WO2023087626A1

WO2023087626A1 - Trajectory prediction method, trajectory prediction apparatus, and storage medium

Info

Publication number: WO2023087626A1
Application number: PCT/CN2022/090552
Authority: WO
Inventors: 段志祥; 杨奎元
Original assignee: 北京小米移动软件有限公司
Priority date: 2021-11-19
Filing date: 2022-04-29
Publication date: 2023-05-25
Also published as: CN114537432A

Abstract

A trajectory prediction method and apparatus, and a storage medium. The trajectory prediction method comprises: obtaining a map image, the map image comprising a lane center line, the lane center line having several nodes arranged at equal intervals, each node having a distance feature, and the distance feature of the node being fused with distance features of other lane center line nodes that have different distances from the node (S101); on the basis of the position of a vehicle in the map image, determining a lane center line node closest to the position of the vehicle as a target node (S102); performing vector merging on a distance feature of the target node and a historical trajectory feature of the vehicle to obtain a trajectory prediction vector (S103); and predicting a driving trajectory of the vehicle in the map image on the basis of the trajectory prediction vector (S104). According to the trajectory prediction method, the distance feature and a historical trajectory of the vehicle are merged to obtain the trajectory prediction vector, so as to predict the driving trajectory of the vehicle on the basis of the trajectory prediction vector, such that the topological structure of a lane line in the map image can be reserved, and the accuracy of trajectory prediction is improved.

Description

Trajectory prediction method, trajectory prediction device and storage medium

Cross References to Related Applications

This disclosure claims the priority of a Chinese patent application with application number 202111401099.2 and titled "Track Prediction Method, Trajectory Prediction Device, and Storage Medium" filed with the China Patent Office on November 19, 2021, the entire contents of which are incorporated herein by reference. In public.

technical field

The present disclosure relates to the technical field of intelligent driving, and in particular to a trajectory prediction method, a trajectory prediction device and a storage medium.

Background technique

In the fields of robot navigation, automatic driving, etc., forecasting is made in the dynamically changing traffic environment, the driving route of the physical object is planned, and the driving of the physical object is guided.

The movement trajectory of the current object will be affected by the behavior of other entity objects in the traffic scene. Therefore, in the process of guiding the driving behavior of the entity object, it is necessary to predict the driving trajectory of the surrounding entities, as the basis for the driving guidance of the current object, and reasonably avoid obstacles for the driving of the current object.

Contents of the invention

In order to overcome the problems existing in the related technologies, the present disclosure provides a trajectory prediction method, a trajectory prediction device and a storage medium.

According to the first aspect of an embodiment of the present disclosure, a trajectory prediction method is provided, the trajectory prediction method includes: acquiring a map image, the map image includes a lane centerline, and the lane centerline has several nodes arranged at equal intervals , each node has a distance feature, and the distance features of the node are fused with the distance features of other lane centerline nodes with different distances from the node; based on the position of the vehicle in the map image, determine the The lane centerline node with the closest position is used as the target node; the distance feature of the target node is vector-merged with the historical track feature of the vehicle to obtain a trajectory prediction vector; based on the trajectory prediction vector, it is predicted that the vehicle is at driving track in the above map image.

In some embodiments, the distance feature of each node is determined in the following manner: the first node is determined on the map image; based on the position of the first node, an area map image is determined, and the area map image includes The first node, and a plurality of second nodes having different distances from the first node; based on the distances between the plurality of second nodes and the first node, extracting the plurality of second nodes distance features of two nodes, and combine the distance features of the plurality of second nodes to obtain the distance features of the first node.

In some embodiments, based on the distance between the plurality of second nodes and the first node, the distance features of the plurality of second nodes are extracted, and the distance features of the plurality of second nodes are performed Merge to obtain the distance feature of the first node, including: dividing the plurality of second nodes into multiple categories according to the distance from the first node, wherein the second nodes in the same category The same as the distance between the first nodes; respectively extracting the distance features of the second nodes of the same category in the multiple categories, and splicing the extracted distance features of different categories to obtain the distance features of the first node distance feature.

In some embodiments, determining an area map image based on the position of the first node includes: centering on the first node, determining an area map image with a preset range, within the preset range at least including a second node directly adjacent to the first node, and a second node indirectly adjacent to the first node.

In some embodiments, extracting the distance features of the plurality of second nodes includes: extracting the distance features of the plurality of second nodes based on a deep learning model of point cloud data.

In some embodiments, the characteristics of the historical trajectory of the vehicle are determined in the following manner: acquiring the historical trajectory of the vehicle; extracting the characteristics of the historical trajectory based on a long-short-term memory regression neural network to obtain the characteristics of the historical trajectory of the vehicle.

According to the second aspect of the embodiments of the present disclosure, there is provided a trajectory prediction device. The trajectory prediction device includes: an acquisition unit, configured to acquire a map image, the map image includes a lane centerline, and the lane centerline has an equal interval Several nodes are set, each node has a distance feature, and the distance feature of the node is fused with the distance features of other lane centerline nodes having different distances from the node; the determination unit is used for position in the image, determine the lane centerline node closest to the vehicle position as the target node, and combine the distance feature of the target node with the historical trajectory feature of the vehicle to obtain a trajectory prediction vector; prediction unit , for predicting the driving trajectory of the vehicle in the map image based on the trajectory prediction vector.

In some embodiments, the determining unit extracts the distance features of the plurality of second nodes based on the distance between the plurality of second nodes and the first node in the following manner, and the plurality of Merge the distance features of the second node to obtain the distance feature of the first node: divide the plurality of second nodes into multiple categories according to the distance from the first node, wherein the same category The distance between the second node and the first node is the same; the distance features of the second nodes of the same category in the multiple categories are respectively extracted, and after the extracted distance features of different categories are spliced, the obtained Describe the distance feature of the first node.

In some embodiments, the determination unit determines an area map image based on the position of the first node in the following manner: with the first node as the center, an area map image with a preset range is determined, and in the preset It is assumed that the range includes at least a second node directly adjacent to the first node, and a second node indirectly adjacent to the first node.

In some embodiments, the determining unit extracts the distance features of the plurality of second nodes in the following manner: extracts the distance features of the plurality of second nodes based on a deep learning model of point cloud data.

According to a third aspect of an embodiment of the present disclosure, there is provided a trajectory prediction device, including: a processor; a memory for storing processor-executable instructions; wherein the processor is configured to: execute the trajectory described in any one of the foregoing method of prediction.

According to the fourth aspect of the embodiments of the present disclosure, there is provided a non-transitory computer-readable storage medium. When the instructions in the storage medium are executed by the processor of the mobile terminal, the mobile terminal can execute the trajectory described in any one of the foregoing method of prediction.

According to a fifth aspect of the embodiments of the present disclosure, there is provided a computer program, including computer readable code, which, when the computer readable code is run on a computing processing device, causes the computing processing device to execute the implementation of the first aspect of the present disclosure. Example of the proposed trajectory prediction method.

Additional aspects and advantages of the disclosure will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the disclosure.

Description of drawings

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

Fig. 1 is a flow chart showing a trajectory prediction method according to an exemplary embodiment of the present disclosure.

Fig. 2 is a schematic diagram of lane centerlines and nodes in a map image according to an exemplary embodiment of the present disclosure.

Fig. 3 is a flowchart showing a method for determining a distance feature of a node according to an exemplary embodiment of the present disclosure.

Fig. 4 shows a method of extracting distance features of multiple second nodes based on the distance between multiple second nodes and the first node according to an exemplary embodiment of the present disclosure, and combining the distance features of multiple second nodes Merging is performed to obtain the flow chart of the distance feature method of the first node.

Fig. 5 is a schematic diagram illustrating determining a distance feature of a first node according to an exemplary embodiment of the present disclosure.

Fig. 6 is a flow chart showing a method for determining a distance feature of a node according to an exemplary embodiment of the present disclosure.

Fig. 7 is a flow chart showing a method for determining historical trajectory characteristics of a vehicle according to an exemplary embodiment of the present disclosure.

Fig. 8 is a block diagram of a trajectory prediction device according to an exemplary embodiment of the present disclosure.

Fig. 9 is a block diagram showing a device for trajectory prediction according to an exemplary embodiment of the present disclosure.

Detailed ways

Reference will now be made in detail to the exemplary embodiments, examples of which are illustrated in the accompanying drawings. When the following description refers to the accompanying drawings, the same numerals in different drawings refer to the same or similar elements unless otherwise indicated. The implementations described in the following exemplary examples do not represent all implementations consistent with the present disclosure. Rather, they are merely examples of apparatuses and methods consistent with aspects of the present disclosure as recited in the appended claims.

In the field of autonomous driving, it is necessary to predict the motion of the self and surrounding moving objects. Since the autonomous vehicle will inevitably interact with other traffic participants during actual road driving, the driving strategy planning of the autonomous vehicle is closely related to the driving of surrounding vehicles. Behavior is closely related. When planning a driving strategy, predict the behavioral intentions of moving objects in the surrounding range for a period of time in the future, and convert the predicted results into trajectories of time and space dimensions. The predicted trajectories of moving objects such as obstacle vehicles, pedestrians, and non-motorized vehicles are used as input to enable self-driving vehicles to make reasonable driving decisions and plan reasonable and safe vehicle movement behaviors. For example, if it is predicted that other vehicles will merge into the lane where the current vehicle is located, deceleration needs to be considered in advance. The higher the accuracy of the prediction, the more accurate the decision-making, and the higher the reliability of unmanned driving.

Motion prediction combines the attribute information of moving objects, historical running track information and high-precision map information to give the prediction of the moving behavior of moving objects in the future. Rendering the map into an overhead view image will lose the topological structure of the map, and cannot reflect the difference between the opposite lane and the lane where the vehicle is located, which will affect the accuracy of trajectory prediction.

Therefore, the present disclosure provides a trajectory prediction method, which can preserve the topological structure of the lane lines in the map image where the vehicle is located, thereby improving the accuracy of trajectory prediction.

Fig. 1 is a flow chart showing a trajectory prediction method according to an exemplary embodiment of the present disclosure. As shown in Fig. 1 , the trajectory prediction method includes the following steps.

In step S101, a map image is obtained, the map image includes the centerline of the lane, and there are several nodes arranged at equal intervals on the centerline of the lane, each node has a distance feature, and the distance feature of the node is fused with nodes having different distances from the node Distance features for other lane centerline nodes.

In step S102, based on the position of the vehicle in the map image, the centerline node of the lane closest to the position of the vehicle is determined as the target node.

In step S103, the distance feature of the target node and the historical track feature of the vehicle are vector-merged to obtain a track prediction vector.

In step S104, based on the trajectory prediction vector, the trajectory of the vehicle in the map image is predicted.

In the embodiment of the present disclosure, when predicting the driving trajectory of the vehicle, a map image including the location of the vehicle is acquired, and the map image may be rendered to a map to obtain a top view picture of the map.

Fig. 2 is a schematic diagram of lane centerlines and nodes in a map image according to an exemplary embodiment of the present disclosure. As shown in Fig. 2 , the map image includes multiple lanes. Can include lanes aligned with the direction of travel as well as opposite lanes. The lane has a lane centerline, and several nodes are arranged at preset intervals on the lane centerline, and the distances between adjacent nodes are equal. Taking any node as the current node as an example, there are other lane centerline nodes different from the current node around the current node, there can be multiple other lane centerline nodes, and the distance between other lane centerline nodes and the current node can be different , the distance features of other lane centerline nodes with different distances are fused as the distance feature of the current node.

Based on the position of the vehicle in the map image, determine the lane centerline node closest to the vehicle position as the target node, and combine the distance features at the target node with the historical trajectory features of the vehicle to obtain the trajectory prediction vector, which is used as the target node. basis for trajectory prediction.

In the embodiment of the present disclosure, the trajectory prediction vector of the vehicle can be input through a multilayer perceptron (MLP) neural network, etc., and the position where the vehicle may appear and the probability of appearing at the position can be output, so as to realize the prediction of the vehicle on the map. The driving trajectory in the image.

According to the embodiment of the present disclosure, the target node where the vehicle is located is determined in the map image, and the distance feature of the target node is determined. The distance feature is fused with the features of other nodes with different distances from the node, and the distance feature is combined with the historical track of the vehicle to obtain The trajectory prediction vector can predict the vehicle trajectory, which can preserve the topology of the lane lines in the map image and improve the accuracy of trajectory prediction.

Fig. 3 is a flowchart showing a method for determining a distance characteristic of a node according to an exemplary embodiment of the present disclosure. As shown in Fig. 3 , the method for determining a distance characteristic of a node includes the following steps.

In step S201, a first node is determined on a map image.

In step S202, an area map image is determined based on the position of the first node, and the area map image includes the first node and a plurality of second nodes having different distances from the first node.

In step S203, based on the distance between the multiple second nodes and the first node, extract the distance features of the multiple second nodes, and combine the distance features of the multiple second nodes to obtain the distance features of the first node .

In the embodiment of the present disclosure, when predicting the driving trajectory of the vehicle, the acquired map image includes the centerline of the lane, and there are several nodes arranged at equal intervals on the centerline of the lane, and the distance features of the nodes are fused with the Distance features for other lane centerline nodes at different distances. When determining the distance feature of each node in the map image, it is necessary to select a certain area, and determine the node distance feature in the area map image corresponding to the area range. The area map image includes a first node and a plurality of second nodes with different distances from the first node.

It can be understood that in the embodiment of the present disclosure, the first node can be understood as the current node, that is, the position of the current vehicle, relative to the current vehicle, there are other surrounding vehicles, and the trajectories of other surrounding vehicles are predicted to predict the current vehicle's trajectories. Carry out reasonable guidance planning for driving.

Still referring to FIG. 2, FIG. 2 shows the current node, that is, the first node, the first node included in the elliptical area, and the second node directly adjacent to the first node, that is, the second node included in the elliptical area and The first node has a distance of 1. Outside the ellipse area, the second node indirectly adjacent to the first node with a distance of 2, the second node indirectly adjacent to the first node with a distance of 3, etc. are sequentially shown. For multiple second nodes, based on the distance between them and the first node, the distance features of the second nodes are extracted, and the distance features of multiple second nodes are combined to obtain the distance features of the first node, so that the first The distance feature of the node is fused with the distance feature of the second node having a different distance from the first node.

According to an embodiment of the present disclosure, in the map image, determine the first node, determine the area map image, extract the distance features of multiple second nodes included in the area map image, and combine the distance features of the multiple second nodes , get the distance feature of the first node, preserve the topological structure of the lane line, and provide guarantee for vehicle trajectory prediction.

Fig. 4 shows a method of extracting distance features of multiple second nodes based on the distance between multiple second nodes and the first node according to an exemplary embodiment of the present disclosure, and combining the distance features of multiple second nodes Merging is performed to obtain the flow chart of the distance feature method of the first node, as shown in FIG. 4 , the method includes the following steps.

In step S301, a plurality of second nodes are divided into a plurality of categories according to the distance between them and the first node, wherein the distance between the second nodes and the first nodes in the same category is the same.

In step S302, the distance features of the second nodes of the same category among the multiple categories are respectively extracted, and the extracted distance features of different categories are concatenated to obtain the distance features of the first node.

In an embodiment of the present disclosure, when determining the distance feature of each node in the map image, determine in the map image an area map image including a first node and a plurality of second nodes with different distances from the first node , to determine the distance features of each node in the region map image.

A plurality of second nodes around the first node are divided into a plurality of categories according to distances from the first node, and second nodes with the same distance from the first node are divided into the same category. For the second node of the same category, the distance feature is extracted, and the extracted distance features of different categories are stitched together to obtain the distance feature of the first node. Understandably, the distance feature of the first node obtained by splicing is fused with the The distance feature of the adjacent second node.

Fig. 5 is a schematic diagram of determining the distance feature of the first node according to an exemplary embodiment of the present disclosure. Referring to Fig. 2 and Fig. 5, Fig. 2 shows the first node included in the area map image and the relationship with the first node The distance between the second node directly adjacent to the first node and the first node is 1. A second node indirectly adjacent to the first node at a distance of 2, a second node indirectly adjacent to the first node at a distance of 3, and so on. For multiple second nodes, they are classified into different categories according to their distances from the first node.

Figure 5 shows that the second node is divided into three categories, that is, the second node with a distance of 1 from the first node, the second node with a distance of 2 from the first node, and a distance from the first node of 3 for the second node. Extract the distance feature of the second node of each of the three categories, that is, extract the distance feature of the second node whose distance from the first node is 1, and the distance feature of the second node whose distance from the first node is 2 , and the distance feature of the second node whose distance from the first node is 3, after splicing the extracted distance features of different categories, the distance feature of the first node is obtained. It can be understood that the distance feature of the second node of each category may be represented by a vector, and the distance features of different categories are concatenated, that is, the vectors corresponding to the distance features of different categories are merged.

According to an embodiment of the present disclosure, the first node is determined in the map image, and the area map image is determined, the distance features of multiple second nodes included in the area map image are extracted, and the distance features of the multiple second nodes are combined, The distance feature of the first node is obtained, and the topological structure of the lane line is preserved, which provides guarantee for vehicle trajectory prediction.

Fig. 6 is a flowchart showing a method for determining a distance characteristic of a node according to an exemplary embodiment of the present disclosure. As shown in Fig. 6 , the method for determining a distance characteristic of a node includes the following steps.

In step S401, a first node is determined on a map image.

In step S402, with the first node as the center, determine an area map image with a preset range, including at least a second node directly adjacent to the first node and a second node indirectly adjacent to the first node within the preset range second node.

In step S403, based on the distances between the multiple second nodes and the first node, the distance features of multiple second nodes are extracted, and the distance features of multiple second nodes are combined to obtain the distance features of the first node .

In the embodiment of the present disclosure, the area map image may be an image corresponding to an area centered on the first node and having a preset range, the first node is included in the preset range, and multiple nodes with different distances from the first node a second node. The adjacency relationship between the second node and the first node may be direct adjacency or indirect adjacency. The preset range may be a regular graph or an irregular graph composed of the first node as the center and the preset distance as the side length. When the distance feature of each node is determined in the map image, an area map image including a first node and a plurality of second nodes having different distances from the first node is determined in the map image, so that the determined area map image The distance feature of each node in .

According to an embodiment of the present disclosure, the area map image is determined in the map image, and in the area map image, based on the distance between multiple second nodes and the first node, the distance features of the second nodes are extracted and combined to obtain the first node The distance feature can improve the calculation speed and improve the calculation efficiency.

In the embodiment of the present disclosure, multiple second nodes with different distances around the first node, based on the distance between the multiple second nodes and the first node, extract the distance features of the multiple second nodes, and combine the multiple The distance features of the second node are combined to obtain the distance features of the first node. Extracting the distance feature of the second node may be performed using a deep learning model (PointNet) of point cloud data. Point cloud data is a collection of unordered data points. A certain number of point clouds with spatial relationships in a specific space constitute an object. Based on PointNet, the overall characteristics of point cloud data can be extracted to provide guarantee for vehicle trajectory prediction.

Fig. 7 is a flow chart showing a method for determining historical trajectory characteristics of a vehicle according to an exemplary embodiment of the present disclosure. As shown in Fig. 7 , the method for determining historical trajectory characteristics of a vehicle includes the following steps.

In step S501, the historical trajectory of the vehicle is obtained.

In step S502, the features of the historical trajectory are extracted based on the long-short-term memory regression neural network, and the historical trajectory features of the vehicle are obtained.

In the embodiment of the present disclosure, based on the position of the vehicle in the map image, the lane centerline node closest to the vehicle position is determined as the target node, and the distance feature at the target node is vector-merged with the historical track feature of the vehicle to obtain the track The prediction vector is used as the basis for predicting the driving trajectory of the vehicle. Based on the trajectory prediction vector, the driving trajectory of the vehicle in the map image is predicted. Obtaining the historical trajectory of the vehicle can be described by the moving trajectory formed by the center of the vehicle, and the characteristics of the historical trajectory of the vehicle are extracted based on the long short-term memory regression neural network (Long Short-Term Memory, LSTM). The LSTM neural network is a variant based on the Recurrent Neural Network (RNN), which can realize the transmission of information from the previous moment to the next moment, and effectively solve the problems of gradient disappearance and gradient explosion in the training process.

Based on the same idea, an embodiment of the present disclosure also provides a trajectory prediction device.

Fig. 8 is a block diagram of a trajectory prediction device according to an exemplary embodiment of the present disclosure. Referring to FIG. 8 , the trajectory prediction device 100 includes: an acquisition unit 101 , a determination unit 102 and a prediction unit 103 .

The acquisition unit 101 is used to acquire a map image. The map image includes a lane centerline, and there are several nodes arranged at equal intervals on the lane centerline. Each node has a distance feature, and the distance feature of the node is fused with a node having a different distance from the node. The distance features of other lane centerline nodes.

The determining unit 102 is configured to determine, based on the position of the vehicle in the map image, the centerline node of the lane closest to the vehicle position as the target node, and vector-merge the distance feature of the target node with the historical track feature of the vehicle to obtain track prediction vector.

The predicting unit 103 is configured to predict the driving trajectory of the vehicle in the map image based on the trajectory prediction vector.

In some embodiments, the distance feature of each node is determined in the following manner: determine the first node on the map image; determine the area map image based on the position of the first node, the area map image includes the first node, and A plurality of second nodes with different distances between the first nodes; based on the distance between the plurality of second nodes and the first node, the distance features of the plurality of second nodes are extracted, and the distance features of the plurality of second nodes are extracted Merge to obtain the distance feature of the first node.

In some embodiments, the determination unit 102 extracts the distance features of multiple second nodes based on the distance between the multiple second nodes and the first node in the following manner, and combines the distance features of the multiple second nodes, Obtaining the distance feature of the first node: dividing a plurality of second nodes into multiple categories according to the distance between them and the first node, wherein the distance between the second node and the first node in the same category is the same; The distance features of the second nodes of the same category in the multiple categories are respectively extracted, and the extracted distance features of different categories are spliced to obtain the distance features of the first node.

In some embodiments, the determining unit 102 determines an area map image based on the position of the first node in the following manner: centering on the first node, determine an area map image with a preset range, including at least the first node within the preset range A second node directly adjacent to a node, and a second node indirectly adjacent to the first node.

In some embodiments, the determining unit 102 extracts distance features of multiple second nodes in the following manner: extracts distance features of multiple second nodes based on a deep learning model of point cloud data.

In some embodiments, the characteristics of the historical trajectory of the vehicle are determined in the following manner: the historical trajectory of the vehicle is obtained; the characteristics of the historical trajectory are extracted based on the long-short-term memory regression neural network to obtain the characteristics of the historical trajectory of the vehicle.

It can be understood that, in order to realize the above-mentioned functions, the apparatus provided by the embodiments of the present disclosure includes corresponding hardware structures and/or software modules for performing various functions. Combining the units and algorithm steps of each example disclosed in the embodiments of the present disclosure, the embodiments of the present disclosure can be implemented in the form of hardware or a combination of hardware and computer software. Whether a certain function is executed by hardware or computer software drives hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art may use different methods to implement the described functions for each specific application, but such implementation should not be regarded as exceeding the scope of the technical solutions of the embodiments of the present disclosure.

Regarding the apparatus in the foregoing embodiments, the specific manner in which each module executes operations has been described in detail in the embodiments related to the method, and will not be described in detail here.

Fig. 9 is a block diagram showing a device for trajectory prediction according to an exemplary embodiment of the present disclosure. For example, the apparatus 200 may be a mobile phone, a computer, a digital broadcast terminal, a messaging device, a game console, a tablet device, a medical device, a fitness device, a personal digital assistant, and the like.

Referring to FIG. 9, the device 200 may include one or more of the following components: a processing component 202, a memory 204, a power component 206, a multimedia component 208, an audio component 210, an input/output (I/O) interface 212, a sensor component 214, and communication component 216 .

The processing component 202 generally controls the overall operations of the device 200, such as operations associated with display, telephone calls, data communications, camera operations, and recording operations. The processing component 202 may include one or more processors 220 to execute instructions to complete all or part of the steps of the above method. Additionally, processing component 202 may include one or more modules that facilitate interaction between processing component 202 and other components. For example, processing component 202 may include a multimedia module to facilitate interaction between multimedia component 208 and processing component 202 .

The memory 204 is configured to store various types of data to support operations at the device 200 . Examples of such data include instructions for any application or method operating on device 200, contact data, phonebook data, messages, pictures, videos, etc. The memory 204 can be implemented by any type of volatile or non-volatile storage device or their combination, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable Programmable Read Only Memory (EPROM), Programmable Read Only Memory (PROM), Read Only Memory (ROM), Magnetic Memory, Flash Memory, Magnetic or Optical Disk.

The power component 206 provides power to various components of the device 200 . Power components 206 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power for device 200 .

The multimedia component 208 includes a screen that provides an output interface between the device 200 and the user. In some embodiments, the screen may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen may be implemented as a touch screen to receive input signals from a user. The touch panel includes one or more touch sensors to sense touches, swipes, and gestures on the touch panel. The touch sensor may not only sense a boundary of a touch or swipe action, but also detect duration and pressure associated with the touch or swipe action. In some embodiments, the multimedia component 208 includes a front camera and/or a rear camera. When the device 200 is in an operation mode, such as a shooting mode or a video mode, the front camera and/or the rear camera can receive external multimedia data. Each front camera and rear camera can be a fixed optical lens system or have focal length and optical zoom capability.

The audio component 210 is configured to output and/or input audio signals. For example, the audio component 210 includes a microphone (MIC), which is configured to receive external audio signals when the device 200 is in operation modes, such as call mode, recording mode and voice recognition mode. Received audio signals may be further stored in memory 204 or sent via communication component 216 . In some embodiments, the audio component 210 also includes a speaker for outputting audio signals.

The I/O interface 212 provides an interface between the processing component 202 and a peripheral interface module, which may be a keyboard, a click wheel, a button, and the like. These buttons may include, but are not limited to: a home button, volume buttons, start button, and lock button.

Sensor assembly 214 includes one or more sensors for providing various aspects of status assessment for device 200 . For example, the sensor component 214 can detect the open/closed state of the device 200, the relative positioning of components, such as the display and keypad of the device 200, and the sensor component 214 can also detect a change in the position of the device 200 or a component of the device 200 , the presence or absence of user contact with the device 200 , the device 200 orientation or acceleration/deceleration and the temperature change of the device 200 . The sensor assembly 214 may include a proximity sensor configured to detect the presence of nearby objects in the absence of any physical contact. Sensor assembly 214 may also include an optical sensor, such as a CMOS or CCD image sensor, for use in imaging applications. In some embodiments, the sensor component 214 may also include an acceleration sensor, a gyroscope sensor, a magnetic sensor, a pressure sensor or a temperature sensor.

The communication component 216 is configured to facilitate wired or wireless communication between the apparatus 200 and other devices. The device 200 can access wireless networks based on communication standards, such as WiFi, 2G or 3G, or a combination thereof. In an exemplary embodiment, the communication component 216 receives broadcast signals or broadcast related information from an external broadcast management system via a broadcast channel. In an exemplary embodiment, the communication component 216 also includes a near field communication (NFC) module to facilitate short-range communication. For example, the NFC module may be implemented based on Radio Frequency Identification (RFID) technology, Infrared Data Association (IrDA) technology, Ultra Wideband (UWB) technology, Bluetooth (BT) technology and other technologies.

In an exemplary embodiment, apparatus 200 may be programmed by one or more application specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable A gate array (FPGA), controller, microcontroller, microprocessor or other electronic component implementation for performing the methods described above.

In an exemplary embodiment, there is also provided a non-transitory computer-readable storage medium including instructions, such as the memory 204 including instructions, which can be executed by the processor 220 of the device 200 to implement the above method. For example, the non-transitory computer readable storage medium may be ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, optical data storage device, and the like.

In order to realize the above-mentioned embodiments, the present disclosure also proposes a computer program, including computer readable codes, which, when the computer readable codes are run on a computing processing device, cause the computing processing device to execute the aforementioned trajectory prediction method.

It can be understood that "plurality" in the present disclosure refers to two or more, and other quantifiers are similar. "And/or" describes the association relationship of associated objects, indicating that there may be three types of relationships, for example, A and/or B may indicate: A exists alone, A and B exist simultaneously, and B exists independently. The character "/" generally indicates that the contextual objects are an "or" relationship. The singular forms "a", "said" and "the" are also intended to include the plural unless the context clearly dictates otherwise.

It can be further understood that the terms "first", "second", etc. are used to describe various information, but the information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another, and do not imply a specific order or degree of importance. In fact, expressions such as "first" and "second" can be used interchangeably. For example, without departing from the scope of the present disclosure, first information may also be called second information, and similarly, second information may also be called first information.

It can be further understood that, unless otherwise specified, "connection" includes a direct connection without other components between the two, and also includes an indirect connection between the two with other elements.

It can be further understood that although operations are described in a specific order in the drawings in the embodiments of the present disclosure, it should not be understood as requiring that these operations be performed in the specific order shown or in a serial order, or that Do all of the operations shown to get the desired result. In certain circumstances, multitasking and parallel processing may be advantageous.

Other embodiments of the disclosure will be readily apparent to those skilled in the art from consideration of the specification and practice of the disclosure disclosed herein. This application is intended to cover any modification, use or adaptation of the present disclosure, and these modifications, uses or adaptations follow the general principles of the present disclosure and include common knowledge or conventional technical means in the technical field not disclosed in the present disclosure . The specification and examples are to be considered as illustrative only, with the true scope and spirit of the disclosure indicated by the following claims.

It should be understood that the present disclosure is not limited to the precise constructions which have been described above and shown in the drawings, and various modifications and changes may be made without departing from the scope thereof. The scope of the present disclosure is limited only by the scope of the appended claims.

Claims

A trajectory prediction method, characterized in that the method comprises:

Obtain a map image, the map image includes a lane centerline, the lane centerline has several nodes arranged at equal intervals, each node has a distance feature, and the distance feature of the node is fused with the distance feature of the node Distance features of other lane centerline nodes at different distances;

Based on the position of the vehicle in the map image, determine the lane centerline node closest to the vehicle position as the target node;

Carrying out vector merging of the distance feature of the target node and the historical track feature of the vehicle to obtain a track prediction vector;

Based on the trajectory prediction vector, a trajectory of the vehicle in the map image is predicted.
The trajectory prediction method according to claim 1, wherein the distance feature of each node is determined in the following manner:

determining a first node on said map image;

determining an area map image based on the position of the first node, the area map image including the first node and a plurality of second nodes having different distances from the first node;

Based on the distance between the plurality of second nodes and the first node, extracting distance features of the plurality of second nodes, and merging the distance features of the plurality of second nodes to obtain the first node A node distance feature.
The trajectory prediction method according to claim 2, wherein, based on the distance between the plurality of second nodes and the first node, the distance features of the plurality of second nodes are extracted, and the The distance features of multiple second nodes are merged to obtain the distance features of the first node, including:

Dividing the plurality of second nodes into a plurality of categories according to the distance between them and the first node, wherein the distance between the second nodes in the same category and the first node is the same;

The distance features of the second nodes of the same category among the plurality of categories are respectively extracted, and the extracted distance features of different categories are spliced to obtain the distance features of the first node.
The trajectory prediction method according to any one of claims 2 or 3, wherein, based on the position of the first node, determining an area map image includes:

Centering on the first node, determine an area map image with a preset range, including at least a second node directly adjacent to the first node, and a second node indirectly connected to the first node within the preset range adjacent second node.
trajectory prediction method according to claim 4, is characterized in that, extracting the distance feature of described multiple second nodes comprises:

The distance features of the plurality of second nodes are extracted based on a deep learning model of point cloud data.
The trajectory prediction method according to claim 4, wherein the historical trajectory characteristics of the vehicle are determined in the following manner:

Obtain the historical trajectory of the vehicle;

The characteristics of the historical trajectory are extracted based on a long-short-term memory regression neural network to obtain the characteristics of the historical trajectory of the vehicle.
A trajectory prediction device, characterized in that the device comprises:

The acquisition unit is used to acquire a map image, the map image includes a lane centerline, and the lane centerline has several nodes arranged at equal intervals, each node has a distance feature, and the distance features of the nodes are fused with distance characteristics of other lane centerline nodes having different distances from said node;

A determination unit, configured to determine, based on the position of the vehicle in the map image, a lane centerline node closest to the position of the vehicle as a target node, and combine the distance feature of the target node with the historical track feature of the vehicle Perform vector merging to obtain trajectory prediction vectors;

A predicting unit, configured to predict the driving trajectory of the vehicle in the map image based on the trajectory prediction vector.
The trajectory prediction device according to claim 7, wherein the distance feature of each node is determined in the following manner:

determining a first node on said map image;

determining an area map image based on the position of the first node, the area map image including the first node and a plurality of second nodes having different distances from the first node;

Based on the distance between the plurality of second nodes and the first node, extracting distance features of the plurality of second nodes, and merging the distance features of the plurality of second nodes to obtain the first node A node distance feature.
The trajectory prediction device according to claim 8, wherein the determination unit extracts the distances of the plurality of second nodes based on the distance between the plurality of second nodes and the first node in the following manner: distance feature, and merge the distance features of the plurality of second nodes to obtain the distance feature of the first node:

Dividing the plurality of second nodes into a plurality of categories according to the distance between them and the first node, wherein the distance between the second nodes in the same category and the first node is the same;

Extract the distance features of the second nodes of the same category in the plurality of categories respectively, and after splicing the extracted distance features of different categories, obtain the distance features of the first node.
The trajectory prediction device according to any one of claims 8 or 9, wherein the determining unit determines an area map image based on the position of the first node in the following manner:

Centering on the first node, determine an area map image with a preset range, including at least a second node directly adjacent to the first node, and a second node indirectly connected to the first node within the preset range adjacent second node.
The trajectory prediction device according to claim 10, wherein the determining unit extracts the distance features of the plurality of second nodes in the following manner:

The distance features of the plurality of second nodes are extracted based on a deep learning model of point cloud data.
The trajectory prediction device according to claim 10, wherein the historical trajectory characteristics of the vehicle are determined in the following manner:

Obtain the historical trajectory of the vehicle;

The characteristics of the historical trajectory are extracted based on a long-short-term memory regression neural network to obtain the characteristics of the historical trajectory of the vehicle.
A trajectory prediction device, characterized in that it comprises:

processor;

memory for storing processor-executable instructions;

Wherein, the processor is configured to: execute the trajectory prediction method according to any one of claims 1-6.
A storage medium, characterized in that instructions are stored in the storage medium, and when the instructions in the storage medium are executed by the processor of the terminal, the terminal can execute the method described in any one of claims 1 to 6. Trajectory prediction method.
A computer program, comprising computer readable code, which, when the computer readable code is run on a computing processing device, causes the computing processing device to perform the method of heating the battery pack according to any one of claims 1-5 Control Method.