Trajectory prediction method and device, and map

Abstract

The present application discloses a trajectory prediction method and apparatus, and a map, which are related to the fields of V2X, intelligent vehicles, autonomous driving, Internet of Vehicles, and network driving. In the method, at least one target point is first determined, where the distance between the target point and the vehicle is included in a first range; at least one reference trajectory between the target and the at least one target point is obtained based on the at least one target point and the geographical position of the target; and a predicted trajectory of the target is determined from the at least one reference trajectory. During the trajectory prediction performed by using the method, not only the historical movement information of the target is used, but also information about the target point is used, and more diverse information can be used. Therefore, the method can not only accurately predict the trajectory of the target in the next relatively short period, but also improve the accuracy of predicting the trajectory of the target in the next relatively long period.

Description

[Technical Field]

[0001] This application relates to the field of intelligent transportation technology, and more particularly to trajectory prediction methods and apparatus, and maps. [Background technology]

[0002] In the driving process, an intelligent vehicle usually needs to predict the trajectory of a target on the road. The target may be a pedestrian or another vehicle on the road. Therefore, the intelligent vehicle plans the driving behavior of the intelligent vehicle based on the prediction result to avoid a collision between the intelligent vehicle and the target.

[0003] Currently, when predicting a target's trajectory, an intelligent vehicle typically obtains historical movement information of the target based on sensors configured within the intelligent vehicle. The historical movement information typically includes the target's historical movement speed and direction, as well as lane boundary information for the road on which the target is located. The vehicle then predicts the target's trajectory based on the information to obtain a predicted trajectory of the target.

[0004] In other words, in the prior art, historical movement information of the target is used during the trajectory prediction of the target. Therefore, in the prior art, the trajectory of the target in the next relatively short period can be predicted with relatively high accuracy, but the accuracy of predicting the trajectory of the target in the next relatively long period is relatively low. For example, it has been found through multiple tests that the prior art can accurately predict the trajectory of the target only in the next 3 to 5 seconds, and the accuracy of predicting the trajectory in a longer period is relatively low. Summary of the Invention

[0005] To solve the problem that existing trajectory prediction methods can only accurately predict the trajectory of a target for the next relatively short period of time, the embodiments of the present application provide a trajectory prediction method and apparatus, and a map.

[0006] According to a first aspect, an embodiment of the present application comprises: determining at least one target point, wherein a distance between the target point and the vehicle is within a first range; obtaining at least one reference trajectory between a target and the at least one target point based on the at least one target point and a geographical position of the target; and determining a predicted trajectory of the target from said at least one reference trajectory; A method for trajectory prediction is disclosed, comprising:

[0007] According to the present solution, during the trajectory prediction of the target, not only the historical movement information of the target but also the information about the target point are used, so that the present solution can not only accurately predict the trajectory of the target in the next relatively short period, but also improve the accuracy of predicting the trajectory of the target in the next relatively long period.

[0008] In any design, the step of determining at least one target point may further comprise: determining a first reference point by consulting a map on which reference points are marked or by consulting a text file indicating said reference points, wherein the distance between said first reference point and said vehicle is within said first range; and If the number of first reference points is not greater than n, determining the first reference points as the target points, where n is a predetermined natural number; or if the number of first reference points is greater than n, selecting the at least one target point from the first reference points; Includes.

[0009] According to the aforementioned design, the at least one target point may be determined based on a map on which the reference field is marked and / or a text file indicating the reference field.

[0010] In any design, obtaining at least one reference trajectory between a target and the at least one target point comprises: determining a historical trajectory from the target to the at least one target point as the at least one reference trajectory; or determining the shortest trajectory between the target and any one of the at least one target points as a reference trajectory when a traffic rule is satisfied; Includes.

[0011] In any design, the step of determining a predicted trajectory from the target to the at least one reference trajectory comprises: determining a short-term trajectory of the target for the next s seconds based on a movement trend of the target; determining a reliability of the at least one reference trajectory based on the short-term trajectory, wherein the reliability of the reference trajectory indicates a degree of fit between the reference trajectory and the short-term trajectory; and determining the reference trajectory having the highest reliability as the predicted trajectory of the target; Includes.

[0012] According to the above design, a predicted trajectory can be determined from a reference trajectory based on the reliability of the reference trajectory. A higher reliability of the reference trajectory indicates that the reference trajectory fits the short-term trajectory better. Therefore, the predicted trajectory determined in the above design is the reference trajectory that best fits the movement trend of the target in the reference trajectory.

[0013] In any design, determining a short-term trajectory of the target for the next s seconds based on a movement trend of the target may include: determining the trajectory obtained after the target has moved for s seconds under a current movement trend as the short-term trajectory; or determining second parameters of the target based on the movement trend of the target, where the second parameters of the target include a historical movement speed and a historical movement direction of the target, a geographic location of the target, and environmental information of the target; and obtaining the short-term trajectory output by the neural network model of the target for the next s seconds after the second parameter is input into the neural network model; Includes.

[0014] In any design, determining the reliability of the at least one reference trajectory based on the short-term trajectory comprises: determining a projection value of the short-term orbit of the at least one reference orbit according to the formula: wherein the projection value is the confidence of the at least one reference trajectory;

number

number

number

number

number

number

number

number

[0015] In any design, the method further comprises: adjusting a driving behavior of the vehicle after determining based on the predicted trajectory that the vehicle is about to collide with the target.

[0016] According to this design, the driving behavior of the vehicle is planned based on the predicted trajectory, which can improve the driving planning ability of the vehicle and ensure the safety and comfort of the vehicle.

[0017] According to a second aspect, an embodiment of the present application comprises: Processor and Transceiver Interface where the transceiver interface is configured to receive information about the sensor; The processor is configured to: determine at least one target point based on the information about the sensor, where a distance between the target point and the vehicle is within a first range; obtain at least one reference trajectory between the target and the at least one target point based on the at least one target point and a geographic location of the target; and determine a predicted trajectory of the target from the at least one reference trajectory. A trajectory prediction device is disclosed.

[0018] In any design, the processor may: determining a first reference point by consulting a map on which reference points are marked or by consulting a text file indicating said reference points, wherein the distance between said first reference point and said vehicle is within said first range; and If the number of first reference points is not greater than n, determining the first reference points as the target points, where n is a predetermined natural number; or if the number of first reference points is greater than n, selecting the at least one target point from the first reference points. The present invention is specifically configured to:

[0019] In any design, the processor is specifically configured to determine a historical trajectory from the target to the at least one target point as the at least one reference trajectory, or the processor is specifically configured to determine a shortest trajectory between the target and any one of the at least one target points as the reference trajectory when traffic rules are satisfied.

[0020] In any design, the processor may: determining a short-term trajectory of the target for the next s seconds based on a movement trend of the target; determining a reliability of the at least one reference trajectory based on the short-term trajectory, where the reliability of the reference trajectory indicates a degree of fit between the reference trajectory and the short-term trajectory; and determining the reference trajectory having the highest reliability as the predicted trajectory of the target; The present invention is specifically configured to:

[0021] In any design, the processor is specifically configured to determine as the short-term trajectory a trajectory obtained after the target has moved for s seconds under a current movement trend; or The processor determines second parameters of the target based on the movement trend of the target, where the second parameters of the target include a historical movement speed and a historical movement direction of the target, a geographic location of the target, and environmental information of the target; and obtaining the short-term trajectory output by the neural network model of the target for the next s seconds after the second parameter is input into the neural network model; The present invention is specifically configured to:

[0022] In any design, the processor may: specifically configured to determine a projection value of the short-term trajectory of the at least one reference trajectory according to the following formula, wherein the projection value is the confidence level of the at least one reference trajectory;

number

number

number

number

number

number

number

number

[0023] In any design, the processor further comprises: and adjusting a driving behavior of the vehicle after determining, based on the predicted trajectory, that the vehicle is about to collide with the target.

[0024] According to a third aspect, an embodiment of the present application provides a map including a first layer, wherein: The first tier includes at least one reference point, the reference point including at least one of locations such as business industry locations, enterprise locations, and service industry locations.

[0025] In the process of predicting the trajectory of a target, the map provided in this embodiment of the present application helps to find reference points to which the target may travel, and can be used to determine a target point based on the found reference points and the survey information of the target. Therefore, during the trajectory prediction of a target by using the map provided in this embodiment of the present application, not only the historical movement information of the target but also information about the target point is used. This helps to improve the accuracy of predicting the trajectory of the target in the next relatively long period of time.

[0026] In any design, the reference point is a location where a number of people pass through in a first period of time that is not less than a second threshold.

[0027] According to a fourth aspect, an embodiment of the present application comprises: At least one processor and memory where the memory configured to store program instructions; The processor is configured to call and execute the program instructions stored in the memory, so that the terminal device executes the trajectory prediction method according to the first aspect. A terminal device is provided.

[0028] According to a fifth aspect, an embodiment of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores instructions that, when executed on a computer, enable the computer to perform a method according to the first aspect.

[0029] According to a sixth aspect, embodiments of the present application provide a computer program product comprising instructions, which when executed on an electronic device, enable the electronic device to perform a method according to the first aspect.

[0030] An embodiment of the present application provides a trajectory prediction method, in which at least one target point is first determined, at least one reference trajectory between a target of a vehicle and the at least one target point is determined, and a predicted trajectory of the target is determined from the at least one reference trajectory.

[0031] In other words, in the solution provided in the embodiment of the present application, a target point needs to be determined during the trajectory prediction of the target. The target point is usually a location where the target is more likely to travel. In this case, during the trajectory prediction of the target by using the solution provided in the embodiment of the present application, not only the historical movement information of the target but also the information about the target point is used.

[0032] Compared with existing methods for predicting a target trajectory, the solution provided in the embodiments of the present application can use more diverse information during target trajectory prediction, so that the solution provided in the embodiments of the present application can not only accurately predict the target trajectory in the next relatively short period, but also improve the accuracy of predicting the target trajectory in the next relatively long period.

[0033] In the prior art, usually, only the trajectory of a target for the next 3 to 5 seconds can be accurately predicted. However, the solution provided in the embodiment of the present application can accurately predict the trajectory of a target for the next 5 to 20 seconds. Compared with the prior art, the solution provided in the embodiment of the present application improves the accuracy of predicting the trajectory of a target for the next relatively long period. [Brief explanation of the drawings]

[0034] [Figure 1] FIG. 1 is a schematic diagram of a vehicle structure.

[0035] [Figure 2] 1 is a schematic diagram of the architecture of a computer system in a vehicle.

[0036] [Figure 3] FIG. 2 is a schematic diagram of the connection relationship between the vehicle and the cloud side.

[0037] [Figure 4] 1 is a schematic diagram of an application scenario of a trajectory prediction method in the prior art;

[0038] [Figure 5] FIG. 1 is a schematic diagram of an application scenario of another trajectory prediction method in the prior art;

[0039] [Figure 6] FIG. 2 is a schematic diagram of the operation procedure of the trajectory prediction method according to an embodiment of the present application;

[0040] [Figure 7(a)] 1 is an exemplary diagram of a map according to the prior art;

[0041] [Figure 7(b)] 1 is an exemplary diagram of a map in a trajectory prediction method according to an embodiment of the present application; FIG.

[0042] [Figure 8] FIG. 1 is an exemplary illustration of a map of a trajectory prediction method according to an embodiment of the present application.

[0043] [Figure 9] 1 is an exemplary diagram of an application scenario of a trajectory prediction method according to an embodiment of the present application; FIG.

[0044] [Figure 10] 1 is a schematic diagram of the structure of a trajectory prediction device according to an embodiment of the present application;

[0045] [Figure 11]1 is a schematic diagram of the structure of a terminal device according to an embodiment of the present application; DETAILED DESCRIPTION OF THE INVENTION

[0046] The following describes the technical solutions in the embodiments of the present application with reference to the accompanying drawings in the embodiments of the present application.

[0047] The terms used in the following embodiments are intended to merely describe particular embodiments and are not intended to limit the present application. As used in this specification and the appended claims, the singular terms "one," "a," "the," "the foregoing," "this," and "the one" are intended to include terms such as "one or more," unless the context clearly indicates otherwise. In the following embodiments of the present application, it should be further understood that "at least one" and "one or more" mean one, two, or more. The term "or" is used to describe an association relationship between related objects and indicates that three relationships may exist. For example, A or B may represent the following cases: only A exists, both A and B exist, or only B exists, where A and B may each be singular or plural. The " / " character generally indicates that the associated objects are in an "or" relationship.

[0048] References to "an embodiment" or "some embodiments" described herein indicate that one or more embodiments of the present application include the particular features, structures, or characteristics described with reference to the embodiment. Thus, phrases such as "in an embodiment," "in some embodiments," "in some other embodiments," and "in other embodiments" appearing in different places in this specification do not necessarily refer to the same embodiment; instead, they mean "one or more but not all of the embodiments" unless specifically emphasized otherwise. The terms "include," "comprise," "have," and variations thereof all mean "include but are not limited to," unless specifically emphasized otherwise.

[0049] Existing trajectory prediction methods can only accurately predict the trajectory of a target in the next relatively short period of time, and have relatively low accuracy in predicting the trajectory of a target in the next relatively long period of time. To solve this technical problem, embodiments of the present application provide a trajectory prediction method and apparatus. The trajectory prediction method and apparatus can be applied to fields such as vehicle-to-everything (V2X), intelligent vehicles, autonomous driving, Internet of Vehicles, and networked driving.

[0050] The method may be applied to a vehicle. The vehicle is typically an intelligent vehicle. In a possible implementation, the vehicle may be configured to be in a fully or partially autonomous driving mode. For example, when the vehicle is in the autonomous driving mode, the vehicle may further use manual operations to determine the current status of the vehicle and the vehicle's surrounding environment, determine the possible behavior of at least one target, such as another vehicle or a pedestrian, in the surrounding environment, and control the vehicle based on the determined information. Additionally, when the vehicle is in the autonomous driving mode, the vehicle may be configured to perform operations without human interaction.

[0051] 1 is a functional block diagram of a vehicle 100 according to an embodiment of the present invention. As shown in FIG. 1, the vehicle 100 may include various subsystems, such as a driving system 102, a sensor system 104, a planning and control system 106, one or more peripheral devices 108, a power source 110, a computer system 101, and a user interface 116.

[0052] Optionally, vehicle 100 may include more or fewer subsystems, and each subsystem may include multiple components. Additionally, all subsystems and components in vehicle 100 may be connected to each other in a wired or wireless manner.

[0053] The traction system 102 may include components that provide power to the vehicle 100. In an embodiment, the traction system 102 may include an engine 118, an energy source 119, a transmission 120, and wheels 121. The engine 118 may be an internal combustion engine, a motor, an air-compression engine, another type of engine, or a combination of multiple engines. A combination of multiple engines herein may include, for example, a hybrid engine formed by a gasoline engine and a motor, or a hybrid engine formed by an internal combustion engine and an air-compression engine. The engine 118 converts the energy source 119 into mechanical energy.

[0054] Examples of energy source 119 include gasoline, diesel, another petroleum-based fuel, propane, another compressed gas-based fuel, ethanol, a solar panel, a battery, or another power source. Energy source 119 may also provide energy for other systems in vehicle 100.

[0055] The transmission 120 may transmit mechanical power from the engine 118 to the wheels 121. The transmission 120 may include a gearbox, a differential, and a drive shaft. In embodiments, the transmission 120 may further include other components, such as a clutch. The drive shaft may include one or more shafts that may be coupled to one or more wheels 121.

[0056] The sensor system 104 may include several sensors that sense information about the vehicle 100 and its surrounding environment. For example, the sensor system 104 may include a positioning system 122 (which may be a GPS system, the Beidou system, or another positioning system), an inertial measurement unit (IMU) 124, a radar 126, a laser rangefinder 128, a camera 130, a computer vision system 138, and a sensor fusion algorithm 140. The sensor system 104 may also include sensors within the vehicle 100's internal systems (e.g., an on-board air quality monitor, a fuel gauge, or an engine oil temperature gauge). Data from one or more of these sensors may be used to detect objects and corresponding characteristics of the objects (e.g., position, shape, orientation, speed, etc.). Such detection and recognition is a primary function of the vehicle 100 used to implement safe operation.

[0057] The global positioning system 122 may be configured to estimate the geographic position of the vehicle 100. The IMU 124 is configured to detect changes in position and orientation of the vehicle 100 based on inertial acceleration. In one embodiment, the IMU 124 may be a combination of an accelerometer and a gyroscope.

[0058] By using radio signals, radar 126 may detect objects in the environment surrounding vehicle 100. In some embodiments, in addition to objects, radar 126 may be configured to detect the speed or direction of travel of the objects.

[0059] Laser rangefinder 128 may use lasers to detect objects in the environment in which vehicle 100 is located. In some embodiments, laser rangefinder 128 may include one or more laser sources, a laser scanner, one or more detectors, and other system components.

[0060] Camera 130 may be configured to capture multiple images of the environment surrounding vehicle 100. Camera 130 may be a still camera or a video camera.

[0061] The computer vision system 138 may perform operations to process and analyze images captured by the camera 130 to recognize objects or features in the environment surrounding the vehicle 100. The objects or features may include traffic signals, road boundaries, and landmarks. The computer vision system 138 may use object recognition algorithms, structure from motion (SFM) algorithms, video tracking, and other computer vision techniques. In some embodiments, the computer vision system 138 may be configured to perform tasks such as drawing a map of the environment, tracking objects, and estimating the speed of objects.

[0062] The planning and control system 106 is configured to control the operation of the vehicle 100 and components of the vehicle 100. The planning and control system 106 may include various components, including a steering system 132, an accelerator 134, a braking unit 136, a route control system 142, and a target avoidance system 144.

[0063] The forward direction of the vehicle 100 may be adjusted by manipulating the steering system 132. For example, in an embodiment, the steering system 132 may be a steering wheel system.

[0064] The axel 134 is configured to control the operating speed of the engine 118 and thereby further control the speed of the vehicle 100 .

[0065] Braking unit 136 is configured to control and decelerate vehicle 100. Braking unit 136 may use frictional force to slow down wheels 121. In another embodiment, braking unit 136 may convert the kinetic energy of wheels 121 into electrical current. Braking unit 136 may alternatively use another form of reducing the rotational speed of wheels 121 to control the speed of vehicle 100.

[0066] The route planning system 142 is configured to determine a driving route for the vehicle 100. In some embodiments, the route planning system 142 may plan a driving route for the vehicle 100 that may avoid potential targets in the environment based on data from the sensors 138, the GPS 122, and one or more predetermined maps. The trajectory planning method provided in the embodiments of the present application may be executed by the route planning system 142 to output a target driving trajectory for the vehicle 100. The target driving trajectory includes a plurality of target waypoints. Each of the plurality of target waypoints includes coordinates of the waypoint, as well as an allowable lateral error and an allowable speed error for the waypoint. The allowable lateral error described herein includes a value range of the allowable lateral error, and in some cases may be understood as abbreviated to a value range of the allowable lateral error. In this specification, the term "lateral" refers to a direction perpendicular or approximately perpendicular to the vehicle's direction of travel. The allowable lateral error essentially means the range of allowable lateral displacement errors, i.e., the allowable displacement error of the vehicle 100 in a direction perpendicular or nearly perpendicular to the direction of travel of the vehicle, and will not be described in detail again below.

[0067] The control system 144 is configured to generate control amounts for the accelerator, brake, and steering angle based on the driving route / trajectory output by the route planning system, and control the steering system 132, the accelerator 134, and the braking unit 136.

[0068] Indeed, in some examples, the planning and control system 106 may include additional or alternative components other than those shown and described. Alternatively, some of the components shown above may be omitted from the planning and control system 106.

[0069] The vehicle 100 interacts with external sensors, other vehicles, other computer systems, or a user using peripheral devices 108. The peripheral devices 108 may include a wireless communication system 146, an on-board computer 148, a microphone 150, or a speaker 152.

[0070] In some embodiments, peripheral devices 108 provide a means for interaction between a user of vehicle 100 and user interface 116. For example, onboard computer 148 may provide information to the user of vehicle 100. User interface 116 may further operate onboard computer 148 to receive user input. In some implementations, onboard computer 148 may be operated by using a touchscreen. In other cases, peripheral devices 108 may provide a means for communication between vehicle 100 and another device located within the vehicle. For example, microphone 150 may receive audio (e.g., voice commands or other audio input) from the user of vehicle 100. Similarly, speaker 152 may output audio to the user of vehicle 100.

[0071] The wireless communication system 146 may wirelessly communicate with one or more devices directly or through a communication network. For example, the wireless communication system 146 may use 3G cellular communication such as CDMA, EVD0, or GSM / GPRS, 4G cellular communication such as LTE, or 5G cellular network communication. The wireless communication system 146 may communicate with a wireless local area network (WLAN) through Wi-Fi. In some embodiments, the wireless communication system 146 may communicate directly with devices through an infrared link, Bluetooth, or ZigBee. Other wireless protocols, such as various vehicle communication systems, for example, the wireless communication system 146, may include one or more dedicated short range communication (DSRC) devices, which may include devices that perform public and / or private data communication with vehicles or roadside stations.

[0072] The power source 110 may provide power for the various components of the vehicle 100. In some embodiments, the power source 110 may be a rechargeable lithium-ion battery or a lead-acid battery. One or more battery packs of such batteries may be configured as a power source to provide power to the various components of the vehicle 100. In some embodiments, the power source 110 and the energy source 119 may be implemented together, for example, in an all-electric vehicle.

[0073] Some or all of the functions of vehicle 100 are controlled by computer system 101. Computer system 101 may include at least one processor 113, which executes instructions 115 stored on a non-transitory computer-readable medium, such as memory 114. Computer system 101 may alternatively be multiple computing devices that control individual components or subsystems of vehicle 100 in a distributed manner.

[0074] The processor 113 may be any conventional processor, such as, for example, a commercially available CPU. Alternatively, the processor may be a dedicated device, such as an ASIC or another hardware-based processor. While FIG. 1 technically illustrates the processor, memory, and other components of the computer system 101, those skilled in the art should understand that the processor and memory may actually include multiple other processors or memories that are not located within the same physical housing. For example, the memory may be a hard disk drive or another storage medium located within a different housing than that of the computer system 101. Thus, references to a processor should be understood to include references to a set of processors or memories that may or may not perform operations in parallel. Unlike using a single processor to perform the steps described herein, some components, such as the steering component and the deceleration component, may each include their own processor, which performs only calculations related to the component's specific function; or, subsystems, such as the cruise system, sensor system, and planning and control system, may also include their own processors to implement the calculations associated with the corresponding subsystem's tasks to implement the corresponding function.

[0075] In various aspects described herein, the processor may be located remotely from the vehicle and may communicate wirelessly with the vehicle. In other aspects, some of the processes described herein are executed on a processor located within the vehicle, while others are executed by a remote processor and include performing the steps necessary to perform a single operation.

[0076] In some embodiments, memory 114 may include instructions 115 (e.g., program logic) that may be executed by processor 113 to perform various functions of vehicle 100, including those described above. Memory 114 may also include additional instructions, including instructions used to send data to, receive data from, interact with, or control one or more of cruise system 102, sensor system 104, planning and control system 106, and peripheral devices 108.

[0077] In addition to instructions 115, memory 114 may store other relevant data, such as road maps, route information, and vehicle position, direction, speed, and other relevant information. Such information may be used by vehicle 100 or specifically by computer system 101 while vehicle 100 is operating in an autonomous, semi-autonomous, or manual mode.

[0078] User interface 116 is configured to provide information to or receive information from a user of vehicle 100. Optionally, user interface 116 may include one or more input / output devices in the set of peripheral devices 108, such as a wireless communication system 146, an onboard computer 148, a microphone 150, and a speaker 152.

[0079] The computer system 101 may control the functions of the vehicle 100 based on inputs received from various subsystems (e.g., the cruise system 102, the sensor system 104, and the planning and control system 106) and the user interface 116. In some embodiments, the computer system 101 may perform operations to provide control over many aspects of the vehicle 100 and its subsystems.

[0080] Optionally, one or more of the aforementioned components may be located separately from or associated with vehicle 100. For example, memory 114 may be partially or completely separate from vehicle 100. The aforementioned components may be communicatively coupled together in a wired or wireless manner.

[0081] Optionally, the above-mentioned components are merely examples. In practical applications, components may be added or deleted from the components in the above-mentioned modules based on practical requirements. FIG. 1 should not be understood as being limited to the embodiments of the present invention.

[0082] An autonomous vehicle traveling on a road, such as vehicle 100 described above, may recognize objects in the environment surrounding vehicle 100 and determine to make adjustments to the vehicle's trajectory, including adjustments to the vehicle's speed. The object may be another vehicle, a traffic control device, or another type of object. In some examples, each recognized object may be considered independently, and based on characteristics of each object, such as the object's current speed, a trajectory plan for the autonomous vehicle may be determined, including the object's acceleration, the spacing between the object and the vehicle, and the adjusted speed.

[0083] Optionally, a computing device associated with autonomous vehicle 100 (e.g., computer system 101 and computer vision system 138 in FIG. 1 ) may predict the behavior of a recognized object based on the characteristics of the recognized object and the status of the recognized object's surrounding environment (traffic volume on the road, rain, ice, etc.). Optionally, all recognized objects depend on each other's behavior, and therefore all recognized objects may be considered together to predict the behavior of a single recognized object. Vehicle 100 may plan the driving trajectory (including speed) of vehicle 100 based on the predicted behavior of the recognized objects. In addition, a specific status in which the autonomous vehicle needs to be adjusted (e.g., for speed adjustment, such as increasing the vehicle's speed, decreasing the vehicle's speed, or stopping the vehicle) is determined for the vehicle based on the planning result. In this process, the driving trajectory of vehicle 100 may also be determined by considering other factors, such as the horizontal position of vehicle 100 on the road on which the vehicle is traveling, the curvature of the road, the proximity between static objects and the vehicle, and the proximity between dynamic objects and the vehicle.

[0084] In addition to performing speed adjustments for the autonomous vehicle, based on the planning results, the computing device may provide instructions to modify the steering angle of the vehicle 100 so that the autonomous vehicle can follow a given trajectory or maintain a safe horizontal distance and a safe vertical distance from objects close to the autonomous vehicle (e.g., cars in adjacent lanes on the road).

[0085] The vehicle 100 may be a car, truck, motorcycle, bus, boat, airplane, helicopter, lawn mower, recreational vehicle, amusement park vehicle, construction device, motorized trolley, golf cart, train, push cart, etc. This is not a specific limitation of the embodiments of the present invention.

[0086] As shown in FIG. 2, computer system 101 may include a processor 103, a system bus 105, a display adapter (video adapter) 107, a display 109, a bus bridge 111, an input / output bus (in / out bus, I / O bus) 113, an input / output (in / out, I / O) interface 115, a Universal Serial Bus (USB) port 125, a network interface 129, a hard disk drive interface 131, a hard disk drive 133, and a system memory 135.

[0087] The processor 103 is coupled to a system bus 105. The processor 103 may be one or more processors, and each processor may include one or more processor cores. A display adapter may drive a display 109, which is coupled to the system bus 105. The system bus 105 is coupled to an input / output bus (I / O bus) 113 by using a bus bridge 111. The I / O interface 115 is coupled to the I / O bus 113. The I / O interface 115 communicates with multiple I / O devices, such as input devices 117 (including, for example, a keyboard, a mouse, and a touchscreen), a media disk (media tray) 121 (e.g., a CD-ROM and a multimedia interface), a transceiver 123 (which may transmit or receive wireless communication signals), a camera 155 (which may capture static and dynamic digital video images), and a USB port 125. Optionally, the interface connected to the I / O interface 115 may be the USB port 125.

[0088] Processor 103 may be any conventional processor, including a reduced instruction set computing ("RISC") processor, a complex instruction set computing ("CISC") processor, or a combination thereof. Optionally, the processor may be a special-purpose device such as an application specific integrated circuit ("ASIC"). Optionally, processor 103 may be a neural network processor, or a combination of a neural network processor and the aforementioned conventional processors.

[0089] Optionally, in various embodiments described herein, computer system 101 may be located remotely from the autonomous vehicle and may communicate wirelessly with autonomous vehicle 100. In some implementations, some of the processes described herein may also be performed by a processor located within the autonomous vehicle, while others are performed by a remote processor, which may include performing the actions necessary to perform a single operation.

[0090] The computer system 101 may communicate with the software deployment server 149 through a network interface 129. The network interface 129 may include, for example, a hardware network interface such as a network interface card. The network 127 may be an external network, such as the Internet, or an internal network, such as Ethernet or a virtual private network (VPN). Optionally, the network 127 may alternatively be a wireless network, for example, a Wi-Fi network or a cellular network.

[0091] Hard disk drive interface 131 is coupled to system bus 105. Hard disk drive interface 131 is connected to hard disk drive 133. System memory 135 is coupled to system bus 105. Executing in system memory 135 may include operating system 137 and application programs 143 for computer system 101.

[0092] Operating system OS 137 includes a computer shell 139 and a kernel 141. Shell 139 is the interface between the user and the operating system. Shell 139 is the outermost layer of the operating system. Shell 139 can be configured to manage the interaction between the user and the operating system: it waits for user input, interprets the user input to the operating system, and processes various output results of the operating system.

[0093] The kernel 141 includes components within an operating system that are configured to manage memory, files, peripherals, and system resources. When interacting directly with hardware, the kernel of an operating system typically executes processes, provides inter-process communication, CPU time slice management, interrupts, memory management, IO management, etc.

[0094] Application programs 143 include programs related to controlling the autonomous driving of the vehicle, such as programs for managing interactions between the autonomous vehicle and objects on the road, programs for controlling (including speed) the autonomous vehicle as it travels along a planned trajectory, or programs for controlling interactions between the autonomous vehicle and other autonomous vehicles on the road. Application programs 143 also reside in the system of software deployment server 149. In an embodiment, when application programs 143 need to be executed, computer system 101 may download application programs 143 from software deployment server 149.

[0095] Sensor 153 is associated with computer system 101. Sensor 153 is configured to detect the surrounding environment of computer system 101. For example, sensor 153 may detect animals, vehicles, targets, and crosswalks. Furthermore, the sensor may detect the surrounding environment of objects such as animals, vehicles, targets, and crosswalks, for example, other objects around the animal, weather conditions, and brightness in the surrounding environment. Optionally, when computer system 101 is located in an autonomous vehicle, the sensor may be a camera, an infrared sensor, a chemical detector, a microphone, or the like.

[0096] The computer system 101 may receive information from or transfer information to another computer system. Alternatively, sensor data collected by the sensor system 104 of the vehicle 100 may be transferred to another computer for data processing. As shown in FIG. 3 , data from the computer system 101 may be transferred over a network to a cloud-based computer 320 for further processing. The network and intermediate nodes may include a variety of structures and protocols, including the Internet, the World Wide Web, an intranet, a virtual private network, a wide area network, a local area network, a private network using one or more company-proprietary communication protocols, Ethernet, a wireless network, a hypertext transport protocol (HTTP) network, and various combinations thereof. Such communication may be performed by any device capable of transferring data to and from another computer, such as a modem or wireless interface.

[0097] In an example, computer 320 may include a server having multiple computers, such as a load-balancing server cluster, that exchanges information with different nodes of a network to receive data from computer system 101, process the data, and forward the processed data. The server may include a structure similar to that of computer system 101, and is provided with a processor 321, memory 322, instructions 323, and data 324.

[0098] Existing trajectory prediction methods can only accurately predict the trajectory of a target in the next relatively short period of time, and have relatively low accuracy in predicting the trajectory of a target in the next relatively long period of time. To clarify the problems existing in the prior art, the following describes currently used trajectory prediction methods.

[0099] In a currently commonly used trajectory prediction method, when predicting a trajectory of a target, a vehicle obtains historical movement information of the target by using sensors configured in the vehicle. The historical movement information usually includes the historical movement speed and historical movement direction of the target, and lane boundary information of the road on which the target is located. For example, the vehicle can continuously capture images of the target by using a camera configured in the vehicle, and determine the historical movement information of the target by using the captured images.

[0100] Next, the vehicle may input the historical movement information of the target into the neural network model and obtain information output by the neural network model, which is the predicted trajectory of the target.

[0101] See, for example, the exemplary diagram of the scenario shown in Figure 4. In Figure 4, the straight dashed lines represent lane markings, a first vehicle 10 is traveling within the drawing, and the first vehicle 10 may be a target for a second vehicle 20, which may predict the trajectory of the first vehicle 10 according to prior art techniques.

[0102] In this case, the second vehicle 20 may photograph the first vehicle by using a camera configured within the second vehicle 20. In Figure 4, the geographic locations of the four numbered circles on the rear side of the body of the first vehicle 10 are the locations of the first vehicle 10 when the camera photographed the first vehicle 10 at four different times. Additionally, in this scenario, the higher the number on the circle, the later the time of the photograph.

[0103] In this example, after a camera in the second vehicle 20 captures an image of the target, the second vehicle 20 processes the image to determine the historical speed and direction of travel of the first vehicle 10, as well as lane boundary information for the road on which the first vehicle 10 is located, and transmits the information to the neural network model.

[0104] After obtaining the information input by the second vehicle 20, the neural network model may output a predicted trajectory of the first vehicle. In this case, the second vehicle 20 may obtain the predicted trajectory of the first vehicle 10 based on the output of the neural network. As shown in Figure 4, the curved dashed line is the predicted trajectory of the first vehicle.

[0105] However, in this solution, the trajectory of the target is predicted only by using the historical movement information of the target. Therefore, the prediction time achievable in this solution is relatively short. In other words, this solution can only accurately predict the trajectory of the target in the next relatively short period, and the accuracy of predicting the trajectory of the target in the next relatively long period is relatively low.

[0106] In addition, currently, another trajectory prediction method can also be used. In this prediction method, when predicting the trajectory of a target, the vehicle can obtain historical movement information of the target by using sensors configured in the vehicle, and the target can be referred to as a first target. Accordingly, the historical movement information of the first target usually includes the historical movement speed and historical movement direction of the first target, and lane boundary information of the road on which the first target is located.

[0107] In the driving process, the first target needs to avoid colliding with another target of the vehicle. The other target of the vehicle may be referred to as a second target. Therefore, the first target adjusts its trajectory based on the second target; in other words, the movement status of the second target may affect the trajectory of the first target. Therefore, in this method, the vehicle may further acquire historical movement information of the second target. Accordingly, the historical movement information of the second target typically includes the historical movement speed and historical movement direction of the second target, and lane boundary information of the road on which the second target is located.

[0108] The vehicle may then input the historical movement information of the first target and the historical movement information of the second target into the neural network model, and the vehicle then uses the output of the neural network model as a predicted trajectory of the first target.

[0109] For example, see the schematic diagram shown in Figure 5. Figure 5 includes one first target 30 and three second targets 40, and the circle behind each target represents the historical trajectory of the target. In this case, the vehicle may capture images of the first target 30 and the second target 40 using a camera configured in the vehicle, determine historical movement information of the first target 30 and the second target 40 based on the captured images, and then input the historical movement information of the first target 30 and the second target 40 into a neural network model. In addition, the vehicle may obtain the output of the neural network model and use the output of the neural network model as the predicted trajectory of the first target 30.

[0110] In this method, the influence of another target on the trajectory of the first target is further taken into account. Therefore, compared with the previous method, the accuracy of trajectory prediction is improved by using this method. However, in this method, only the target's historical movement information is still used during the target's trajectory prediction. Therefore, like the previous method, this method can only accurately predict the trajectory of the target in the next relatively short period of time, and the prediction accuracy of predicting the trajectory of the target in the next relatively long period of time is relatively low.

[0111] Tests have shown that when the two prior art techniques mentioned above are used to predict a target's trajectory, only the trajectory for the next 3 to 5 seconds can be accurately predicted, and the trajectory for a longer period cannot be accurately predicted.

[0112] Furthermore, in the driving process, the vehicle usually plans its driving behavior based on the predicted trajectory of the target. For example, when it determines that the vehicle may collide with the target based on the predicted trajectory of the target, the vehicle may reduce its driving speed, apply the brakes, or stop.

[0113] However, currently used trajectory prediction methods can only accurately predict the trajectory of a target for the next relatively short period. In this case, when planning driving behavior using the current trajectory prediction method, the vehicle can only make temporary decisions based on the predicted trajectory for a relatively short period. Therefore, during the driving process, the vehicle needs to frequently adjust its driving behavior based on the predicted trajectory for a relatively short period, and as a result, the vehicle can only implement conservative planning for its driving behavior. As a result, during the vehicle driving process, phenomena such as emergency braking, emergency avoidance, frequent acceleration and deceleration, frequent stopping and stopping, and low-speed vehicle driving are relatively likely to occur, which affect the vehicle's driving and the passenger experience on the vehicle.

[0114] To solve the problem that current trajectory prediction methods can only accurately predict the trajectory of a target in the next relatively short period of time, and have relatively low accuracy in predicting the trajectory of a target in the next relatively long period of time, embodiments of the present application provide an trajectory prediction method and apparatus, thereby improving the accuracy in predicting the trajectory of a target in the next relatively long period of time.

[0115] To clarify the solution provided in the present application, the solution provided in the present application will be described below with reference to the accompanying drawings by using various embodiments.

[0116] The embodiment of the present application provides a trajectory prediction method. As shown in the schematic diagram of the operation procedure shown in Figure 6, the trajectory prediction method provided in this embodiment of the present application includes the following steps:

[0117] Step S11: Determine at least one target point, where the distance between the target point and the vehicle is included in a first range.

[0118] In this embodiment of the present application, the target points are typically in various locations and are places where the target is more likely to travel, the target may be a pedestrian or another vehicle on the road, the location may be a residence, a workplace, a garage, or a point of consumption (e.g., a supermarket or a pharmacy), etc.

[0119] At this stage, target points within a first range of distances from the vehicle may be determined based on the geographical location of the vehicle.

[0120] The first range may be a specific distance range. For example, the first range may be from 0 meters to 200 meters. In this case, the first range may be set when the vehicle is shipped. In addition, after the vehicle is shipped, the vehicle owner may further adjust the first range.

[0121] Alternatively, in another possible implementation, the vehicle may determine the first range based on the vehicle's traveling speed and a time threshold. In this case, the product of the traveling speed and the time threshold is set as the second distance. The first range is typically from the first distance to the second distance, and the first distance may be 10 meters. For example, if the vehicle's traveling speed is 60 km / h and the time threshold is 12 seconds, the second distance is 200 meters. Accordingly, the first range is from 10 meters to 200 meters.

[0122] Step S12: Obtain at least one reference trajectory between the target and the at least one target point based on the at least one target point and the geographical position of the target.

[0123] In this embodiment of the present application, a target point is a location to which the target can travel, where a reference trajectory between the target and the target point is a feasible trajectory of the target.

[0124] Step S13: Determine a predicted trajectory of the target from at least one reference trajectory.

[0125] In this embodiment of the present application, the at least one reference trajectory between the target and the at least one target point can be one or more reference trajectories. If the at least one reference trajectory between the target and the at least one target point is one reference trajectory, at this stage, the reference trajectory is determined as the predicted trajectory of the target.

[0126] In addition, if at least one reference trajectory between the target and at least one target point is a plurality of reference trajectories, according to this step, one reference trajectory is selected from the plurality of reference trajectories, and the reference trajectory is used as a predicted trajectory of the target.

[0127] While determining the predicted trajectory from the multiple reference trajectories, the degree of fit between each of the multiple reference trajectories and the movement trend of the target may be determined, and the reference trajectory that is within the reference trajectories and best fits the movement trend of the target is used as the predicted trajectory of the target.

[0128] This embodiment of the present application provides a trajectory prediction method, in which at least one target point is first determined, at least one reference trajectory between a target of a vehicle and the at least one target point is determined, and a predicted trajectory of the target is determined from the at least one reference trajectory.

[0129] In other words, the solution provided in this embodiment of the present application needs to determine a target point during target trajectory prediction. The target point is usually a location where the target is more likely to travel. In this case, during target trajectory prediction by using the solution provided in this embodiment of the present application, not only the target's historical movement information is used, but also information about the target point is used.

[0130] Compared with existing methods for predicting a target trajectory, the solution provided in this embodiment of the present application can use more diverse information during target trajectory prediction, so that the solution provided in this embodiment of the present application can not only accurately predict the target trajectory in the next relatively short period, but also improve the accuracy of predicting the target trajectory in the next relatively long period.

[0131] In the prior art, usually only the trajectory of the target for the next 3 to 5 seconds can be accurately predicted. However, the solution provided in this embodiment of the present application can accurately predict the trajectory of the target for the next 5 to 20 seconds. Compared with the prior art, the solution provided in this embodiment of the present application improves the accuracy of predicting the trajectory of the target for the next relatively long period.

[0132] Furthermore, the solution provided in this embodiment of the present application can accurately predict a target trajectory in the next relatively long period of time, and the vehicle plans its driving behavior by using the target predicted trajectory, so compared with the prior art, when the vehicle plans its driving behavior by using the predicted trajectory determined in this embodiment of the present application, the vehicle can plan its driving behavior in advance for the next relatively long period of time. In this way, the vehicle can make more appropriate plans in advance, thereby alleviating the current problem of inappropriate planning.

[0133] In addition, since the vehicle can plan its driving behavior for the next relatively long period of time, phenomena such as emergency braking, emergency avoidance, and frequent stopping and running of the vehicle can be reduced. Therefore, the solution provided in this embodiment of the present application can further improve the planning ability of the vehicle for autonomous driving while ensuring the safety and comfort of the vehicle.

[0134] In the solution provided in this embodiment of the present application, at least one target point needs to be determined to obtain a reference trajectory based on the target point. The operation of determining at least one target point may include the following steps:

[0135] First, a first reference point is determined by referencing a map on which the reference points are marked or by referencing a text file indicating the reference points, and the distance between the first reference point and the vehicle is within a first range.

[0136] In the solution provided in this embodiment of the present application, a map can be applied during determining the first reference point. Compared with the existing map, a layer marked with at least one reference point is added to the map applied in this embodiment of the present application. The reference point is usually a location to which the target can travel, such as an entrance to an apartment complex, a garage, a hospital, a government agency, a company, a bus stop, a shopping mall, a supermarket, an overpass, a crossroad, or a pedestrian crossing.

[0137] To clarify the differences between the map applied in this embodiment of the present application and existing maps, an example is disclosed below. The example is provided in Figures 7(a) and 7(b). Figure 7(a) is a schematic diagram of a map in the prior art, which represents a university campus. Figure 7(b) is a map with reference points marked and applied in this embodiment of the present application. The map not only includes the university campus shown in Figure 7(a), but also includes layers with multiple reference points marked. The reference points include office buildings 22 to 25 and the entrances to the university campus.

[0138] The map with the marked reference points may be stored in a cloud server, and the vehicle performing the trajectory prediction interacts with the cloud server to query the map. In addition, the map with the marked reference points may alternatively be stored in the vehicle. In addition, the vehicle may determine the vehicle's geographic location on the map through positioning, and then determine a first reference point whose distance from the vehicle falls within a first range by querying the map.

[0139] In addition, the map may be pre-drawn. In addition, in the application process, the cloud server or the vehicle may further update the map based on the received operation, and add or remove reference points marked on the map.

[0140] In another implementation solution, the vehicle may determine the first reference point by consulting a text file. The text file indicates the reference points. For example, the text file may be a table file that records the longitude and latitude positions of the reference points. In this case, the first reference point may alternatively be determined by consulting the text file and the vehicle's geographic location.

[0141] In this implementation solution, the text file can also be stored in the cloud server or the vehicle, and in the application process, the cloud server or the vehicle can further update the text file based on the received operation to add or delete a reference point.

[0142] Next, if the number of first reference points is not greater than n, the first reference points are determined as target points, where n is a predetermined natural number; or, if the number of first reference points is greater than n, at least one target point is selected from the first reference points.

[0143] The first reference points may be determined by referencing a map on which the reference points are marked or by referencing a text file indicating the reference points. There may be one or more first reference points. In this case, if the number of first reference points is not greater than n, it indicates that the number of first reference points is relatively small, and all of the first reference points are determined to be target points; or if the number of first reference points is greater than n, it indicates that the number of first reference points is relatively large, and the first reference points may be further screened to determine target points from the first reference points.

[0144] In this specification, n is a predetermined natural number. In a possible implementation, n may be 3.

[0145] In a possible screening method, among the plurality of first reference points, n first reference points corresponding to the largest flow of people within a certain period of time can be determined as target points. The period can be 24 hours before the current time, or the period can be a period relatively close to the current time. For example, if the current time is 11:00 AM, the period can be from 10:00 to 11:00 yesterday.

[0146] Alternatively, in another possible screening method, the vehicle may determine a target point from the first reference point based on the target's survey information. When the target is confirmed to have a given permission, the target's survey information may be the target's appearance frequency, appearance time, or consumption record at the first reference point. The consumption record typically records the appearance time and number of appearances of the target at each consumption location. The appearance frequency and appearance time of the target at the consumption location may be determined based on the consumption record.

[0147] In this case, while selecting a target point from the first reference points, a number of first reference points within the first reference points and corresponding to the highest frequency towards which the target has traveled may be determined as the target point based on the survey information of the target; or a first reference point towards which the target has traveled with a historical frequency greater than a first threshold value may be determined as the target point; or a first number of b first reference points towards which the target has traveled at a time closest to the current time may be determined as the target point, where both a and b are predetermined natural numbers.

[0148] For example, when a vehicle is traveling in an industrial park, each first reference point is a location in the industrial park, such as an office building, a supermarket, or a cafeteria in the industrial park, and each person traveling in the industrial park is a staff member of the industrial park. To improve the safety of the vehicle traveling in the industrial park, the staff member of the industrial park agrees that the vehicle may acquire survey information of the staff member. The survey information may include the appearance frequency or consumption record of the staff member at each first reference point.

[0149] In this case, when a staff member of the industrial park is walking within the industrial park, the staff member is the target and the vehicle may select a target point from the first reference point by using survey information of the staff member.

[0150] After the target point is determined, at least one reference trajectory between the target and the at least one target point can be determined. Since the target point is a location to which the target can travel, the at least one reference trajectory usually includes a trajectory along which the target is trying to travel. In the solution provided in this embodiment of the present application, the reference trajectory can be determined in several ways.

[0151] In a possible implementation, the step of obtaining at least one reference trajectory between the target and the at least one target point comprises the steps of: Determining a historical trajectory from the target to at least one target point as at least one reference trajectory.

[0152] In this embodiment of the present application, the target point is typically a location to which the target has traveled, in which case the historical trajectory from the target to the target point may be used as the reference trajectory.

[0153] Alternatively, in another possible implementation, determining at least one reference trajectory between the target and the at least one target point comprises the following steps: determining the shortest trajectory between the target and any one of the at least one target points as a reference trajectory when the traffic rules are satisfied;

[0154] In this implementation, when the traffic rules are satisfied, fitting can be performed on the geographical location of the target point and the geographical location of the target, possible trajectories between the target point and the target are determined by fitting, and the shortest trajectory between the target point and the target is determined as the reference trajectory.

[0155] This application provides a method for determining a reference trajectory based on the historical trajectory of a target, and a method for determining a reference trajectory based on the shortest trajectory between the target and a target point. In addition, the reference trajectory can alternatively be determined based on both methods. In this case, the historical trajectory from the target to the target point is determined to be the reference trajectory, and when traffic rules are satisfied, the shortest trajectory between the target and the target point is also determined to be the reference trajectory.

[0156] After at least one reference trajectory is determined, the predicted trajectory of the target needs to be determined from the at least one reference trajectory. If there is only one reference trajectory, the predicted trajectory of the target is determined from the at least one reference trajectory, and the operation can be implemented according to the following steps:

[0157] Step 1: Determine the target's short-term trajectory in the next s seconds based on the target's movement trend.

[0158] In this embodiment of the present application, the movement trend of the target may include information such as the current movement direction and movement speed of the target, etc. In this case, the trajectory of the target in the next s seconds may be predicted based on the movement trend of the target to determine the short-term trajectory of the target.

[0159] The value of s may be determined based on the accuracy of the method for determining the short-term orbit. Generally, the value of s may range from 1 to 3.

[0160] Step 2: Determine the reliability of at least one reference trajectory based on the short-term trajectory.

[0161] The confidence of the reference trajectory indicates the degree of fit between the reference trajectory and the short-term trajectory. Additionally, a higher confidence of the reference trajectory indicates that the reference trajectory fits better to the short-term trajectory.

[0162] Step 3: The reference trajectory with the highest reliability is determined as the predicted trajectory of the target.

[0163] In this embodiment of the present application, a short-term trajectory of a target in the next short period is predicted based on the movement trend of the target, and a reliability of each of at least one reference trajectory and the short-term trajectory is determined. The reliability may reflect the degree of fit between each of the at least one reference trajectory and the short-term trajectory, and the reference trajectory with the highest reliability is the reference trajectory that best fits the short-term trajectory. In this case, when the reference trajectory with the highest reliability is determined as the predicted trajectory of the target, the predicted trajectory is the reference trajectory that best fits the movement trend of the target among the at least one reference trajectory. Therefore, according to the above steps, the reference trajectory that best fits the movement trend of the target can be determined as the predicted trajectory of the target from the at least one reference trajectory.

[0164] In this embodiment of the present application, the short-term trajectory can be determined in several ways. In a possible implementation, determining the short-term trajectory of the target in the next s seconds based on the movement trend of the target includes: A step of determining the trajectory obtained after the target moves for s seconds under the current movement trend as the short-term trajectory.

[0165] The current movement trend of the target includes the current movement speed and current movement direction of the target, in which case the trajectory n acquired after the target moves for s seconds under the current movement trend can be determined as the short-term trajectory.

[0166] In another possible implementation, determining the short-term trajectory of the target in the next s seconds based on the movement trend of the target includes:

[0167] Step 1: determining a second parameter of the target based on the movement trend of the target, where the second parameter of the target includes the historical movement speed and historical movement direction of the target, the geographical location of the target, and environmental information of the target.

[0168] The target's environmental information typically includes lane boundary information, traffic light information, cross road information, etc. at the target's geographic location.

[0169] In a possible implementation, the target is continuously photographed, and then image processing is performed on the images including the target, and a second parameter of the target can be determined based on the image processing results.

[0170] Furthermore, since the target needs to avoid other targets during the traveling process, the second parameter of the target may further include related information of other targets around the target, which usually includes the historical moving speed and direction of the other targets, the geographical location of the other targets, etc.

[0171] Step 2: After the second parameter is input into the neural network model, obtain the short-term trajectory, which is the target short-term trajectory in the next s seconds and is output by the neural network model.

[0172] The neural network model is typically a kinematic model. For example, the neural network model may be a long-short term memory (LSTM) artificial neural network model. If the LSTM model applied in this embodiment of the present application can accurately predict the target trajectory in approximately 3 seconds, s may typically be set to 3.

[0173] The neural network model may receive a second parameter of the target, predict a trajectory of the target based on the second parameter of the target, and output a short-term trajectory of the target in the next s seconds, where the short-term trajectory of the target in the next s seconds may be determined based on the output of the neural network model.

[0174] In the above embodiment, two solutions are provided to determine the short-term trajectory of the target in the next s seconds. In practical applications, the two solutions can be applied simultaneously and alternatively, so that the predicted trajectory is jointly determined based on the short-term trajectory determined by using the two solutions.

[0175] In the solution provided in this embodiment of the present application, the predicted trajectory in the reference estimation needs to be determined based on the reliability of the reference trajectory. In a possible implementation solution, the step of determining the reliability of the at least one reference trajectory based on the short-term trajectory includes the following steps: determining a projection value of the short-term orbit of the at least one reference orbit according to the formula: wherein the projection value is the confidence of the at least one reference trajectory;

number

number

number

number

number

number

number

number

[0176] In this implementation, the reliability of at least one reference trajectory is determined by using a projection method. The projection value of a short-term trajectory relative to a reference trajectory refers to the length of the projection obtained after the short-term trajectory is projected onto the projection plane on which the reference trajectory is located. The larger the projection value of the reference trajectory, the more closely the reference trajectory fits the short-term trajectory, and the higher the reliability of the reference trajectory.

[0177] To clarify the solution for determining the reliability of a reference trajectory in this application, the following example is disclosed.

[0178] 8 is a schematic diagram of a map applied in an embodiment of the present application. Reference points are marked on the map, and the reference points include a hospital, an office building, a cafeteria, a fruit shop, a school, and a supermarket. The distances between the vehicle and each of the school and the supermarket exceed a first range, and the distances between the vehicle and each of the hospital, the office building, the cafeteria, and the fruit shop are included in the first range. Therefore, in this example, the hospital, the office building, the cafeteria, and the fruit shop can be determined as first reference points.

[0179] In this example, the target is a pedestrian, as shown in Figure 9. Additionally, the pedestrian passes through three geographic locations at times t1, t2, and t3 in the time sequence. Each of the three geographic locations is marked by a portrait in Figure 9, and the current time is time t3.

[0180] Based on the identity information of the pedestrian, it is determined that the historical frequency of the pedestrian going to the hospital is 1, the historical frequency of the pedestrian going to the office building is 26, the historical frequency of the pedestrian going to the diner is 50, and the historical frequency of the pedestrian going to the fruit shop is 10. In addition, in this example, the first threshold is 5, and the first reference points to which the target goes with a historical frequency greater than the first threshold are determined as target points. In this case, the three first reference points, i.e., the office building, the diner, and the fruit shop, may be determined as target points in this example. In addition, in FIG. 9, reference trajectories between the pedestrian and each target point are plotted, and the reference trajectories are represented by connecting lines with arrows pointing to each target point.

[0181] In this example, the pedestrian's short-term trajectory for the next s seconds can be determined based on the pedestrian's movement trend. In Figure 9, the short-term trajectory is represented by a line segment with an arrow. In addition, the short-term trajectory is represented by a thicker line segment to separate it from the reference trajectory, compared to the connecting line representing the reference trajectory.

[0182] In this case, the projection value of the short-term trajectory onto the reference trajectory needs to be determined. In Figure 9, the dashed line represents a perpendicular line to the reference trajectory, there is an intersection between the perpendicular line and the reference trajectory, the connecting line between the intersection and the geographical position of the pedestrian at time t3 is the projection of the short-term trajectory onto the reference trajectory, and the length of the projection is the projection value of the short-term trajectory onto the reference trajectory.

[0183] 9, it can be determined that the projection value of the reference trajectory between the pedestrian and the fruit stand is the largest, and the projection value can be used as the reliability of the reference trajectory between the pedestrian and the fruit stand, so that the reliability of the reference trajectory between the pedestrian and the fruit stand is the largest. In this case, it can be determined that the reference trajectory between the pedestrian and the fruit stand is the predicted trajectory of the pedestrian.

[0184] In the above embodiment, a method for determining a predicted trajectory based on a reliability is described. In addition, the predicted trajectory can be determined in other manners. In another possible implementation solution of the present application, the step of determining a predicted trajectory of the target from at least one reference trajectory includes the following steps:

[0185] First, the historical frequency or time that the target travels on at least one reference trajectory is determined.

[0186] Next, a predicted trajectory is determined from the at least one reference trajectory based on a historical frequency or a historical time that the target has traveled the at least one reference trajectory, where the predicted trajectory is the reference trajectory along which the target has traveled most historically frequently, or the predicted trajectory is the reference trajectory along which the target has traveled the closest historical time to the current time.

[0187] In other words, in this implementation, the reference trajectory along which the target travels with the highest historical frequency is used as the predicted trajectory, or the reference trajectory along which the target travels with the closest historical time to the current time is used as the predicted trajectory.

[0188] According to the above-mentioned solution, a target predicted trajectory can be determined. In addition, the vehicle can plan its driving behavior based on the target predicted trajectory. In this case, the solution provided in this embodiment of the present application further includes the following steps: adjusting a driving behavior of the vehicle after determining based on the predicted trajectory that the vehicle is about to collide with the target.

[0189] After determining that the vehicle is about to collide with a target based on the predicted trajectory, the vehicle must adjust its driving behavior to avoid the collision. For example, the vehicle may avoid the collision by performing maneuvers such as accelerating and swerving, driving at a slower speed, or stopping and waiting.

[0190] Additionally, if it is determined based on the predicted trajectory that the vehicle will not collide with the target, the vehicle may maintain its current driving behavior.

[0191] Existing trajectory prediction technologies can only accurately predict the trajectory of a target in the next relatively short period. Therefore, while planning the driving behavior of a vehicle based on the trajectory predicted by using the conventional technology, the driving behavior of the vehicle can only be planned for a short period. In this case, the vehicle usually needs to frequently adjust the driving behavior of the vehicle, and in some cases, the vehicle may even stop and start frequently or the vehicle speed may become unstable.

[0192] However, according to the above-mentioned steps of the present application, the vehicle can plan its driving behavior based on the predicted trajectory. Furthermore, since the solution in this embodiment of the present application can accurately predict the trajectory for the next relatively long period, compared with the prior art, when the vehicle plans its driving behavior by using the predicted trajectory determined in this embodiment of the present application, the vehicle can plan its driving behavior for the next relatively long period in advance. This avoids frequent adjustments of the driving behavior, reduces phenomena such as emergency braking, emergency avoidance, and frequent stopping and running of the vehicle, and ensures the safety and comfort of the vehicle.

[0193] The following provides apparatus embodiments of the present application, which can be used to perform method embodiments of the present application. For details not disclosed in the apparatus embodiments of the present application, please refer to the method embodiments of the present application.

[0194] In the implementation of the above-mentioned embodiment, the embodiment of the present application discloses a trajectory prediction device. As shown in the structural schematic diagram shown in Figure 10, the driving behavior monitoring device disclosed in this embodiment of the present application includes a processor 1110 and a transceiver interface 1120.

[0195] The transceiver interface 1120 is configured to receive information about the sensor.

[0196] In a possible implementation, the trajectory prediction device disclosed in this embodiment of the present application is applied to a vehicle, in which case the sensor may be a sensor disposed within the vehicle.

[0197] In this case, the sensor may include a camera, and the camera may continuously capture images of the target, and then determine relevant information of the target by analyzing the images including the target. For example, based on the images obtained by continuously capturing images of the target using the camera, historical movement information of the target may be determined, and the movement trend of the target may be determined.

[0198] Additionally, the sensor may further include a radar, in which case the vehicle may determine the moving speed of the target by using the radar.

[0199] The processor 1110 The vehicle is configured to: determine at least one target point based on the information about the sensor, wherein a distance between the target point and the vehicle is within a first range; obtain at least one reference trajectory between the target and the at least one target point based on the at least one target point and a geographical position of the target; and determine a predicted trajectory of the target from the at least one reference trajectory.

[0200] During the trajectory prediction of a target by using the solution provided in this embodiment of the present application, the target point needs to be determined. Therefore, during the trajectory prediction of a target by using the solution provided in this embodiment of the present application, not only the historical movement information of the target is used, but also the information about the target point is used.

[0201] Compared with existing methods for predicting a target trajectory, the solution provided in this embodiment of the present application can use more diverse information during target trajectory prediction, so that the solution provided in this embodiment of the present application can not only accurately predict the target trajectory in the next relatively short period, but also improve the accuracy of predicting the target trajectory in the next relatively long period.

[0202] In the prior art, usually only the trajectory of the target for the next 3 to 5 seconds can be accurately predicted. However, the solution provided in this embodiment of the present application can accurately predict the trajectory of the target for the next 5 to 20 seconds. Compared with the prior art, the solution provided in this embodiment of the present application improves the accuracy of predicting the trajectory of the target for the next relatively long period.

[0203] In the solution provided in this embodiment of the present application, the processor: determining a first reference point by consulting a map on which reference points are marked or by consulting a text file indicating said reference points, wherein the distance between said first reference point and said vehicle is within said first range; and If the number of first reference points is not greater than n, determining the first reference points as the target points, where n is a predetermined natural number; or if the number of first reference points is greater than n, selecting the at least one target point from the first reference points. The present invention is specifically configured to:

[0204] In the solution provided in this embodiment of the present application, a map may be applied while determining the first reference point. Compared with an existing map, a layer marked with at least one reference point is added to the map applied in this embodiment of the present application. The reference point is usually a location toward which a target may travel, such as an entrance to an apartment complex, a garage, a hospital, a government agency, a company, a bus stop, a shopping mall, a supermarket, an overpass, a crossroad, or a pedestrian crossing. In this case, the processor may determine the first reference point by consulting the map marked with the reference point.

[0205] The first reference points may be determined by referencing a map on which the reference points are marked or by referencing a text file indicating the reference points. There may be one or more first reference points. In this case, if the number of first reference points is not greater than n, it indicates that the number of first reference points is relatively small, and all of the first reference points are determined to be target points; or if the number of first reference points is greater than n, it indicates that the number of first reference points is relatively large, and the first reference points may be further screened to determine target points among the first reference points.

[0206] n is a predetermined natural number. In a possible implementation, n may be 3.

[0207] In the solution provided in this embodiment of the present application, the processor may determine a reference trajectory. In a possible implementation solution, the processor is specifically configured to determine a historical trajectory from the target to at least one target point as the at least one reference trajectory.

[0208] Alternatively, in another possible implementation solution, the processor is specifically configured to determine the shortest trajectory between the target and any one of the at least one target points as the reference trajectory when the traffic rule is satisfied. In this case, when the traffic rule is satisfied, the processor may perform fitting on the geographical position of the target point and the geographical position of the target, determine possible trajectories between the target point and the target by fitting, and determine the shortest trajectory between the target point and the target as the reference trajectory.

[0209] In the solution provided in this embodiment of the present application, the processor: determining a short-term trajectory of the target for the next s seconds based on a movement trend of the target; determining a reliability of the at least one reference trajectory based on the short-term trajectory, where the reliability of the reference trajectory indicates a degree of fit between the reference trajectory and the short-term trajectory; and determining the reference trajectory having the highest reliability as the predicted trajectory of the target; The present invention is specifically configured to:

[0210] The processor may determine the short-term trajectory in several ways. In a possible implementation solution, the processor is specifically configured to determine the trajectory obtained after the target moves for s seconds under the current movement trend as the short-term trajectory.

[0211] Alternatively, in another possible implementation solution, the processor determines second parameters of the target based on the movement trend of the target, where the second parameters of the target include a historical movement speed and a historical movement direction of the target, a geographical location of the target, and environmental information of the target; and obtaining the short-term trajectory output by the neural network model of the target for the next s seconds after the second parameter is input into the neural network model; The present invention is specifically configured to:

[0212] In the solution provided in this embodiment of the present application, the processor: specifically configured to determine a projection value of the short-term trajectory of the at least one reference trajectory according to the following formula, wherein the projection value is the confidence level of the at least one reference trajectory;

number

number

number

number

number

number

number

number

[0213] Additionally, the processor may further determine the predicted trajectory from the reference trajectory based on a historical frequency or historical time that the target travels toward the reference trajectory. Determining a historical frequency or time that the target travels toward at least one reference trajectory; and determining a predicted trajectory from at least one reference trajectory based on a historical frequency or a historical time that the target travels toward at least one reference trajectory, the predicted trajectory being the reference trajectory along which the target travels most frequently in history, or the predicted trajectory being the reference trajectory along which the target travels with the closest historical time to the current time; The present invention is specifically configured to:

[0214] Furthermore, a vehicle to which the device provided in this embodiment of the present application is applied may adjust the vehicle's driving behavior based on the predicted trajectory of the target.

[0215] In this case, the processor further comprises: and adjusting a driving behavior of the vehicle after determining, based on the predicted trajectory, that the vehicle is about to collide with the target.

[0216] After determining that the vehicle is about to collide with the target based on the predicted trajectory, the vehicle needs to adjust its driving behavior to avoid the collision. For example, the vehicle may avoid the collision by performing operations such as accelerating and diverting, driving at a slower speed, or stopping and waiting. The safety of the vehicle may be improved by adjusting the driving behavior of the vehicle.

[0217] Additionally, if it is determined based on the predicted trajectory that the vehicle will not collide with the target, the vehicle may maintain its current driving behavior.

[0218] Existing trajectory prediction technologies can only accurately predict the trajectory of a target in the next relatively short period. Therefore, while planning the driving behavior of a vehicle based on the trajectory predicted by using the conventional technology, the driving behavior of the vehicle can only be planned for a short period. In this case, the vehicle usually needs to frequently adjust the driving behavior of the vehicle, and in some cases, the vehicle may even stop and start frequently or the vehicle speed may become unstable.

[0219] However, according to the above-mentioned steps of the present application, the vehicle can plan its driving behavior based on the predicted trajectory. Furthermore, since the solution in this embodiment of the present application can accurately predict the trajectory for the next relatively long period, compared with the prior art, when the vehicle plans its driving behavior by using the predicted trajectory determined in this embodiment of the present application, the vehicle can plan its driving behavior for the next relatively long period in advance. This avoids frequent adjustments of the driving behavior, reduces phenomena such as emergency braking, emergency avoidance, and frequent stopping and running of the vehicle, and ensures the safety and comfort of the vehicle.

[0220] An embodiment of the present application provides a map. The map includes a first layer. The first layer includes at least one reference point, the reference point including at least one of a business industry location, an enterprise location, and a service industry location.

[0221] Compared with existing maps, the map provided in this embodiment of the present application additionally has reference points marked thereon, which are located on the first layer of the map. The reference points may include business industry locations, such as shopping malls and supermarkets. In addition, the reference points may further include enterprise locations, such as office buildings. Alternatively, the reference points may further include service industry locations, such as hospitals.

[0222] Additionally, reference points may further include other locations where the target may reside, such as, for example, apartment complex entrances, garages, cross streets, and crosswalks.

[0223] For the maps provided in this embodiment of the present application, please refer to the schematic diagrams shown in FIG. 7(b) and FIG.

[0224] The first reference point can be determined by consulting the map provided in this embodiment of the present application. The first reference point is a reference point whose distance from the vehicle is within a first range. In this case, at least one target point can be determined by using the first reference point, so that the vehicle predicts the trajectory of the target by using the at least one target point.

[0225] In the process of predicting the trajectory of a target, the map provided in this embodiment of the present application helps to find reference points to which the target may travel, and the vehicle determines the target point based on the discovered reference points and the survey information of the target. Therefore, when the vehicle predicts the trajectory of a target by using the map provided in this embodiment of the present application, not only the historical movement information of the target but also information about the target point is used, which helps to improve the accuracy of the vehicle's prediction of the trajectory of the target in the next relatively long period of time.

[0226] In a possible implementation, the map provided in this embodiment of the present application is a map acquired after the first layer is set on an existing high-resolution map. The accuracy of absolute coordinates of the high-resolution map is relatively high. The accuracy of absolute coordinates is the accuracy between targets on the map and objects in the actual external world. In addition, the road traffic information elements included in the high-resolution map are richer and more detailed. In this case, each reference point marked on the map provided in this embodiment of the present application is more accurate, which helps to improve the accuracy of trajectory prediction performed by the vehicle.

[0227] Furthermore, in the map provided in this embodiment of the present application, the reference point is a location where a number of people not less than a second threshold passes during a first period of time.

[0228] The first time period can be 24 hours prior to the present time, or the first time period can be a period relatively close to the present time. For example, if the present time is 11:00 AM, the first time period can be from 10:00 to 11:00 yesterday.

[0229] The reference point is a location where a number of people not less than the second threshold passes during the first period, so the reference point is a location where the flow of people is relatively high and the probability that the target will travel to the reference point is relatively high. In this case, the first reference point determined by the vehicle by consulting the map is a location where the target is more likely to travel toward. Therefore, when the vehicle performs trajectory prediction by using the first reference point, the accuracy of performing trajectory prediction can be improved.

[0230] Accordingly, in accordance with the above-mentioned method, the embodiment of the present application further discloses a terminal device. As shown in the structural schematic diagram shown in Figure 11, the terminal device includes: It includes at least one first processor 1101 and a memory.

[0231] The memory is configured to store program instructions.

[0232] The first processor calls and executes program instructions stored in the memory, so that the terminal device is configured to perform all or some of the steps in the embodiment corresponding to FIG.

[0233] Furthermore, the terminal device may further include a transceiver 1102 and a bus 1103 , and the memory includes a random access memory 1104 and a read-only memory 1105 .

[0234] The first processor is coupled to the transceiver, the random access memory, and the read-only memory via a bus. When the terminal needs to run, a basic input / output system, such as a basic input / output system (BIO) built in the read-only memory, or a boot loader boot system in an embedded system, initiates booting of the terminal and enters a normal execution state. After the terminal enters the normal execution state, application programs and an operating system are executed in the random access memory, so that the terminal executes all or some of the steps in the embodiment corresponding to FIG. 6.

[0235] The apparatus in this embodiment of the present invention may correspond to the trajectory prediction apparatus in the embodiment corresponding to Figure 6. In addition, a processor or the like in the apparatus may implement the functions of the trajectory prediction apparatus in the embodiment corresponding to Figure 6, or various steps and methods implemented by the trajectory prediction apparatus. For the sake of brevity, details will not be described again in this specification.

[0236] In a specific implementation, the embodiment of the present application further provides a computer-readable storage medium. The computer-readable storage medium includes instructions. When the computer-readable storage medium disposed in any device is executed on a computer, all or some of the steps in the embodiment corresponding to FIG. 6 can be implemented. The storage medium of the computer-readable medium can be a magnetic disk, an optical disk, a read-only memory (ROM for short), a random access memory (RAM for short), etc.

[0237] In addition, another embodiment of the present application further discloses a computer program product including instructions, which, when executed on an electronic device, enable the electronic device to implement all or some of the steps in the embodiment corresponding to FIG.

[0238] The various exemplary logic units and circuits described in the embodiments of this application may be general-purpose processors. The described functions may be implemented or operated using designs of digital information processors, application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs) or other programmable logic devices, discrete gate or transistor logic, discrete hardware components, or any combination thereof. The general-purpose processor may be a microprocessor. Optionally, the general-purpose processor may be any conventional processor, controller, microcontroller, or state machine. The processor may alternatively be implemented by a combination of computing devices, such as, for example, a digital information processor and a microprocessor, multiple microprocessors, one or more microprocessors and a digital information processor core, or any other similar structure.

[0239] The steps of a method or algorithm described in the embodiments of the present application may be embodied directly in hardware, in a software unit executed by a processor, or a combination thereof. The software unit may be stored in RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, a hard disk, a removable magnetic disk, a CD-ROM, or any other form of storage medium known in the art. For example, the storage medium may be connected to the processor so that the processor can read information from and write information to the storage medium. Alternatively, the storage medium may be integrated with the processor. The processor and the storage medium may be located in an ASIC, and the ASIC may be located in the UE. Optionally, the processor and the storage medium may be located in different components of the UE.

[0240] It should be understood that the sequence numbers in the processes do not imply the order of execution in various embodiments of the present application, and the order of execution of the processes should be determined based on the functions and internal logic of the processes, and should not be construed as any limitation on the process of implementation of the embodiments of the present application.

[0241] All or some of the above-described embodiments may be implemented by software, hardware, firmware, or any combination thereof. When software is used to implement an embodiment, all or part of the embodiment may be implemented in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the procedures or functions according to the embodiments of the present application are generated. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or another programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted from a computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions may be transmitted from a website, computer, server, or data center to another website, computer, server, or data center via a wired (e.g., coaxial cable, optical fiber, or digital subscriber line (DSL)) or wireless (e.g., infrared, radio wave, or microwave) method. The computer-readable storage medium may be any available medium accessible by a computer, or may be a data storage device integrating one or more available media, such as a server or data center. The available media may be magnetic media (e.g., floppy disks, hard disks, or magnetic tapes), optical media (e.g., DVDs), or semiconductor media (e.g., solid state disks (SSDs)).

[0242] Cross-references may be made to the same and similar parts between the embodiments in this specification. Each embodiment will focus on the differences from other embodiments. In particular, the device and system embodiments are basically similar to the method embodiments, and therefore will be briefly described. For relevant parts, please refer to the partial description in the method embodiments.

[0243] Those skilled in the art can clearly understand that the technology in the embodiments of the present invention can be implemented by software in addition to a required general hardware platform. Based on such understanding, the technical solutions of the embodiments of the present invention can be essentially implemented, or the parts that contribute to the current technology can be implemented in the form of a software product. The computer software product is stored in a storage medium such as a ROM / RAM, a hard disk, or an optical disk, and includes several instructions that instruct a computer device (which may be a personal computer, a server, a network device, etc.) to execute the methods described in the embodiments or several parts of the embodiments of the present invention.

[0244] Cross-references may be made to the same and similar parts between the embodiments in this specification. In particular, the road constraint determination device embodiment disclosed in this application is basically similar to the method embodiment, and therefore will be briefly described. For relevant parts, please refer to the description in the method embodiment.

[0245] The above implementations of the present invention are not intended to limit the protection scope of the present invention.

Claims

1. 1. A method for trajectory prediction executed by a processor included in a computer system, comprising: determining, by the processor, at least one target point, wherein a distance between the target point and the vehicle is within a first range; obtaining, by the processor, at least one reference trajectory between the target and the at least one target point based on the at least one target point and a geographical position of the target; and determining, by the processor, a reference trajectory along which the target moves most frequently in history or a reference trajectory along which the target moves for a historical time closest to the present time from the at least one reference trajectory based on a historical frequency or a historical time at which the target moves to the at least one reference trajectory, as a predicted trajectory of the target; A trajectory prediction method comprising:

2. 1. A method for trajectory prediction executed by a processor included in a computer system, comprising: determining, by the processor, at least one target point, wherein a distance between the target point and the vehicle is within a first range; obtaining, by the processor, at least one reference trajectory between the target and the at least one target point based on the at least one target point and a geographical position of the target; and determining, by the processor, a predicted trajectory of the target from the at least one reference trajectory based on a historical frequency or a historical time that the target has traveled to the at least one reference trajectory. Equipped with determining the at least one target point includes determining the at least one target point from a plurality of reference points based on survey information of the target for the plurality of reference points; A trajectory prediction method, wherein the target survey information includes the target's appearance frequency, appearance time, or consumption record at the multiple reference points.

3. The step of determining at least one target point comprises: determining a first reference point by consulting a map on which reference points are marked or by consulting a text file indicating said reference points, wherein the distance between said first reference point and said vehicle is within said first range; and If the number of first reference points is not greater than n, determining the first reference points as the target points, where n is a predetermined natural number; or selecting the at least one target point from the first reference points if the number of first reference points is greater than n; The trajectory prediction method of claim 1 , comprising:

4. The step of obtaining at least one reference trajectory between a target and the at least one target point comprises: determining a historical trajectory from the target to the at least one target point as the at least one reference trajectory; or determining a reference trajectory as the shortest trajectory between the target and any one of the at least one target points when a traffic rule is satisfied; The trajectory prediction method of claim 1 , comprising:

5. 1. A method for trajectory prediction executed by a processor included in a computer system, comprising: determining, by the processor, at least one target point, wherein a distance between the target point and the vehicle is within a first range; obtaining, by the processor, at least one reference trajectory between the target and the at least one target point based on the at least one target point and a geographical position of the target; and determining, by the processor, a predicted trajectory of the target from the at least one reference trajectory based on a historical frequency or a historical time that the target has traveled to the at least one reference trajectory. Equipped with determining a predicted trajectory from the target to the at least one reference trajectory, determining a short-term trajectory of the target for the next s seconds based on a movement trend of the target; determining a reliability of the at least one reference trajectory based on the short-term trajectory, wherein the reliability of the reference trajectory indicates a degree of fit between the reference trajectory and the short-term trajectory; and determining the reference trajectory having the highest reliability as the predicted trajectory of the target; Including, The step of determining a short-term trajectory of the target for the next s seconds based on a movement trend of the target includes: determining the trajectory obtained after the target has moved for s seconds under the current movement trend as the short-term trajectory; or determining second parameters of the target based on the movement trend of the target, wherein the second parameters of the target include a historical movement speed and a historical movement direction of the target, a geographic location of the target, and environmental information of the target; and obtaining the short-term trajectory output by the neural network model of the target for the next s seconds after the second parameter is input into the neural network model; A method for predicting a trajectory, comprising:

6. 1. A method for trajectory prediction executed by a processor included in a computer system, comprising: determining, by the processor, at least one target point, wherein a distance between the target point and the vehicle is within a first range; obtaining, by the processor, at least one reference trajectory between the target and the at least one target point based on the at least one target point and a geographical position of the target; and determining, by the processor, a predicted trajectory of the target from the at least one reference trajectory. Equipped with determining a predicted trajectory from the target to the at least one reference trajectory, determining a short-term trajectory of the target for the next s seconds based on a movement trend of the target; determining a reliability of the at least one reference trajectory based on the short-term trajectory, wherein the reliability of the reference trajectory indicates a degree of fit between the reference trajectory and the short-term trajectory; and determining the reference trajectory having the highest reliability as the predicted trajectory of the target; Including, determining a reliability of the at least one reference trajectory based on the short-term trajectory, determining a projection value of said short-term orbit on said at least one reference orbit according to the formula: wherein the projection value is the confidence of the at least one reference trajectory; [Equation 33] and [Equation 34] is the reference orbit [Equation 35] above short-term orbit [Equation 36] is the projection value of [Equation 37] is the short-term orbit [Number 38] and the reference trajectory [0.39] and the reference orbit is an included angle between [Equation 40] is any one of the at least one reference trajectory.

7. adjusting, by the processor, a driving behavior of the vehicle after the processor determines, based on the predicted trajectory, that the vehicle is about to collide with the target. The trajectory prediction method according to claim 1 , further comprising:

8. Processor and Transceiver Interface where the transceiver interface is configured to receive information about the sensor; The processor is configured to: determine at least one target point based on the information about the sensor, wherein a distance between the target point and the vehicle is within a first range; obtain at least one reference trajectory between the target and the at least one target point based on the at least one target point and a geographical position of the target; and determine, based on a historical frequency or a historical time at which the target has moved to at least one reference trajectory, a reference trajectory along which the target moves with the highest historical frequency or a reference trajectory along which the target moves with the closest historical time to a current time from the at least one reference trajectory as a predicted trajectory of the target. Trajectory prediction device.

9. The processor: determining a first reference point by consulting a map on which reference points are marked or by consulting a text file indicating said reference points, wherein a distance between said first reference point and said vehicle is within said first range; and If the number of first reference points is not greater than n, determining the first reference points as the target points, where n is a predetermined natural number; or if the number of first reference points is greater than n, selecting the at least one target point from the first reference points. The trajectory prediction device of claim 8 specifically configured to:

10. 9. The trajectory prediction device of claim 8, wherein the processor is specifically configured to determine a historical trajectory from the target to the at least one target point as the at least one reference trajectory, or the processor is specifically configured to determine, when a traffic rule is satisfied, the shortest trajectory between the target and any one of the at least one target point as the reference trajectory.

11. Processor and Transceiver Interface where the transceiver interface is configured to receive information about the sensor; The processor is configured to: determine at least one target point based on the information about the sensor, wherein a distance between the target point and the vehicle is within a first range; obtain at least one reference trajectory between the target and the at least one target point based on the at least one target point and a geographic location of the target; and determine a predicted trajectory of the target from the at least one reference trajectory based on a historical frequency or a historical time that the target has traveled to the at least one reference trajectory; The processor: determining a short-term trajectory of the target for the next s seconds based on a movement trend of the target; determining a reliability of the at least one reference trajectory based on the short-term trajectory, where the reliability of the reference trajectory indicates a degree of fit between the reference trajectory and the short-term trajectory; and determining the reference trajectory having the highest reliability as the predicted trajectory of the target; is specifically configured to: The processor is specifically configured to determine as the short-term trajectory a trajectory obtained after the target has moved for s seconds under a current movement trend; or The processor determines second parameters of the target based on the movement trend of the target, where the second parameters of the target include a historical movement speed and a historical movement direction of the target, a geographic location of the target, and environmental information of the target; and obtaining the short-term trajectory output by the neural network model of the target for the next s seconds after the second parameter is input into the neural network model; trajectory prediction device specifically configured to:

12. Processor and Transceiver Interface where the transceiver interface is configured to receive information about the sensor; the processor is configured to: determine at least one target point based on the information about the sensor, wherein a distance between the target point and the vehicle is within a first range; obtain at least one reference trajectory between the target and the at least one target point based on the at least one target point and a geographic position of the target; and determine a predicted trajectory of the target from the at least one reference trajectory; The processor: determining a short-term trajectory of the target for the next s seconds based on a movement trend of the target; determining a reliability of the at least one reference trajectory based on the short-term trajectory, where the reliability of the reference trajectory indicates a degree of fit between the reference trajectory and the short-term trajectory; and determining the reference trajectory having the highest reliability as the predicted trajectory of the target; is specifically configured to: The processor: specifically configured to determine a projection value of the short-term trajectory on the at least one reference trajectory according to the following formula, wherein the projection value is the confidence of the at least one reference trajectory; [Equation 41] and [Equation 42] is the reference orbit [Equation 43] above short-term orbit [Equation 44] is the projection value of [Equation 45] is the short-term orbit [Equation 46] and the reference trajectory [Equation 47] and the reference orbit is an included angle between [Number 48] is any one of the at least one reference trajectory.

13. The processor further comprises:

13. The trajectory prediction device of claim 8, configured to adjust a driving behavior of the vehicle after determining, based on the predicted trajectory, that the vehicle is about to collide with the target.

14. A computer-readable storage medium storing map data and a computer program, the computer program, when executed by a computer, causing the computer to: determining at least one target point, wherein a distance between the target point and the vehicle is within a first range; Obtaining at least one reference trajectory between a target and the at least one target point based on the at least one target point and a geographical position of the target; and determining, based on a historical frequency or a historical time at which the target has moved to at least one reference trajectory, a reference trajectory that the target has moved to with the highest historical frequency from the at least one reference trajectory, or a reference trajectory that the target has moved to with the closest historical time to the current time, as a predicted trajectory of the target; Execute Determining at least one target point includes: determining a first reference point by consulting map data on which reference points are marked, wherein a distance between the first reference point and the vehicle is within the first range; and If the number of first reference points is not greater than n, determining the first reference points as the target points, where n is a predetermined natural number; or if the number of first reference points is greater than n, selecting the at least one target point from the first reference points. Including, The map data is a first layer, the first layer includes at least one reference point, the reference point including at least one of a location such as a business industry location, an enterprise location, and a service industry location; A computer-readable storage medium.

15. The computer-readable storage medium of claim 14 , wherein the reference point is a location where a number of people not less than a second threshold passes during a first period of time.

16. At least one processor and memory A terminal device comprising: the memory configured to store program instructions; The processor is configured to call and execute the program instructions stored in the memory, so that the terminal device executes the trajectory prediction method according to any one of claims 1 to 7. Terminal device.

17. A computer program for causing a computer to execute the trajectory prediction method according to any one of claims 1 to 7.

Images