene according to any one of the first aspects is implemented.

[0026] According to a third aspect of the present disclosure, a computer-readable storage medium is provided, on which computer instructions are stored. When the computer instructions are executed by a processor, the method for predicting obstacles in an autonomous driving scene according to any one of the first aspects is implemented.

[0027] The obstacle prediction for autonomous driving scenarios provided by the present disclosure uniformly models the information of each obstacle and the environmental information in the target scene. This information is converted into unbiased features through a global encoder. Unbiased here means that the global encoder can extract corresponding feature values ​​from each input information. The global encoder can be shared for the prediction of each target obstacle, eliminating the problem of repeated modeling of the same information, solving the problem of large space storage occupation, improving training speed, and improving the joint modeling capability of multiple target obstacle prediction trajectories.

[0028] Features and advantages of the embodiments of the present specification will become apparent from the following detailed description of exemplary embodiments of the present specification with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments of the specification and, together with the description, serve to explain the principles of the embodiments of the specification.

[0030] FIG1 is a schematic diagram of a hardware configuration of an electronic device that can be used to implement an embodiment of the present disclosure, provided by an embodiment of the present disclosure;

[0031] FIG2 is a block diagram of a trained prediction model according to an embodiment of the present disclosure;

[0032] FIG3 is a processing flow chart of a method for predicting obstacles in an autonomous driving scenario according to one embodiment of the present disclosure;

[0033] FIG4 is a block diagram of a trained prediction model according to an embodiment of the present disclosure;

[0034] FIG5 is a block diagram of a trained prediction model according to an embodiment of the present disclosure;

[0035] FIG6 is a structural block diagram of an electronic device according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0036] Various exemplary embodiments of the present disclosure will now be described in detail with reference to the accompanying drawings.

[0037] The following will be combined with the accompanying drawings in the embodiments of this application to clearly describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are 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 are within the scope of protection of this application.

[0038] The terms "first," "second," and the like in the specification and claims of this application are used to distinguish similar objects, and are not used to describe a specific order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate, so that the embodiments of this application can be implemented in an order other than that illustrated or described herein, and that the objects distinguished by "first," "second," and the like are generally of the same type, and do not limit the number of objects; for example, the first object can be one or more. In addition, the term "and / or" in the specification and claims represents at least one of the connected objects, and the character " / " generally indicates that the objects associated with each other are in an "or" relationship.

[0039] It should be noted that all actions of acquiring signals, information or data in this application are carried out in compliance with the relevant data protection laws and policies of the country where they are located and with the authorization given by the owner of the corresponding device.

[0040] <Hardware Configuration>

[0041] FIG1 is a schematic diagram of the structure of an electronic device that can be used to implement an embodiment of the present disclosure. The electronic device can be used to implement the method for predicting obstacles in an autonomous driving scenario according to an embodiment of the present disclosure.

[0042] The electronic device 1000 can be a server, a smart phone, a portable computer, a desktop computer, a tablet computer, etc., which is not limited here.

[0043] The electronic device 1000 may include, but is not limited to, a processor 1100, a memory 1200, an interface device 1300, a communication device 1400, a display device 1500, an input device 1600, a speaker 1700, a microphone 1800, and the like. The processor 1100 may be a central processing unit (CPU), a graphics processing unit (GPU), a microprocessor (MCU), and the like, and is configured to execute computer programs / instructions, which may be written using an instruction set such as an x86, Arm, RISC, MIPS, or SSE architecture. The memory 1200 may include, for example, ROM (read-only memory), RAM (random access memory), or a non-volatile memory such as a hard disk. The interface device 1300 may include, for example, a USB interface, a serial interface, or a parallel interface. The communication device 1400 may be capable of wired communication using, for example, optical fiber or cable, or wireless communication, and may specifically include WiFi, Bluetooth, 2G / 3G / 4G / 5G, and the like. The display device 1500 may be, for example, an LCD display or a touch screen display. The input device 1600 may include, for example, a touch screen, a keyboard, a somatosensory input device, etc. The speaker 1700 is used to output audio signals. The microphone 1800 is used to collect audio signals.

[0044] As used in the embodiment of the present disclosure, the memory 1200 of the electronic device 1000 is used to store computer programs / instructions, which are used to control the processor 1100 to operate to implement the automatic driving scene obstacle prediction method according to the embodiment of the present disclosure. Technicians can design the computer program / instructions based on the scheme disclosed in the present disclosure. How the computer program / instructions control the processor to operate is well known in the art and will not be described in detail here. The electronic device 1000 can be installed with an intelligent operating system (such as Windows, Linux, Android, IOS, etc.) and application software.

[0045] Those skilled in the art will appreciate that, although FIG1 shows multiple devices of the electronic device 1000 , the electronic device 1000 of the present disclosure may only involve some of the devices, for example, only the processor 1100 and the memory 1200 .

[0046] Existing technology modeling processes are redundant, computationally complex, slow to train, and require large amounts of storage space. Furthermore, each modeling process, for example, involves repeated modeling of the same lane, preventing the reuse of this information across different predicted target obstacles.

[0047] Hereinafter, various embodiments and examples of the fundamental disclosure are described with reference to the accompanying drawings.

[0048] <Method Example>

[0049] In this embodiment, a method for predicting obstacles in an autonomous driving scenario is provided. The method is implemented using a trained prediction model.

[0050] The trained prediction model architecture is shown in Figure 2. As shown in Figure 2, the trained prediction model includes a global encoder, a first local encoder, and a first decoder.

[0051] The global encoder is used to construct feature information based on the state information and environmental information of each obstacle in the target scene from the perspective of the autonomous vehicle. The first local encoder is used to construct feature information based on the prompt information of the target obstacle obtained from the perspective of the autonomous vehicle.

[0052] As shown in FIG3 , the method for predicting obstacles in an autonomous driving scene of this embodiment may include the following steps S310 to S350 .

[0053] Step S310: Acquire status information and environmental information of each obstacle in the target scene from the perspective of the autonomous driving vehicle; wherein the target obstacle is at least one obstacle in the target scene.

[0054] The status information and environmental information of each obstacle can be collected by the equipment installed on the vehicle.

[0055] In one embodiment, the status information of each obstacle includes the position information and motion information of each obstacle. The position information of each obstacle is the position information in the coordinate system established with the autonomous driving vehicle as the origin. The position information of each obstacle includes the current position information and historical position information of each obstacle. The historical position information is the position information of the obstacle at the moment before the current moment, and may also be the position information of the obstacle at multiple historical moments in the time period before the current moment. The motion information of each obstacle includes the distance between each obstacle and the autonomous driving vehicle, the speed of each obstacle, and the acceleration of each obstacle. The motion information of each obstacle includes the current motion information and historical motion information. The historical motion information is the motion information of the obstacle at the moment before the current moment, and may also be the motion information of the obstacle at multiple historical moments in the time period before the current moment.

[0056] Obstacles include movable obstacles and stationary obstacles.

[0057] Environmental information includes lane information and traffic light information. Lane information includes the lane information, isolation belt information, and sidewalk information of the autonomous vehicle.

[0058] Step S320: Input the state information and environment information of each obstacle in the target scene into the global encoder to obtain first feature information.

[0059] In one embodiment, step S320 includes: inputting the state information and environmental information of each obstacle in the target scene into the global encoder to obtain a global feature vector and a position vector; wherein the position vector includes the position information corresponding to each eigenvalue in the global feature vector; adding the global feature vector and the position vector to obtain the first feature information.

[0060] Step S330: Obtain prompt information of the target obstacle.

[0061] Prompt information of target obstacles can be collected through equipment installed on the vehicle.

[0062] In one embodiment, the prompt information of the target obstacle includes the predicted task type information of the target obstacle and the category information of the target obstacle. The predicted task type information of the target obstacle can be any of trajectory prediction, intention prediction, intelligent simulation, and path planning. Trajectory prediction is the movement trajectory information of a movable obstacle at the next moment. Intent prediction is the movement intention of a stationary obstacle at the current moment at the next moment. The movement intention here can be stationary, or any of the actions of moving forward, backward, turning left, and turning right. The category information of the target obstacle can be any of motor vehicles, non-motor vehicles, pedestrians, and stationary obstacles.

[0063] The target obstacle may be one obstacle in the target scene or multiple obstacles in the target scene.

[0064] Step S340: Input the prompt information of the target obstacle into the first local encoder to obtain second feature information.

[0065] In step S350 , the first feature information is used as input information of the first decoder, and the second feature information is used as prompt information of the first decoder, and inputted into the first decoder to obtain prediction information of the target obstacle.

[0066] The first feature information represents the state and environmental information of the obstacles surrounding the target obstacle. The second feature information represents information about the target obstacle. Using the attention mechanism, feature information related to the predicted information about the target obstacle is obtained from the first and second feature information. Based on the feature information related to the predicted information about the target obstacle, the predicted information about the target obstacle is obtained.

[0067] In one embodiment, the prediction of obstacles in an autonomous driving scenario may be trajectory prediction. Thus, the predicted information of the target obstacle obtained based on the autonomous driving scenario obstacle prediction method is trajectory prediction information. The predicted trajectory information of each target obstacle may be one or more predicted trajectories.

[0068] The obstacle prediction method for autonomous driving scenarios provided by the present disclosure uniformly models the information of each obstacle and the environmental information in the target scene. This information is converted into unbiased features through a global encoder. Unbiased here means that the global encoder can extract corresponding feature values ​​from each input information. The global encoder can be shared for the prediction of each target obstacle, eliminating the problem of repeated modeling of the same information, solving the problem of large space storage usage, improving training speed, and improving the joint modeling capability of multiple target obstacle prediction trajectories.

[0069] The feature information constructed by the trained global encoder is unbiased and suitable not only for obstacle trajectory prediction, but also for prediction of other tasks, such as intent understanding, intelligent simulation, and path planning. This allows the global encoder to be shared when predicting different tasks, while the corresponding local encoders and decoders can be used to complete the tasks.

[0070] In one embodiment, referring to FIG4 , the trained prediction model further includes a second local encoder and a second decoder. The method further includes: obtaining prompt information for a second prediction task, inputting the prompt information for the second prediction task into the second local encoder to obtain third feature information, using the first feature information as input information for a second decoder, and inputting the third feature information as prompt information for the second decoder into the second decoder to obtain prediction result information for the second prediction task. The first prediction task is a prediction task in obstacle prediction, and the first prediction task and the second prediction task are different prediction tasks.

[0071] The second local encoder is used to construct feature information based on prompt information of the second prediction task obtained from the perspective of the autonomous driving vehicle.

[0072] In one embodiment, the method further includes: when the prediction model is trained using the first training sample to obtain a trained global encoder, a trained first local encoder and a trained first decoder, freezing the trained global encoder, and training only the second local encoder and the second decoder using the second training sample to obtain a trained second local encoder and a trained second decoder.

[0073] Freezing the trained global encoder freezes the parameters that were optimized during the training of the global encoder. When the local encoder and the second decoder are trained using the second training sample, the global encoder no longer participates in the training.

[0074] Since the global encoder converts input information into unbiased features, the output information of the global encoder can be shared with different tasks to share the global encoder.

[0075] In one embodiment, the prediction method for the first prediction task and the prediction method for the second prediction task are described in conjunction with Figure 5. The first prediction task is obstacle trajectory prediction. The second prediction task is a path planning task.

[0076] First, the status information of movable obstacles, the status information of stationary obstacles, lane information, and traffic light information in the target scene from the vehicle's perspective are obtained.

[0077] Next, the state information of movable obstacles, stationary obstacles, lane information, and traffic light information in the target scene are input into the global encoder to generate a global feature vector and a position vector. The position vector includes the position information corresponding to each eigenvalue in the global feature vector. The global feature vector and the position vector are added together to generate the first feature information.

[0078] Obtain prompt information for multiple target obstacles. Each target obstacle prompt information includes the target obstacle's prediction task type and target obstacle category information. Each target obstacle prompt information is input into a first local encoder to obtain multiple second feature information. The three target obstacle prompts shown in Figure 5 are merely examples and do not limit the present disclosure in any way.

[0079] The first feature information is used as input information of the first decoder, and each second feature information is used as prompt information of the first decoder, and is input into the first decoder to obtain the predicted trajectory information of each target obstacle.

[0080] Obtain prompt information for the path planning task, input the prompt information of the path planning task into the second local encoder, obtain third feature information, use the first feature information as input information of the second decoder, and use the third feature information as prompt information of the second decoder, input it into the second decoder, and obtain prediction result information of the path planning task.

[0081] The above-mentioned obstacle trajectory prediction task and vehicle path planning task can be performed simultaneously or separately according to actual needs, without any limitation here.

[0082] <Device Example>

[0083] FIG6 is a schematic diagram of an electronic device 600 provided by one embodiment of the present disclosure. The electronic device 600 includes a processor 610 and a memory 620. The memory 620 stores computer instructions that, when executed by the processor 610, implement the obstacle prediction method for autonomous driving scenarios disclosed in any of the aforementioned embodiments.

[0084] The embodiments of the present disclosure also provide a computer-readable storage medium having computer instructions stored thereon. When the computer instructions are executed by a processor, the method for predicting obstacles in autonomous driving scenarios disclosed in any of the aforementioned embodiments is implemented.

[0085] <Vehicle Example>

[0086] An embodiment of the present disclosure further provides a vehicle, which includes the electronic device disclosed in any of the above embodiments.

[0087] The various embodiments of this disclosure are described in a progressive manner. Similar portions between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from the other embodiments. In particular, the device and apparatus embodiments are generally similar to the method embodiments, so their descriptions are relatively simple. For relevant portions, reference can be made to the descriptions of the method embodiments.

[0088] The foregoing description describes specific embodiments of the present disclosure. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims can be performed in an order different from that described in the embodiments and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order shown or the sequential order to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0089] The embodiments of the present disclosure may be systems, methods, and / or computer program products. The computer program product may include a computer-readable storage medium carrying computer-readable program instructions for causing a processor to implement various aspects of the embodiments of the present disclosure.

[0090] A computer-readable storage medium can be a tangible device that can hold and store instructions for use by an instruction execution device. A computer-readable storage medium can be, for example, but not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination thereof. More specific examples (a non-exhaustive list) of computer-readable storage media include: 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), a static random access memory (SRAM), a portable compact disc read-only memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, a mechanical encoding device, such as a punch card or a raised structure in a groove on which instructions are stored, and any suitable combination thereof. As used herein, a computer-readable storage medium is not to be construed as a transient signal per se, such as a radio wave or other freely propagating electromagnetic wave, an electromagnetic wave propagating through a waveguide or other transmission medium (e.g., a light pulse through a fiber optic cable), or an electrical signal transmitted through an electrical wire.

[0091] The computer-readable program instructions described herein can be downloaded from a computer-readable storage medium to each computing / processing device, or downloaded to an external computer or external storage device via a network, such as the Internet, a local area network, a wide area network, and / or a wireless network. The network can include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers, and / or edge servers. The network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards the computer-readable program instructions to be stored in the computer-readable storage medium in each computing / processing device.

[0092] The computer program instructions for performing the operation of the embodiments of the present disclosure can be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-related instructions, microcode, firmware instructions, state setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Smalltalk, C++, and conventional procedural programming languages ​​such as "C" language or similar programming languages. Computer-readable program instructions can be executed entirely on a user's computer, partially on a user's computer, executed as an independent software package, partially on a user's computer and partially on a remote computer, or completely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer via 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., using an Internet service provider to connect via the Internet). In some embodiments, by utilizing the state information of computer-readable program instructions to personalize an electronic circuit, such as a programmable logic circuit, a field programmable gate array (FPGA), or a programmable logic array (PLA), the electronic circuit can execute computer-readable program instructions, thereby realizing the various aspects of the embodiments of the present disclosure.

[0093] Various aspects of the embodiments of the present disclosure are described herein with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present disclosure. It should be understood that each block of the flowcharts and / or block diagrams, and combinations of blocks in the flowcharts and / or block diagrams, can be implemented by computer-readable program instructions.

[0094] These computer-readable program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, thereby producing a machine, so that when these instructions are executed by the processor of the computer or other programmable data processing device, a device is generated that implements the functions / actions specified in one or more blocks in the flowchart and / or block diagram. These computer-readable program instructions can also be stored in a computer-readable storage medium, where these instructions cause the computer, programmable data processing device, and / or other device to operate in a specific manner. Thus, the computer-readable medium storing the instructions comprises an article of manufacture that includes instructions for implementing various aspects of the functions / actions specified in one or more blocks in the flowchart and / or block diagram.

[0095] Computer-readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device so that a series of operational steps are performed on the computer, other programmable data processing apparatus, or other device to produce a computer-implemented process, thereby causing the instructions executed on the computer, other programmable data processing apparatus, or other device to implement the functions / actions specified in one or more blocks in the flowchart and / or block diagram.

[0096] The flowcharts and block diagrams in the accompanying drawings show the possible architectures, functions and operations of the systems, methods and computer program products according to multiple embodiments of the present disclosure. In this regard, each box in the flowchart or block diagram can represent a module, program segment or part of an instruction, and a part of the module, program segment or instruction contains one or more executable instructions for implementing the specified logical function. In some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of boxes in the block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or action, or can be implemented using a combination of dedicated hardware and computer instructions. It is well known to those skilled in the art that implementation by hardware, implementation by software, and implementation by a combination of software and hardware are all equivalent.

[0097] While various embodiments of the present disclosure have been described above, the above descriptions are illustrative, non-exhaustive, and not intended to be limiting of the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope of the described embodiments. The terminology used herein is selected to best explain the principles of the embodiments, their practical applications, or improvements to existing technologies, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. A method for predicting obstacles in an autonomous driving scenario, wherein: This is achieved by a trained prediction model, wherein the trained prediction model includes a global encoder, a first local encoder, and a first decoder, and the method includes: Acquire state information and environment information of each obstacle in a target scene from the perspective of the autonomous driving vehicle; wherein the target obstacle is at least one obstacle in the target scene; Inputting state information and environment information of each obstacle in the target scene into the global encoder to obtain first feature information; Get prompt information of target obstacles; Inputting the prompt information of the target obstacle into the first local encoder to obtain second feature information; The first feature information is used as input information of the first decoder, and the second feature information is used as prompt information of the first decoder, and is input into the first decoder to obtain prediction information of the target obstacle.

2. The method according to claim 1, wherein: The global encoder is used to construct feature information based on the state information and environmental information of each obstacle in the target scene from the perspective of the autonomous driving vehicle, with the autonomous driving vehicle as the center. The first local encoder is used to construct feature information based on the prompt information of the target obstacle obtained from the perspective of the autonomous driving vehicle.

3. The method according to claim 1, wherein: The state information of each obstacle includes the position information and movement information of each obstacle, and the environmental information includes lane information and traffic light information.

4. The method according to claim 1, wherein: The prompt information of the target obstacle includes prediction task type information of the target obstacle and category information of the target obstacle.

5. The method according to claim 1, wherein: The step of inputting the state information and environment information of each obstacle in the target scene into the global encoder to obtain the first feature information includes: Inputting the state information and environment information of each obstacle in the target scene into the global encoder to obtain a global feature vector and a position vector; wherein the position vector includes the position information corresponding to each eigenvalue in the global feature vector; The global feature vector and the position vector are added to obtain the first feature information.

6. The method according to claim 1, wherein: The trained prediction model also includes a second local encoder and a second decoder; wherein, Obtain prompt information of a second prediction task; wherein the first prediction task is a prediction task in the obstacle prediction, and the first prediction task and the second prediction task are different prediction tasks; Inputting the prompt information of the second prediction task into the second local encoder to obtain third feature information; The first feature information is used as input information of the second decoder, and the third feature information is used as prompt information of the second decoder, and is input into the second decoder to obtain prediction result information of the second prediction task.

7. The method according to claim 6, wherein: The first prediction task and the second prediction task are any two of a trajectory prediction task, an intention understanding task, an intelligent simulation task, and a path planning task.

8. The method according to claim 6, wherein: The method further comprises: When the prediction model is trained by using the first training sample to obtain a trained global encoder, a trained first local encoder, and a trained first decoder, freezing the trained global encoder; Only the second local encoder and the second decoder are trained using the second training sample to obtain a trained second local encoder and a trained second decoder.

9. An electronic device, wherein: It includes a memory and a processor, the memory stores computer instructions, and the computer instructions, when executed by the processor, implement the automatic driving scene obstacle prediction method according to any one of claims 1-8.

10. A computer-readable storage medium, wherein: Computer instructions are stored thereon, and when the computer instructions are executed by a processor, the method for predicting obstacles in an autonomous driving scene according to any one of claims 1-8 is implemented.

Images