Non-motor vehicle motion trajectory prediction method, device and related equipment

Abstract

The present application discloses a method, device and related equipment for predicting the motion trajectory of a non-motor vehicle, and relates to the field of vehicle technology. It is applied to a vehicle-mounted terminal, and the method includes: obtaining first position information and driving path information associated with the vehicle, and obtaining driving state information and road information associated with the non-motor vehicle sensed by the vehicle, and the road information includes second position information; based on the first position information, the driving path information and the second position information, determining a target driving scene, and the target driving scene is used to indicate whether there is interaction between the vehicle and the non-motor vehicle; in a preset trajectory prediction model, determining a target trajectory prediction model corresponding to the target driving scene; inputting the driving state information and road information into the target trajectory prediction model, and predicting the first motion trajectory of the non-motor vehicle by the target trajectory prediction model.

Description

Technical Field

[0001] The present application belongs to the field of vehicle technology, and in particular, relates to a method, device and related equipment for predicting the motion trajectory of a non-motor vehicle. Background Art

[0002] With the rapid development of vehicle technology, in order to improve vehicle driving safety, it is necessary to identify obstacles that affect vehicle driving and predict their movement trajectories. However, the existing obstacle movement trajectory prediction is mainly aimed at motor vehicles and pedestrians. Non-motor vehicles are limited by their driving characteristics such as high degree of freedom, complex interactive relationships and unstable driving speed. It is not only difficult to predict their movement trajectories, but also cannot be applied to the existing motor vehicle trajectory prediction scheme. Summary of the invention

[0003] The embodiments of the present application provide a method, device and related equipment for predicting the motion trajectory of a non-motor vehicle, which can identify non-motor vehicles that affect vehicle driving and predict their motion trajectories, thereby improving the safety of vehicle driving.

[0004] In a first aspect, an embodiment of the present application provides a method for predicting a motion trajectory of a non-motor vehicle, which is applied to a vehicle-mounted terminal. The method includes:

[0005] Acquire first position information and driving path information associated with the vehicle, and acquire driving state information and road information associated with a non-motor vehicle sensed by the vehicle, wherein the driving state information includes a driving speed and a direction angle between a driving direction and a lane direction, and the road information includes second position information, wherein the first position information is used to indicate a first position of the vehicle, and the second position information is used to indicate a second position of the non-motor vehicle;

[0006] Determining a target driving scene based on the first position information, the driving path information, and the second position information, wherein the target driving scene is used to indicate whether there is an interaction between the vehicle and the non-motor vehicle;

[0007] Determining a target trajectory prediction model corresponding to the target driving scenario in a preset trajectory prediction model;

[0008] The driving state information and the road information are input into the target trajectory prediction model, and the first motion trajectory of the non-motor vehicle is predicted by the target trajectory prediction model.

[0009] In a second aspect, an embodiment of the present application provides a device for predicting a motion trajectory of a non-motor vehicle, which is applied to a vehicle-mounted terminal, and the device includes:

[0010] an acquisition module, configured to acquire first position information and driving path information associated with the vehicle, and to acquire driving state information and road information associated with a non-motor vehicle sensed by the vehicle, wherein the driving state information includes a driving speed and a direction angle between a driving direction and a lane direction, and the road information includes second position information, wherein the first position information is used to indicate a first position of the vehicle, and the second position information is used to indicate a second position of the non-motor vehicle;

[0011] A first determination module, configured to determine a target driving scene based on the first position information, the driving path information, and the second position information, wherein the target driving scene is used to indicate whether there is an interaction between the vehicle and the non-motor vehicle;

[0012] A second determination module is used to determine a target trajectory prediction model corresponding to the target driving scene in a preset trajectory prediction model;

[0013] The prediction module is used to input the driving state information and the road information into the target trajectory prediction model, and predict the first motion trajectory of the non-motor vehicle by the target trajectory prediction model.

[0014] In a third aspect, an embodiment of the present application provides an electronic device, comprising: a processor and a memory storing computer program instructions; when the processor executes the computer program instructions, the method for predicting the motion trajectory of a non-motor vehicle as described in any one of the above items is implemented.

[0015] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium having computer program instructions stored thereon, and when the computer program instructions are executed by a processor, a method for predicting the motion trajectory of a non-motor vehicle as described in any one of the above is implemented.

[0016] In a fifth aspect, an embodiment of the present application provides a vehicle, comprising: an electronic device, wherein the electronic device is used to implement the method for predicting the motion trajectory of a non-motor vehicle as described in any one of the above.

[0017] The prediction method, device and related equipment of the non-motor vehicle motion trajectory of the embodiment of the present application are applied to the vehicle-mounted terminal, and the target driving scene is determined based on the first position information, the driving path information and the second position information. The target driving scene is used to indicate whether there is interaction between the vehicle and the non-motor vehicle; in the preset trajectory prediction model, the target trajectory prediction model corresponding to the target driving scene is determined; the driving state information and the road information are input into the target trajectory prediction model, and the first motion trajectory of the non-motor vehicle is predicted by the target trajectory prediction model. In this way, the present application can identify the non-motor vehicle that affects the driving of the vehicle and predict its motion trajectory, thereby improving the safety of vehicle driving. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solution of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0019] Figure 1 It is a schematic diagram of the flow of a method for predicting a motion trajectory of a non-motor vehicle provided in an embodiment of the present application;

[0020] Figure 2 It is a schematic diagram of the framework of the method for predicting the motion trajectory of a non-motor vehicle provided in an embodiment of the present application;

[0021] Figure 3 is a schematic diagram of an interactive driving scenario provided in an embodiment of the present application;

[0022] Figure 4 It is a schematic diagram of the structure of a device for predicting the motion trajectory of a non-motor vehicle provided in an embodiment of the present application;

[0023] Figure 5 It is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0024] The features and exemplary embodiments of various aspects of the present application will be described in detail below. In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain the present application, rather than to limit the present application. For those skilled in the art, the present application can be implemented without the need for some of these specific details. The following description of the embodiments is only to provide a better understanding of the present application by illustrating the examples of the present application.

[0025] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the statement "include..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0026] With the rapid development of vehicle technology, in order to improve vehicle driving safety, it is necessary to identify obstacles that affect vehicle driving and predict their movement trajectories. However, the existing obstacle movement trajectory prediction is mainly aimed at motor vehicles and pedestrians. Non-motor vehicles are limited by their driving characteristics such as high degree of freedom, complex interactive relationships and unstable driving speed. It is not only difficult to predict their movement trajectories, but also cannot be applied to the existing motor vehicle trajectory prediction scheme.

[0027] In order to solve the problems of the prior art, the present application provides a method, device and related equipment for predicting the motion trajectory of a non-motor vehicle. The following first introduces the method for predicting the motion trajectory of a non-motor vehicle provided in the present application.

[0028] Figure 1 FIG. 1 is a flow chart showing a method for predicting a motion trajectory of a non-motor vehicle provided by an embodiment of the present application. Figure 1 As shown, a method for predicting a motion trajectory of a non-motor vehicle is applied to a vehicle-mounted terminal. The method may include the following steps S101 to S104.

[0029] S101, obtaining first position information and driving path information associated with the vehicle, and obtaining driving state information and road information associated with the non-motor vehicle sensed by the vehicle. The driving state information includes the driving speed and the angle between the driving direction and the lane direction, and the road information includes second position information, the first position information is used to indicate the first position of the vehicle, and the second position information is used to indicate the second position of the non-motor vehicle.

[0030] S102: Determine a target driving scene based on the first position information, the driving path information, and the second position information. The target driving scene is used to indicate whether there is interaction between the vehicle and the non-motor vehicle.

[0031] S103: Determine a target trajectory prediction model corresponding to the target driving scenario in a preset trajectory prediction model.

[0032] S104: Input the driving state information and the road information into a target trajectory prediction model, and use the target trajectory prediction model to predict a first motion trajectory of the non-motor vehicle.

[0033] The method for predicting the motion trajectory of a non-motor vehicle in an embodiment of the present application is applied to a vehicle-mounted terminal, and based on the first position information, the driving path information, and the second position information, a target driving scene is determined, and the target driving scene is used to indicate whether there is interaction between the vehicle and the non-motor vehicle; in a preset trajectory prediction model, a target trajectory prediction model corresponding to the target driving scene is determined; the driving state information and road information are input into the target trajectory prediction model, and the first motion trajectory of the non-motor vehicle is predicted by the target trajectory prediction model. In this way, the present application can identify non-motor vehicles that affect vehicle driving and predict their motion trajectories, thereby improving the safety of vehicle driving.

[0034] The specific implementation methods of the above steps are introduced below.

[0035] In S101, the first position information may be used to indicate the first position of the vehicle.

[0036] The driving state information may include driving speed and an angle between the driving direction and the lane direction.

[0037] The above-mentioned road information may include second position information, and the second position information may be used to indicate the second position of the non-motor vehicle.

[0038] The above-mentioned vehicle can be a motor vehicle, specifically an autonomous driving motor vehicle.

[0039] The above-mentioned acquisition of the first position information associated with the vehicle may be the identification of the driving position information of the vehicle through the position positioning of the vehicle.

[0040] The above-mentioned acquisition of the driving path information associated with the vehicle may be acquisition of the preset driving path information of the autonomous driving vehicle, or acquisition of the departure point, destination and real-time traffic conditions information of the motor vehicle to determine the driving path information of the motor vehicle.

[0041] The above-mentioned acquisition of the driving status information and road information associated with the non-motor vehicle sensed by the vehicle may be achieved by setting a sensing module (such as computer video imaging technology) in the vehicle to identify the driving status information and road information of the non-motor vehicle related to the vehicle through the sensing module.

[0042] In S102, the target driving scene may be used to indicate whether there is interaction between the vehicle and the non-motor vehicle.

[0043] In some embodiments, the above S102 may include:

[0044] Calculating a lateral distance between the vehicle and the non-motor vehicle based on the first position information and the second position information;

[0045] When the lateral distance is less than or equal to the distance threshold, the driving path information indicates that the vehicle maintains the lane where the first position is located, and the direction angle is less than the first threshold, it is determined that the target driving scene is a road driving scene.

[0046] In this embodiment, when the lateral distance between the vehicle and the non-motor vehicle is less than or equal to the distance threshold, the driving path information indicates that the vehicle maintains the lane where the first position is located, and the direction angle is less than the first threshold, the target driving scene is determined to be a road driving scene, which means that the vehicle and the non-motor vehicle are driving in parallel at a close distance. The movement trajectory of the non-motor vehicle will affect the driving safety of the vehicle. Therefore, it is also necessary to predict the movement trajectory of the non-motor vehicle in the road driving scene.

[0047] In S103, the preset trajectory prediction model includes trajectory prediction models corresponding to multiple driving scenarios. Different driving scenarios correspond to different trajectory prediction models. Exemplarily, the trajectory prediction model corresponding to the road driving scenario is a road predictor; the trajectory prediction model corresponding to the interactive driving scenario is an interactive predictor; the trajectory prediction model corresponding to the intersection driving scenario is an intersection predictor; and the trajectory prediction model corresponding to the free driving scenario is an extended Kalman filter predictor.

[0048] In S104, in some embodiments, the road information may further include road traffic light information, and the road traffic light information may include a signal of a traffic light on the road where the non-motor vehicle is located, and a first distance between the non-motor vehicle and the traffic light;

[0049] The above S104 may specifically include:

[0050] When the road traffic light information indicates that the traffic light is a stop signal, calculating the acceleration of the non-motor vehicle based on the first distance and the travel speed;

[0051] When the acceleration is a negative value, a first motion trajectory of the non-motor vehicle performing uniform deceleration motion with acceleration is predicted;

[0052] When the road traffic light information indicates that the traffic light is not a stop signal, or the acceleration is not a negative value, a first motion trajectory of the non-motor vehicle moving at a uniform speed at a driving speed is determined.

[0053] In this embodiment, the first motion trajectory of the non-motor vehicle is predicted to be a uniform motion or a uniform deceleration motion through road information, which helps the vehicle avoid the non-motor vehicle in the driving path, thereby improving driving safety.

[0054] As an implementation of the present application, in order to identify a driving scene in which the motion trajectory of other non-motor vehicles may affect the driving safety of the vehicle, the above S102 may also include:

[0055] When the lateral distance is greater than the distance threshold and the direction angle is greater than or equal to the first threshold, the minimum spatial distance between the vehicle and the non-motor vehicle within the preset time period is calculated based on the driving path information, the second position information, the driving speed and the direction angle;

[0056] When the minimum spatial distance is less than the second threshold, calculating the first time for the vehicle to arrive at the location corresponding to the minimum spatial distance according to the driving path information, and calculating the second time for the non-motor vehicle to arrive at the location corresponding to the minimum spatial distance according to the driving speed;

[0057] When the first time is less than or equal to the second time, the target driving scene is determined to be an interactive driving scene.

[0058] Based on the driving path information, the second position information, the driving speed and the direction angle, the minimum spatial distance between the vehicle and the non-motor vehicle in the preset time period is calculated. For example, the future trajectory y of the non-motor vehicle in 4s can be obtained with a resolution of 0.1s based on the second position information, the driving speed and the direction angle. c , the vehicle's 4s future trajectory y is obtained with a resolution of 0.1s based on the driving path information v ; Traverse the future trajectory points of the target non-motor vehicle for 40 steps 40 future trajectory points with the vehicle The spatial distance between them, and the minimum spatial distance ds is obtained:

[0059]

[0060] The second threshold may be a predefined minimum interaction distance threshold dε, which may be, for example, 2.5 m in length of a motor vehicle. Of course, the present embodiment is not limited thereto and is not specifically limited thereto.

[0061] The above-mentioned calculation of the first time when the vehicle arrives at the location corresponding to the minimum spatial distance based on the driving path information may be to determine the time when the vehicle arrives at the location corresponding to the minimum spatial distance in the driving path information as the first time.

[0062] The second time for the non-motor vehicle to reach the location corresponding to the minimum spatial distance is calculated based on the driving speed. The second time may be obtained by dividing the distance the non-motor vehicle reaches the location corresponding to the minimum spatial distance by the driving speed.

[0063] In this embodiment, when the first time is less than or equal to the second time, it means that when the non-motor vehicle intersects the vehicle, the vehicle is driving in front and the non-motor vehicle is driving behind, and the non-motor vehicle will collide with the vehicle. At this time, the target driving scene is determined to be an interactive driving scene, thereby identifying driving scenes in which the movement trajectory of other non-motor vehicles may affect the driving safety of the vehicle.

[0064] In some embodiments, the above S104 may further include:

[0065] Based on the driving path information, the second position information, the driving speed and the direction angle, calculating the interaction time for the non-motor vehicle to arrive at the interaction location, the interaction location being the location when the spatial distance between the vehicle and the non-motor vehicle is equal to the second threshold;

[0066] Determine a parking location for the non-motor vehicle based on the interaction time and the second location information, the driving speed and the direction angle;

[0067] Based on the parking location, a first motion trajectory of the non-motor vehicle performing uniform deceleration motion in a preset time period is predicted.

[0068] The above-mentioned calculation of the interaction time for the non-motor vehicle to arrive at the interaction location based on the driving path information, the second position information, the driving speed and the direction angle can be based on the second position information, the driving speed and the direction angle of the non-motor vehicle and the driving path information of the vehicle to determine the interaction location when the spatial distance between the vehicle and the non-motor vehicle is equal to the second threshold; and calculating the interaction time for the non-motor vehicle to arrive at the interaction location at a uniform driving speed.

[0069] The above-mentioned method of determining the parking location of the non-motor vehicle based on the interaction time and the second position information, the driving speed and the direction angle may be that the non-motor vehicle maintains the direction angle unchanged at the position indicated by the second position information, and performs a uniform deceleration motion with the initial speed as the driving speed, the ending speed as 0, and the corresponding duration of the interaction time, and the location where the uniform deceleration motion stops is determined as the parking location.

[0070] In this embodiment, when the target driving scenario is an interactive driving scenario, the parking location where the non-motor vehicle is forced to stop when the interaction occurs is determined through the interaction time point and the driving speed, and the first motion trajectory of the non-motor vehicle performing uniform deceleration motion in a preset time period can be accurately predicted, thereby helping the vehicle to avoid non-motor vehicles in the driving path and improving driving safety.

[0071] As another implementation of the present application, in order to identify driving scenarios where the movement trajectories of other non-motor vehicles may affect the driving safety of the vehicle, the road information further includes lane exit information; the lane exit information includes at least one lane exit of the target intersection, and lane information of each lane exit;

[0072] The above S102 may further include:

[0073] When the first time is greater than the second time, the loss value of each lane exit is calculated based on the lane information and the second position information, the driving speed, and the direction angle;

[0074] When the optimal exit is found at the target intersection, the target driving scene is determined to be the intersection driving scene, and the optimal exit is the lane exit corresponding to the minimum loss value among the loss values ​​of the lane exits of the target intersection;

[0075] When the optimal exit is not found in the target intersection, the target driving scenario is determined to be a free driving scenario.

[0076] When the first time is greater than the second time, it means that when the non-motor vehicle intersects the vehicle, the non-motor vehicle is driving in front and the vehicle is driving behind, and the non-motor vehicle is in the process of changing lanes. Therefore, it is necessary to find the optimal exit for the non-motor vehicle at the target intersection.

[0077] Based on the lane information and the second position information, the driving speed and the direction angle, the loss value of each lane exit is calculated. Exemplarily, it can be based on the second position information, the driving speed and the direction angle, uniformly estimate the future 8s trajectory segment of the non-motor vehicle, traverse each lane exit in the target intersection, obtain the center point, left and right boundary position points, exit road width and direction of each lane exit, and obtain the straight line expression of each lane intersection through the center point and the boundary position point; for each lane exit, the intersection point of the straight line expression and the future 8s trajectory segment of the non-motor vehicle is obtained. If there is no intersection, the lane exit is filtered; if there is an intersection, and the intersection is within 12 road widths from the center point of the lane exit, it is used as a candidate exit; the loss value of each candidate exit is calculated:

[0078] cost=dist 1 +dist 2 +Δθ+cost type Formula (2)

[0079] Among them, dist 1 is the distance from the intersection to the center of the lane exit, dist 2 is the distance from the non-motor vehicle to the center point of the lane exit, Δθ is the angle difference between the non-motor vehicle speed direction angle and the exit road orientation angle, cost type is the road type penalty item, for example, 0 for non-motorized lanes and 10 for motorized lanes.

[0080] In this embodiment, by calculating the loss value of each lane exit and searching whether there is an optimal exit in at least one lane exit of the target intersection, the driving scenarios in which the movement trajectories of other non-motor vehicles may affect the driving safety of the vehicle can be accurately identified.

[0081] In some embodiments, when the target driving scene is an intersection driving scene, the above S104 may further include:

[0082] Based on the second position information and the driving speed, a first motion trajectory of the non-motor vehicle driving towards the optimal exit within a preset time period and moving at a uniform speed along the road corresponding to the optimal exit is predicted.

[0083] Based on the second position information and the driving speed, the first motion trajectory of the non-motor vehicle driving towards the optimal exit within a preset time period and moving at a uniform speed along the road corresponding to the optimal exit is predicted. For example, the coordinates of 25 position points searched from the exit point to the inside on the optimal exit road can be obtained in combination with the map information in the Frenet coordinate system with a vertical coordinate step of 0.2m, and inserted into the candidate exit point queue; for a shorter lane, continue to search for position points on its subsequent lane; based on the current position information and speed information of the non-motor vehicle, generate a set of future 4s trajectory points through a uniform speed model with a resolution of 0.1s; obtain the candidate exit The exit point in the point queue where the straight-line distance to the future trajectory of the non-motor vehicle reaches the minimum is taken as the optimal entrance and exit point for the non-motor vehicle; assuming that the non-motor vehicle leaves the intersection at a constant speed, the current position and speed of the non-motor vehicle are obtained as the starting point, the position of the optimal entrance and exit point and the vector speed of the non-motor vehicle arriving at this point are taken as the end point, and a cubic curve fitting is performed; for the reverse intersection, the center point of the exit is taken as the optimal exit point; the candidate exit time is constructed with a time step of 0.2s and inserted into the candidate exit time set, the planned arrival speed is obtained by the distance-to-time ratio, and the candidate time whose arrival speed is not within the reasonable range of the current perceived speed is filtered out; the loss function is constructed:

[0084] cost = curvature * v 2 +time formula (3)

[0085] Where curvature is the curvature of the cubic curve, v is the current perceived speed of the non-motor vehicle, and time is the candidate time; the optimal time to leave the intersection is obtained by minimizing the curvature loss function in combination with the cubic curve; the predicted trajectory before reaching the optimal entrance and exit points is obtained by fitting the cubic curve; the predicted trajectory after reaching the optimal entrance and exit points is generated along the center line of the road according to the current driving speed. The values ​​in this example are not limited to this, and can also be other values, which are not specifically limited here.

[0086] In this embodiment, when the target driving scene is an intersection driving scene, the first motion trajectory of the non-motor vehicle traveling towards the optimal exit within a preset time period and moving at a uniform speed along the road corresponding to the optimal exit can be predicted based on the second position information and the driving speed, thereby helping the vehicle to avoid non-motor vehicles in the driving path and improving driving safety.

[0087] In some embodiments, the driving state information further includes a position abscissa value, a position ordinate value, a speed magnitude value, a yaw angle value, and an angular velocity value;

[0088] When the target driving scene is a free driving scene, the above S104 may further include:

[0089] Inputting the position abscissa value, position ordinate value, speed magnitude value, yaw angle value and angular velocity value into the target trajectory prediction model, determining the position abscissa value, position ordinate value, speed magnitude value, yaw angle value and angular velocity value of the non-motor vehicle at each time point within a preset time period, the target trajectory prediction model including the relationship between the position abscissa value, position ordinate value, speed magnitude value, yaw angle value and angular velocity value at different time points;

[0090] Based on the position abscissa value, position ordinate value, speed magnitude value, yaw angle value and angular velocity value at each time point in the preset time period, a first motion trajectory of the non-motor vehicle in the preset time period is determined.

[0091] The above-mentioned position abscissa value, position ordinate value, speed magnitude value, yaw angle value and angular velocity value can be expressed as state quantity The five state variables are the horizontal coordinate x, the vertical coordinate y, the velocity v, the yaw angle θ, and the angular velocity ω. The yaw angle θ has a value range of [-π,π].

[0092] The above target trajectory prediction model can be an extended Kalman filter trajectory EKF filter Function. Among them, EKF filter The function includes the extended Kalman filter prediction (EKF prediction ) and the modified (EKF correction ) the whole process.

[0093] EKF correction Correction:

[0094] K k =P k H(HP k H T +R) -1 Formula (4)

[0095] x k =x k +K k (z k -h(x k )) Formula (5)

[0096] P K =(IK k H)P k Formula (6)

[0097] EKF prediction predict:

[0098] xk+1 =g(x k ,u) Formula (7)

[0099]

[0100] Among them, u is the above process control quantity, g(x k ,u) is the above state transfer function, P k is the state covariance matrix at time k, K k is the Kalman gain at time k, R is the measurement covariance matrix, and I is the identity matrix.

[0101] The covariance matrix is ​​initialized as:

[0102]

[0103] Among them, r1 to r5 can be certain values, which are not specifically limited here.

[0104] Assume that the current time point is k=0, and the state quantity is initialized to the motion state of the target non-motor vehicle 1 second ago (i.e., with a resolution of 0.1 seconds and 10 historical steps ago):

[0105]

[0106] All five status variables are directly obtained from the driving status information.

[0107] Measurement value: The measurement value z of the radar sensor is consistent with the state quantity, where the yaw angle θ is derived from the measurement speed, and the change in the yaw angle per unit time is used as the measurement angular velocity ω. Therefore, the measurement matrix is:

[0108]

[0109] Assume that the duration is t k is related to the discrete time step k, then at time point t k+1 The state quantity The time point t k Status It is deduced that:

[0110]

[0111] After integration, we get the CTRV state transfer function:

[0112]

[0113]

[0114] Solve the Jacobian matrix for the state transfer equation:

[0115]

[0116] Process noise and noise covariance matrix: The noise in the CTRV model is mainly composed of linear acceleration noise and yaw acceleration noise. Assume that the linear acceleration μ a and yaw angular acceleration μ ω The mean is 0 and the variances are and Gaussian distribution of , we can get the effect of these two acceleration noises on the state:

[0117]

[0118] And the covariance matrix of the noise:

[0119] Q=G·E[μμ T ]·G T Formula (16)

[0120] in, The standard deviation of linear acceleration and yaw acceleration in process noise are initialized as σ a =a and σ ω =brad / s, where a and b are certain values.

[0121] The above-mentioned position abscissa value, position ordinate value, speed magnitude value, yaw angle value and angular velocity value are input into the target trajectory prediction model to determine the position abscissa value, position ordinate value, speed magnitude value, yaw angle value and angular velocity value of the non-motor vehicle at each time point in the preset time period. For example, after the extended Kalman filter prediction correction process for the past 10 steps, we obtain a relatively stable state quantity and covariance matrix, and record this state quantity as the prediction initial state quantity at time k=0 as In the next 40 time steps, the extended Kalman filter prediction (EKF prediction ) process gradually generates 40 future prediction state quantities.

[0122] The above-mentioned method of determining the first motion trajectory of the non-motor vehicle within the preset time period based on the position abscissa value, position ordinate value, speed magnitude value, yaw angle value and angular velocity value at each time point within the preset time period may be to use the extended Kalman filter prediction (EKF prediction ) process generates 40 future predicted state quantities, which are determined as the first motion trajectory of the non-motor vehicle.

[0123] In this embodiment, when the target driving scene is a free driving scene, the first motion trajectory of the non-motor vehicle within the preset time period is determined based on the horizontal coordinate value of the position, the vertical coordinate value of the position, the speed magnitude value, the yaw angle value and the angular velocity value at each time point within the preset time period, thereby helping the vehicle to avoid non-motor vehicles in the driving path and improving driving safety.

[0124] As another implementation of the present application, in order to identify non-motor vehicles that interfere with vehicle driving, the road information also includes lane boundary information;

[0125] After the above S104, the following may also be included:

[0126] When the lane boundary information indicates that the first motion trajectory is blocked by the dividing facility, a portion of the first motion trajectory after the intersection with the dividing facility is cut off to obtain a second motion trajectory of the non-motor vehicle.

[0127] The lane boundary information may be lane boundary information in a high-precision map cached in the vehicle terminal, and is used to indicate whether the intersection of the first motion trajectory and the lane boundary is blocked by a separation facility. The separation facility may be a separation belt or a green belt.

[0128] In this embodiment, when the first motion trajectory is blocked by a separation facility at the intersection with the lane boundary, the portion of the first motion trajectory after the intersection with the separation facility is cut off to obtain a second motion trajectory of the non-motor vehicle. The vehicle can know from the second motion trajectory that the non-motor vehicle cannot interfere with its driving, thereby keeping the driving path information unchanged.

[0129] In order to facilitate the understanding of the non-motor vehicle motion trajectory prediction method in the embodiment of the present application, the actual application process of the non-motor vehicle motion trajectory prediction method is described as follows:

[0130] like Figure 2 As shown in FIG. 1 , the technical ideas of the prediction method of the motion trajectory of non-motor vehicles are mainly as follows:

[0131] Step 1: Determine whether the target non-motor vehicle is traveling in the lane. First, determine whether the non-motor vehicle is in the non-motor vehicle lane or the motor vehicle lane based on its current position; if it is in a similar lane, calculate whether the angle between the current speed heading angle of the target non-motor vehicle and the direction of the lane meets the condition of less than 30°.

[0132] Step 2: If the target non-motor vehicle is traveling in a similar lane and the angle is less than 30°, it enters the road evaluator and road predictor.

[0133] Step 3: If the target non-motor vehicle is not traveling in a similar lane and the angle is greater than or equal to 30°, the interaction distance between the target non-motor vehicle and the vehicle is determined to be less than or equal to the threshold to determine whether it is an interaction state. The non-motor vehicle that is determined to be in an interaction state is labeled with an interaction label.

[0134] Step 4: The target non-motor vehicle with interaction labels enters the interaction evaluator and interaction predictor.

[0135] Step 5: For the target non-motor vehicle without an interactive label, determine whether it is currently at a traffic intersection based on its location point.

[0136] Step 6: If the target non-motor vehicle is at a traffic intersection, enter the intersection evaluator.

[0137] Step 7: If the intersection evaluator finds the optimal exit, it enters the intersection predictor.

[0138] Step 8: If the target non-motor vehicle is not at the traffic intersection, or is at the traffic intersection but the intersection evaluator has not found the optimal exit, enter the EKF predictor.

[0139] The above-mentioned non-motor vehicle motion trajectory prediction method also includes non-motor vehicle trajectory interception. The interception processing realizes boundary acquisition, target trajectory truncation, trajectory point completion and other functions, which requires relatively complex geometric relationship calculations. After the predictor generates path points, the trajectory truncation points of each target are comprehensively calculated using boundary information, target geometric information, etc. In the process of generating the trajectory, the trajectory points are depicted point by point. If the trajectory is truncate, the truncation trajectory point information is recorded and the trajectory is completed.

[0140] Based on the above technical ideas, the embodiment of the present application provides a method for predicting the motion trajectory of a non-motor vehicle, which may specifically include the following steps:

[0141] Step 1: For all non-motor vehicles currently sensed, the vehicle terminal will first filter out non-motor vehicles whose predicted trajectories are not considered, which can effectively save computing power and avoid interference with vehicle control caused by too many and too complex predicted trajectories.

[0142] The non-motor vehicle filtering rules may specifically include the following steps:

[0143] Step 1.1: Define the IsCycOnCycLane Boolean state. If the non-motor vehicle is currently on the non-motor vehicle lane (i.e., not in the same lane as the vehicle), and the angle between the driving direction and the lane direction is less than 30°, then set the state IsCycOnCycLane to True;

[0144] Step 1.2: Define the IsCycFarFromEgo Boolean state. For non-motor vehicles, traverse the predicted trajectory points of the non-motor vehicle in the next 8 seconds and the planned trajectory points of the vehicle with a resolution of 1 second. If the minimum spatial distance between the two points is greater than a threshold, set the state IsCycFarFromEgo to true. The threshold depends on the current driving speed of the vehicle at 1,000 km / h and the limited distance of 15 meters, whichever is greater;

[0145] Step 1.3: Define the IsCycFarStill state variable. For a stationary non-motor vehicle 5 meters away from the vehicle, set the state variable IsCycFarStill to true.

[0146] Non-motor vehicles that meet any of the above-mentioned status quantities are filtered.

[0147] Step 2: For non-motor vehicles that have not been filtered, they will enter the non-motor vehicle prediction framework process. Specifically, the following steps or methods are included:

[0148] 2.1 Road driving scenarios

[0149] Step 2.1.1: The road evaluator gives the probability of driving on the road.

[0150] Under the condition that the non-motor vehicle currently has the characteristics of a lane, determine whether the first road sequence id of each possible path is consistent with the road sequence id where the non-motor vehicle is currently located. If they are consistent, the probability of the lane sequence is set to 1, otherwise it is set to 0.

[0151] Step 2.1.2: The road predictor generates a predicted trajectory for the filtered candidate lane sequence.

[0152] First, a lane sequence filter function is used to filter out lane sequences with road type PARKING and lane_change_type other than LEFT, RIGHT, or ONTO_LANE. Then, the distance between the obstacle and the vehicle after the lane change is calculated to filter out lane sequences with a distance from the vehicle less than a threshold. Next, a trajectory is generated based on whether there is a stop light in the lane sequence. If there is a stop light, the acceleration is calculated based on the obstacle’s current speed and the distance from the stop light to determine whether the obstacle has an intention to stop. If there is an intention to stop, a uniform deceleration model is used to generate a trajectory. If there is no stop light or no intention to stop, a trajectory is generated along the road at the current vehicle speed.

[0153] 2.2 Interactive Driving Scenario

[0154] Step 2.2.1: Search for all current moving motor vehicle obstacles (including the vehicle itself).

[0155] Step 2.2.2: Determine the interaction relationship between each mobile obstacle and non-motor vehicle; if there is interaction, determine the influencer and responder between the two, and attach an interaction label to the non-motor vehicle.

[0156] Obtain the 4s future trajectory y of the non-motor vehicle with a resolution of 0.1s c 4s future trajectory with obstacles v ; If the obstacle is the vehicle and has a planned trajectory, the planned trajectory is used as the future trajectory y v ;

[0157] like Figure 3 As shown, the future trajectory points of the non-motor vehicle with a length of 40 steps are traversed. 40 future trajectory points with obstacles The spatial distance between them, and the minimum spatial distance ds is obtained:

[0158]

[0159] Define a minimum interaction distance threshold dε, which is adjusted to 2.5m according to the length of the motor vehicle; if the minimum spatial distance ds is less than the distance threshold dε, then obtain the time t when the obstacle and the non-motor vehicle reach the minimum spatial distance point ds respectively 1 and t 2 If t 1 >t 2 , then the non-motor vehicle reaches the minimum spatial distance point first and becomes the influencing factor in the interaction relationship, and no processing is done; if t 1 <t 2 , then the non-motor vehicle is the responder of the interaction relationship, and the time tε taken to reach the distance threshold dε is calculated and stored in the sequence as the interaction time point, and an interaction label is attached to the target non-motor vehicle.

[0160] Step 2.2.3: After the traversal is completed, the minimum value of the interaction time series is obtained as the final interaction time point of the target non-motor vehicle.

[0161] Step 2.2.4: The target non-motor vehicle with an interaction label enters the interaction evaluator, which obtains the interaction time point tε obtained by the non-motor vehicle in the above process, and obtains the position point where the target non-motor vehicle is forced to stop when the interaction occurs through the interaction time point and the current driving speed.

[0162] Step 2.2.5: The interactive predictor generates a 4-second predicted trajectory based on the location of the target non-motor vehicle forced to stop using a uniform deceleration model.

[0163] 2.3 Driving at intersections

[0164] Step 2.3.1: Search all non-motorized vehicle lane entrances and exits and motor vehicle lane exits in the intersection, and insert them into the non-motorized vehicle lane exit queue under the intersection, for example, there are 8 exits.

[0165] Step 2.3.2: Non-motor vehicles entering the intersection enter the intersection evaluator.

[0166] The intersection evaluator obtains the target non-motor vehicle's future 8s trajectory segment through the uniform speed model; searches for the road exit where the target non-motor vehicle enters the intersection, and removes its road exit from the exit queue; traverses all non-motor vehicle exits, obtains the center point, left and right boundary position points, exit road width and direction of the exit point, and obtains the straight line expression of the intersection through the center point and boundary position points; for each exit, finds the intersection point between the straight line where it is located and the target non-motor vehicle's future 8s trajectory segment, if there is no intersection, filter the exit; if there is an intersection, and the intersection point is within 12 road widths from the center point of the exit, then it is used as a candidate exit to calculate the loss: cost = dist 1 +dist 2 +Δθ+cost type Among them, dist 1 is the distance from the intersection point to the center point of the exit, dist 2 is the distance from the target non-motor vehicle to the center point of the exit, Δθ is the angle difference between the current speed angle of the target non-motor vehicle and the exit road angle, cost type is the road type penalty item, which is 0 for non-motorized lanes and 10 for motorized lanes; the exit with the smallest loss is taken as the optimal exit, and the evaluator result is returned as true; if the optimal exit is not found, the evaluator result is returned as false, and the Kalman trajectory prediction EKF predictor is entered.

[0167] Step 2.3.3: The target non-motor vehicle that successfully finds the optimal exit through the intersection evaluator enters the intersection predictor.

[0168] The current perception information of the target non-motor vehicle (including speed, position, and heading angle) and the optimal exit given by the evaluator are obtained; combined with the map information, the coordinates of the 25 position points searched from the exit point on the optimal exit road are obtained in the Frenet coordinate system with a vertical coordinate step of 0.2m, and the coordinates are inserted into the candidate exit point queue; for shorter lanes, the position points on the subsequent lanes are continued to be searched; according to the current position information and speed information of the target motor vehicle, a set of future 4s trajectory points is generated through a uniform speed model with a resolution of 0.1s; the exit point in the candidate exit point queue with the minimum straight-line distance to the future trajectory of the target non-motor vehicle is obtained. The exit point is taken as the optimal exit and entry point of the target non-motor vehicle; assuming that the non-motor vehicle leaves the intersection at a constant speed, the current position and speed of the target non-motor vehicle are obtained as the starting point, the position of the optimal exit and entry point and the vector speed of the non-motor vehicle arriving at this point are taken as the end point, and a cubic curve fitting is performed; for the reverse intersection, the exit center point is taken as the optimal exit point; the candidate exit time is constructed with a time step of 0.2s and inserted into the candidate exit time set, the planned arrival speed is obtained by the distance-to-time ratio, and the candidate time whose arrival speed is not within the reasonable range of the current perceived speed is filtered out; the loss function is constructed: cost = curvature*v 2 +time, where curvature is the curvature of the cubic curve, v is the current perceived speed of the non-motor vehicle, and time is the candidate time; combined with the cubic curve, the optimal time to leave the intersection is obtained by minimizing the curvature loss function; the predicted trajectory before reaching the optimal entrance and exit points is obtained by fitting the cubic curve; the predicted trajectory after reaching the optimal entrance and exit points is generated along the center line of the road according to the current driving speed.

[0169] 2.4 Extended Kalman Filter Trajectory Prediction EKF Predictor

[0170] Extend the Constant Turn Rate and Velocity (CTRV) model in Kalman filtering. The CTRV model assumes that the object moves at a fixed turning rate and a constant velocity.

[0171] State quantity and transfer matrix: target state quantity The five state variables are the horizontal coordinate x, the vertical coordinate y, the velocity v, the yaw angle θ, and the angular velocity ω. The yaw angle θ has a value range of [-π,π].

[0172] Assume that the duration is t k is related to the discrete time step k, then at time point t k+1 The state quantity The time point t k Status It is deduced that:

[0173]

[0174] After integration, we get the CTRV state transfer function:

[0175]

[0176]

[0177] Solve the Jacobian matrix for the state transfer equation:

[0178]

[0179] Process noise and noise covariance matrix: The noise in the CTRV model is mainly composed of linear acceleration noise and yaw acceleration noise.

[0180] Assuming the linear acceleration μ a and yaw angular acceleration μ ω The mean is 0 and the variances are and Gaussian distribution of , we can get the effect of these two acceleration noises on the state:

[0181]

[0182] And the covariance matrix of the noise:

[0183] Q=G·E[μμ T ]·G T

[0184] in,

[0185] Measurement value: The measurement value z of the radar sensor is consistent with the state quantity, where the yaw angle θ is derived from the measurement speed, and the change in the yaw angle per unit time is used as the measurement angular velocity ω. Therefore, the measurement matrix is:

[0186]

[0187] Extended Kalman filter process:

[0188]

[0189] Correction:

[0190] K k =P k H(HP k H T +R) -1

[0191] x k =x k +K k(z k -h(x k ))

[0192] P K =(IK k H)P k

[0193] predict:

[0194] x k+1 =g(x k ,u)

[0195]

[0196] Among them, u is the above process control quantity, g(x k ,u) is the above state transfer function, P k is the state covariance matrix at time k, K k is the Kalman gain at time k, R is the measurement covariance matrix, and I is the identity matrix.

[0197] Step 2.4.1: Initialization

[0198] The covariance matrix is ​​initialized as;

[0199]

[0200] Among them, r1 to r5 are certain values.

[0201] Assume that the current time point is k=0, and the state quantity is initialized to the motion state of the target non-motor vehicle 1 second ago (i.e., with a resolution of 0.1 seconds and 10 historical steps ago): All five state quantities are obtained directly from sensory information;

[0202] The standard deviation of linear acceleration and yaw acceleration in process noise are initialized as σ a =am / s 2 and σ ω =brad / s, where a and b are certain values.

[0203] Step 2.4.2: Extend the Kalman filter prediction correction process for the historical 10 frames of data

[0204] The above EKF filter The function includes the extended Kalman filter prediction (EKF prediction ) and the modified (EKF correction ) the whole process.

[0205] For the first 10 frames of historical data, the Jacobian matrix J at time k AAnd the predicted state The state quantity at time k-1 It is obtained through the state transfer function. At the same time, according to the formula Predict the covariance matrix at time k, where the covariance matrix of the noise is given by the formula: Q = G·E[μμ T ]·G T Calculated.

[0206] Get the state quantity at time k and the covariance matrix P k After that, the correction process of the extended Kalman filter is entered. The Kalman gain is calculated according to the formula K k =P k H(HP k H T +R) -1 Calculate the state quantity at time k According to the measured value z at time k k and Kalman gain, and P K The measured value z in this process is corrected. k The yaw angle θ in is calculated by the perceived velocity vector, and the angular velocity ω is calculated by the yaw angle ratio time step of the two frames before and after the perceived data.

[0207] Step 2.4.3: Extend the Kalman filter prediction process

[0208] After the extended Kalman filter prediction correction process for the past 10 steps, we obtain a relatively stable state quantity and covariance matrix, which is recorded as the initial state quantity of the prediction at time k = 0 as

[0209] In the next 40 time steps, the extended Kalman filter prediction (EKF prediction ) process gradually generates 40 future prediction state quantities.

[0210] Step 3: Track Interception

[0211] Target trajectory interception includes: boundary maintenance and trajectory processing. Boundary maintenance involves the storage, acquisition, update and cleaning of high-precision map boundaries. Combined with the current system module structure, it is necessary to set up a global cache and update the boundary information when receiving the perception message; trajectory processing is implemented in the prediction stage, and the global boundary information is obtained to intercept unreasonable trajectories.

[0212] Step 3.1: Map boundary thinning

[0213] Theoretically, the straighter the lane, the fewer demarcation points there are. Of course, the number of demarcation points is also related to the demarcation length. Multiple consecutive points on a line segment do not need to be retained, but points at broken lines must be retained, so the following sparse processing rules are set to process demarcation points: set a maximum average distance threshold, such as 10m, and a minimum average distance, such as 5m - calculate the average distance between two points. If the average distance is greater than the maximum threshold, all points are retained; if the average distance is less than the minimum threshold, the average distance is the minimum threshold - traverse point by point, and if the distance between the two points is greater than the average distance, they are retained. Iterate in sequence - both the first and last points are retained - the maximum and minimum average distances can be adjusted according to the demarcation length, or can be used as parameters.

[0214] Step 3.2 Trajectory truncation

[0215] Trajectory truncation is to calculate whether the line segment of the trajectory point and the line segment of the demarcation point have an intersection. Calculation process - take the demarcation and trajectory points one by one - the demarcation points and trajectory points form a line segment in pairs, calculate the intersection of the line segments - construct the line segment through ax+by+c=0, and use the formula to calculate whether the line segments intersect and the intersection position.

[0216] Step 3.3 Consider Box in the trajectory stage

[0217] The trajectory extends outward from the center point of the bbox. According to the cutoff angle and the length and width of the vehicle, the following calculation is performed: the offset 1 is calculated based on half the vehicle width and the intersection angle - half the vehicle length is used as offset 2 - the longitudinal direction s of the target trajectory - the offset is the actual trajectory cutoff distance.

[0218] The interactive prediction method of the present invention can reasonably predict whether the non-motor vehicle will directly pass or avoid the interaction when the non-motor vehicle interacts with the motor vehicle. When avoiding, it finds a reasonable and safe deceleration stop point and gives a uniform deceleration prediction trajectory. The interactive prediction method greatly alleviates the interference of the predicted trajectory of non-motor vehicles crossing or turning at intersections on the self-vehicle control.

[0219] The intersection prediction method of the present invention has a stable intention prediction accuracy rate of more than 80% at the intersection. Under the accurate intention prediction, the trajectory prediction quantitative index is improved to a certain extent.

[0220] The Kalman filter prediction method of the present invention, in a test with 4 bicycles, a total of 53 obstacles, 23 effective obstacles, and a total of 624 evaluation paths, the final displacement error (FDE) and average displacement error (ADE) of the 3s trajectory prediction of non-motor vehicle type obstacles in a free driving state are 3.5 meters and 2.2 meters respectively. The predicted trajectory still has a good stability while ensuring accuracy, which greatly improves the phenomenon of the predicted trajectory drifting caused by the unstable speed of non-motor vehicles.

[0221] Based on the non-motor vehicle motion trajectory prediction method provided in the above embodiment, the present application also provides a specific implementation of a non-motor vehicle motion trajectory prediction device. It can be understood that the relevant descriptions in the following device embodiments can refer to the above method embodiments, and for the sake of brevity, they will not be repeated. Please refer to the following embodiments.

[0222] See also Figure 4 , is a structural diagram of a non-motor vehicle motion trajectory prediction device 400 provided in an embodiment of the present application, which is applied to a vehicle-mounted terminal. The above-mentioned device 400 may include: an acquisition module 401, a first determination module 402, a second determination module 403 and a prediction module 404.

[0223] The acquisition module 401 is used to obtain first position information and driving path information associated with the vehicle, as well as driving status information and road information associated with a non-motor vehicle sensed by the vehicle, the driving status information includes the driving speed and the angle between the driving direction and the lane direction, and the road information includes second position information. The first position information is used to indicate the first position of the vehicle, and the second position information is used to indicate the second position of the non-motor vehicle.

[0224] The first determination module 402 is used to determine a target driving scene based on the first position information, the driving path information and the second position information, where the target driving scene is used to indicate whether there is interaction between the vehicle and the non-motor vehicle.

[0225] The second determination module 403 is used to determine a target trajectory prediction model corresponding to the target driving scene in a preset trajectory prediction model.

[0226] The prediction module 404 is used to input the driving state information and the road information into the target trajectory prediction model, and obtain the first motion trajectory of the non-motor vehicle by predicting the target trajectory prediction model.

[0227] The prediction device of the motion trajectory of non-motor vehicles in the embodiment of the present application is applied to the vehicle-mounted terminal, and determines the target driving scene based on the first position information, the driving path information, and the second position information, and the target driving scene is used to indicate whether there is interaction between the vehicle and the non-motor vehicle; in the preset trajectory prediction model, the target trajectory prediction model corresponding to the target driving scene is determined; the driving state information and the road information are input into the target trajectory prediction model, and the first motion trajectory of the non-motor vehicle is predicted by the target trajectory prediction model. In this way, the present application can identify non-motor vehicles that affect vehicle driving and predict their motion trajectories, thereby improving the safety of vehicle driving.

[0228] In some embodiments, the first determining module 402 may specifically include:

[0229] A first calculation unit, used for calculating a lateral distance between the vehicle and the non-motor vehicle based on the first position information and the second position information;

[0230] The first determination unit is used to determine that the target driving scene is a road driving scene when the lateral distance between the vehicle and the non-motor vehicle is less than or equal to a distance threshold, the driving path information indicates that the vehicle maintains the lane where the first position is located, and the direction angle is less than a first threshold.

[0231] In some embodiments, the road information may further include road traffic light information, and the road traffic light information may include the signal of the traffic light of the road where the non-motor vehicle is located and the first distance between the non-motor vehicle and the traffic light;

[0232] The prediction module 404 may include:

[0233] a second calculation unit, configured to calculate the acceleration of the non-motor vehicle based on the first distance and the travel speed when the road traffic light information indicates that the traffic light is a stop signal;

[0234] A first prediction unit is used to predict a first motion trajectory of the non-motor vehicle performing uniform deceleration motion with acceleration when the acceleration is a negative value;

[0235] The first prediction unit is further used to determine a first motion trajectory of the non-motor vehicle moving at a uniform speed at a driving speed when the road traffic light information indicates that the traffic light is not a stop signal or the acceleration is not a negative value.

[0236] As an implementation of the present application, in order to identify a driving scene in which the motion trajectory of other non-motor vehicles may affect the driving safety of the vehicle, the first determination module 402 may further include:

[0237] A third calculation unit is used to calculate the minimum spatial distance between the vehicle and the non-motor vehicle within a preset time period based on the driving path information, the second position information, the driving speed and the direction angle when the lateral distance is greater than the distance threshold and the direction angle is greater than or equal to the first threshold;

[0238] a fourth calculation unit, for calculating, when the minimum spatial distance is less than a second threshold value, a first time for the vehicle to arrive at a location corresponding to the minimum spatial distance according to the driving path information, and a second time for the non-motor vehicle to arrive at the location corresponding to the minimum spatial distance according to the driving speed;

[0239] The second determining unit is used to determine that the target driving scene is an interactive driving scene when the first time is less than or equal to the second time.

[0240] In some embodiments, the prediction module 404 may further include:

[0241] A fifth calculation unit, for calculating the interaction time for the non-motor vehicle to arrive at the interaction location based on the driving path information, the second position information, the driving speed and the direction angle, the interaction location being a location when the spatial distance between the vehicle and the non-motor vehicle is equal to the second threshold;

[0242] A third determination unit, configured to determine a parking location of the non-motor vehicle based on the interaction time and the second position information, the driving speed and the direction angle;

[0243] The second prediction unit is used to predict, based on the parking location, a first motion trajectory of the non-motor vehicle performing uniform deceleration motion in a preset time period.

[0244] As another implementation of the present application, in order to identify driving scenarios where the movement trajectory of other non-motor vehicles may affect the driving safety of the vehicle, the road information also includes lane exit information, and the lane exit information includes at least one lane exit of the target intersection and lane information of each lane exit;

[0245] The first determining module 402 may further include:

[0246] A sixth calculation unit, configured to calculate a loss value of each lane exit based on each lane information and the second position information, the driving speed, and the direction angle when the first time is greater than the second time;

[0247] A fourth determination unit is used to determine, when an optimal exit is found at a target intersection, that the target driving scene is an intersection driving scene, and the optimal exit is a lane exit corresponding to a minimum loss value among the loss values ​​of lane exits at the target intersection;

[0248] The fourth determining unit is further configured to determine that the target driving scenario is a free driving scenario when no optimal exit is found at the target intersection.

[0249] In some embodiments, when the target driving scene is an intersection driving scene, the prediction module 404 is also used to predict, based on the second position information and the driving speed, a first motion trajectory of the non-motor vehicle driving towards the optimal exit within a preset time period and moving at a uniform speed along the road corresponding to the optimal exit.

[0250] In some embodiments, the driving state information further includes a position abscissa value, a position ordinate value, a speed magnitude value, a yaw angle value, and an angular velocity value;

[0251] When the target driving scenario is a free driving scenario, the prediction module 404 may further include:

[0252] a fifth determination unit, for inputting the position abscissa value, the position ordinate value, the speed magnitude value, the yaw angle value and the angular velocity value into a target trajectory prediction model, and determining the position abscissa value, the position ordinate value, the speed magnitude value, the yaw angle value and the angular velocity value of the non-motor vehicle at each time point within a preset time period, wherein the target trajectory prediction model includes the relationship between the position abscissa value, the position ordinate value, the speed magnitude value, the yaw angle value and the angular velocity value at different time points;

[0253] The sixth determination unit is used to determine the first motion trajectory of the non-motor vehicle within the preset time period based on the position horizontal coordinate value, position vertical coordinate value, speed magnitude value, yaw angle value and angular velocity value at each time point within the preset time period.

[0254] As another implementation of the present application, in order to identify non-motor vehicles that interfere with vehicle driving, the road information also includes lane boundary information;

[0255] The above-mentioned device 400 may further include:

[0256] The trajectory truncation module is used to truncate the part of the first motion trajectory after the intersection with the separation facility to obtain a second motion trajectory of the non-motor vehicle when the lane boundary information indicates that the first motion trajectory is blocked by the separation facility.

[0257] Figure 5 A schematic diagram of the hardware structure of an electronic device provided in an embodiment of the present application is shown.

[0258] The electronic device may include a processor 501 and a memory 502 storing computer program instructions.

[0259] Specifically, the processor 501 may include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or may be configured to implement one or more integrated circuits of the embodiments of the present application.

[0260] The memory 502 may include a large capacity memory for data or instructions. By way of example and not limitation, the memory 502 may include a hard disk drive (HDD), a floppy disk drive, a flash memory, an optical disk, a magneto-optical disk, a magnetic tape, or a universal serial bus (USB) drive or a combination of two or more of these. In appropriate cases, the memory 502 may include a removable or non-removable (or fixed) medium. In appropriate cases, the memory 502 may be inside or outside the integrated gateway disaster recovery device. In a specific embodiment, the memory 502 is a non-volatile solid-state memory.

[0261] In certain embodiments, the memory 502 may include a read-only memory (ROM), a random access memory (RAM), a magnetic disk storage medium device, an optical storage medium device, a flash memory device, an electrical, optical or other physical / tangible memory storage device. Thus, in general, the memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., a memory device) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the method according to an aspect of the present disclosure.

[0262] The processor 501 reads and executes the computer program instructions stored in the memory 502 to implement any one of the non-motor vehicle motion trajectory prediction methods in the above embodiments.

[0263] In one example, the electronic device may further include a communication interface 503 and a bus 510. Figure 5 As shown, the processor 501, the memory 502, and the communication interface 503 are connected via a bus 510 and communicate with each other.

[0264] The communication interface 503 is mainly used to implement communication between various modules, devices, units and / or equipment in the embodiments of the present application.

[0265] Bus 510 includes hardware, software or both, and the parts of electronic equipment are coupled to each other. For example, but not limitation, bus may include accelerated graphics port (AGP) or other graphics bus, enhanced industrial standard architecture (EISA) bus, front side bus (FSB), hypertransport (HT) interconnection, industrial standard architecture (ISA) bus, infinite bandwidth interconnection, low pin count (LPC) bus, memory bus, micro channel architecture (MCA) bus, peripheral component interconnection (PCI) bus, PCI-Express (PCI-X) bus, serial advanced technology attachment (SATA) bus, video electronics standard association local (VLB) bus or other suitable bus or two or more of these combinations. In appropriate cases, bus 510 may include one or more buses. Although the present application embodiment describes and shows a specific bus, the application considers any suitable bus or interconnection.

[0266] The electronic device can execute the method for predicting the motion trajectory of a non-motor vehicle in the embodiment of the present application, thereby realizing the combination of Figure 1 and Figure 4 A method and device for predicting the motion trajectory of a non-motor vehicle are described.

[0267] In addition, in combination with the non-motor vehicle motion trajectory prediction method in the above embodiment, the present application embodiment can provide a computer-readable storage medium for implementation. The computer-readable storage medium stores computer program instructions; when the computer program instructions are executed by the processor, any non-motor vehicle motion trajectory prediction method in the above embodiment is implemented.

[0268] In combination with the non-motor vehicle motion trajectory prediction method in the above embodiment, the present application embodiment can provide a vehicle to implement the non-motor vehicle motion trajectory prediction method in any one of the above embodiments.

[0269] It should be clear that the present application is not limited to the specific configuration and processing described above and shown in the figures. For the sake of simplicity, a detailed description of the known method is omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present application is not limited to the specific steps described and shown, and those skilled in the art can make various changes, modifications and additions, or change the order between the steps after understanding the spirit of the present application.

[0270] The functional blocks shown in the above-described block diagram can be implemented as hardware, software, firmware or a combination thereof. When implemented in hardware, it can be, for example, an electronic circuit, an application specific integrated circuit (ASIC), appropriate firmware, a plug-in, a function card, etc. When implemented in software, the elements of the present application are programs or code segments that are used to perform the required tasks. The program or code segment can be stored in a machine-readable medium, or transmitted on a transmission medium or a communication link by a data signal carried in a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, optical fiber media, radio frequency (RF) links, etc. The code segment can be downloaded via a computer network such as the Internet, an intranet, etc.

[0271] It should also be noted that the exemplary embodiments mentioned in this application describe some methods or systems based on a series of steps or devices. However, this application is not limited to the order of the above steps, that is, the steps can be performed in the order mentioned in the embodiment, or in a different order from the embodiment, or several steps can be performed simultaneously.

[0272] Aspects of the present disclosure are described above with reference to the flowchart and / or block diagram of the method, device (system) and computer program product according to the embodiment of the present disclosure. It should be understood that each box in the flowchart and / or block diagram and the combination of each box in the flowchart and / or block diagram can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device to produce a machine so that these instructions executed by the processor of the computer or other programmable data processing device enable the implementation of the function / action specified in one or more boxes of the flowchart and / or block diagram. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field programmable logic circuit. It can also be understood that each box in the block diagram and / or flowchart and the combination of boxes in the block diagram and / or flowchart can also be implemented by dedicated hardware that performs a specified function or action, or can be implemented by a combination of dedicated hardware and computer instructions.

[0273] The above is only a specific implementation of the present application. Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working processes of the systems, modules and units described above can refer to the corresponding processes in the aforementioned method embodiments, and will not be repeated here. It should be understood that the protection scope of the present application is not limited to this. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed in this application, and these modifications or replacements should be included in the protection scope of this application.

Claims

1. A method for predicting the motion trajectory of a non-motor vehicle, It is characterized in that Applied to a vehicle-mounted terminal, the method comprises: Acquiring first position information and driving path information associated with the vehicle, and acquiring driving state information and road information associated with a non-motor vehicle sensed by the vehicle, wherein the driving state information includes a driving speed and a direction angle between a driving direction and a lane direction, and the road information includes second position information, wherein the first position information is used to indicate a first position of the vehicle, and the second position information is used to indicate a second position of the non-motor vehicle; Determining a target driving scene based on the first position information, the driving path information, and the second position information, wherein the target driving scene is used to indicate whether there is an interaction between the vehicle and the non-motor vehicle; Determining a target trajectory prediction model corresponding to the target driving scenario in a preset trajectory prediction model; Inputting the driving state information and the road information into the target trajectory prediction model, and predicting the first motion trajectory of the non-motor vehicle by the target trajectory prediction model; The determining a target driving scene based on the first position information, the driving path information, and the second position information includes: Calculating a lateral distance between the vehicle and the non-motor vehicle based on the first position information and the second position information; When the lateral distance is less than or equal to a distance threshold, the driving path information indicates that the vehicle maintains the lane where the first position is located, and the direction angle is less than a first threshold, it is determined that the target driving scene is a road driving scene.

2. The method according to claim 1, It is characterized in that The road information further includes road traffic light information, wherein the road traffic light information includes a signal of a traffic light on the road where the non-motor vehicle is located and a first distance between the non-motor vehicle and the traffic light; The step of inputting the driving state information and the road information into the target trajectory prediction model, and predicting the first motion trajectory of the non-motor vehicle by the target trajectory prediction model, comprises: When the road traffic light information indicates that the traffic light is a stop signal, calculating the acceleration of the non-motor vehicle based on the first distance and the travel speed; When the acceleration is a negative value, a first motion trajectory of the non-motor vehicle performing uniform deceleration motion at the acceleration is predicted; When the road traffic light information indicates that the traffic light is not a stop signal, or the acceleration is not a negative value, a first motion trajectory of the non-motor vehicle moving at a uniform speed at the driving speed is determined.

3. The method according to claim 1, It is characterized in that The determining a target driving scene based on the first position information, the driving path information and the second position information further includes: When the lateral distance is greater than the distance threshold and the direction angle is greater than or equal to a first threshold, the minimum spatial distance between the vehicle and the non-motor vehicle within a preset time period is calculated based on the driving path information, the second position information, the driving speed and the direction angle; When the minimum spatial distance is less than a second threshold, calculating, based on the driving path information, a first time for the vehicle to arrive at a location corresponding to the minimum spatial distance, and, based on the driving speed, calculating a second time for the non-motor vehicle to arrive at the location corresponding to the minimum spatial distance; When the first time is less than or equal to the second time, the target driving scene is determined to be an interactive driving scene.

4. The method according to claim 3, It is characterized in that The step of inputting the driving state information and the road information into the target trajectory prediction model, and predicting the first motion trajectory of the non-motor vehicle by the target trajectory prediction model, further comprises: Calculating the interaction time for the non-motor vehicle to reach an interaction location based on the driving path information, the second location information, the driving speed and the direction angle, wherein the interaction location is a location when the spatial distance between the vehicle and the non-motor vehicle is equal to the second threshold value; Determining a parking location of the non-motor vehicle based on the interaction time and the second location information, the driving speed, and the direction angle; Based on the parking location, a first motion trajectory of the non-motor vehicle performing uniform deceleration motion in a preset time period is predicted.

5. The method according to claim 3, It is characterized in that The road information further includes lane exit information, wherein the lane exit information includes at least one lane exit of the target intersection and lane information of each lane exit; The determining a target driving scene based on the first position information, the driving path information and the second position information further includes: When the first time is greater than the second time, based on the lane information and the second position information, the driving speed, and the direction angle, the loss value of each lane exit is calculated; When an optimal exit is found at the target intersection, the target driving scene is determined to be an intersection driving scene, and the optimal exit is a lane exit corresponding to the minimum loss value among the loss values ​​of the lane exits at the target intersection; When no optimal exit is found in the target intersection, the target driving scene is determined to be a free driving scene.

6. The method according to claim 5, It is characterized in that In the case where the target driving scene is the intersection driving scene, the step of inputting the driving state information and the road information into the target trajectory prediction model, and predicting the first motion trajectory of the non-motor vehicle by the target trajectory prediction model, further includes: Based on the second position information and the driving speed, a first motion trajectory of the non-motor vehicle driving towards the optimal exit and moving at a uniform speed along the road corresponding to the optimal exit within a preset time period is predicted.

7. The method according to claim 5, It is characterized in that The driving state information also includes a position abscissa value, a position ordinate value, a speed magnitude value, a yaw angle value and an angular velocity value; In the case where the target driving scene is the free driving scene, the step of inputting the driving state information and the road information into the target trajectory prediction model, and predicting the first motion trajectory of the non-motor vehicle by the target trajectory prediction model, further includes: The position abscissa value, the position ordinate value, the speed magnitude value, the yaw angle value and the angular velocity value are input into the target trajectory prediction model to determine the position abscissa value, the position ordinate value, the speed magnitude value, the yaw angle value and the angular velocity value of the non-motor vehicle at each time point within a preset time period, wherein the target trajectory prediction model includes the relationship between the position abscissa value, the position ordinate value, the speed magnitude value, the yaw angle value and the angular velocity value at different time points; Based on the position abscissa value, position ordinate value, speed magnitude value, yaw angle value and angular velocity value at each time point in the preset time period, a first motion trajectory of the non-motor vehicle in the preset time period is determined.

8. The method according to claim 1, It is characterized in that The road information also includes lane boundary information; After the driving state information and the road information are input into the target trajectory prediction model and the target trajectory prediction model predicts the first motion trajectory of the non-motor vehicle, the method further includes: When the lane boundary information indicates that the first motion trajectory is blocked by a dividing facility, a portion of the first motion trajectory that intersects the dividing facility is cut off to obtain a second motion trajectory of the non-motor vehicle.

9. A method and device for predicting the motion trajectory of a non-motor vehicle, It is characterized in that Applied to a vehicle-mounted terminal, the device comprises: an acquisition module, used to acquire first position information and driving path information associated with the vehicle, and to acquire driving state information and road information associated with a non-motor vehicle sensed by the vehicle, wherein the driving state information includes a driving speed and a direction angle between a driving direction and a lane direction, and the road information includes second position information, wherein the first position information is used to indicate a position of the vehicle, and the second position information is used to indicate a position of the non-motor vehicle; A first determination module, configured to determine a target driving scene based on the first position information, the driving path information, and the second position information, wherein the target driving scene is used to indicate whether there is an interaction between the vehicle and the non-motor vehicle; A second determination module is used to determine a target trajectory prediction model corresponding to the target driving scene in a preset trajectory prediction model; A prediction module, used for inputting the driving state information and the road information into the target trajectory prediction model, and predicting the first motion trajectory of the non-motor vehicle by the target trajectory prediction model; The first determining module includes: A first calculation unit, configured to calculate a lateral distance between the vehicle and the non-motor vehicle based on the first position information and the second position information; The first determination unit is used to determine that the target driving scene is a road driving scene when the lateral distance is less than or equal to a distance threshold, the driving path information indicates that the vehicle maintains the lane where the first position is located, and the direction angle is less than a first threshold.

Images