CN114945961B

CN114945961B - Lane changing prediction regression model training method, lane changing prediction method and apparatus

Info

Publication number: CN114945961B
Application number: CN202080092993.3A
Authority: CN
Inventors: 许家妙; 何明; 叶茂胜; 曹通易
Original assignee: DeepRoute AI Ltd
Current assignee: DeepRoute AI Ltd
Priority date: 2020-11-12
Filing date: 2020-11-12
Publication date: 2024-03-08
Anticipated expiration: 2040-11-12
Also published as: CN114945961A; WO2022099526A1

Abstract

A lane-changing predictive regression model training method comprises the following steps: acquiring a first sample motion characteristic, wherein the first sample motion characteristic is obtained according to motion data of a sample moving object acquired at a sample acquisition time; obtaining standard lane change probability corresponding to the first sample motion feature, wherein the standard lane change probability is determined according to a lane change time interval between the object lane change time corresponding to the first sample motion feature and the sample acquisition time, and the standard lane change probability and the lane change time interval form a negative correlation; inputting the first sample motion characteristics into a lane change prediction regression model to be trained to obtain predicted lane change probability corresponding to the first sample motion characteristics; obtaining a model loss value according to the probability difference between the predicted lane change probability and the standard lane change probability; and adjusting model parameters in the lane change prediction regression model by using the model loss value to obtain the trained lane change prediction regression model.

Description

Lane changing prediction regression model training method, lane changing prediction method and apparatus

Technical Field

The application relates to a lane change prediction regression model training method, a lane change prediction device, computer equipment and a storage medium.

Background

With the development of artificial intelligence, an automatic driving automobile, which is an intelligent automobile realizing unmanned driving through a computer system, relies on cooperation of artificial intelligence, visual computing, a radar, a monitoring device and a global positioning system, so that the computer system automatically and safely controls the automobile to run without active operation of human beings. In the running process of the automatic driving automobile, the movement condition of a moving object in the running process or near a running path needs to be detected so as to avoid the moving object, thereby ensuring the safety of the automatic driving automobile.

However, the inventor has appreciated that the current approach for identifying the state of motion of a moving object is inaccurate, resulting in a low obstacle avoidance capability of the autonomous vehicle, thus making the autonomous vehicle less safe.

Disclosure of Invention

According to various embodiments disclosed herein, a lane change prediction regression model training method, a lane change prediction apparatus, a computer device, and a storage medium are provided.

A lane-changing predictive regression model training method comprises the following steps:

acquiring a first sample motion characteristic, wherein the first sample motion characteristic is obtained according to motion data of a sample moving object acquired at a sample acquisition time;

obtaining standard lane change probability corresponding to the first sample motion feature, wherein the standard lane change probability is determined according to a lane change time interval between the object lane change time corresponding to the first sample motion feature and the sample acquisition time, and the standard lane change probability and the lane change time interval form a negative correlation;

inputting the first sample motion characteristics into a lane change prediction regression model to be trained to obtain predicted lane change probability corresponding to the first sample motion characteristics;

obtaining a model loss value according to the probability difference between the predicted lane change probability and the standard lane change probability; and

And adjusting model parameters in the lane change prediction regression model by using the model loss value to obtain the trained lane change prediction regression model so as to perform lane change prediction according to the trained lane change prediction regression model.

A lane-changing predictive regression model training apparatus comprising:

the first sample motion feature acquisition module is used for acquiring first sample motion features, wherein the first sample motion features are obtained according to motion data of a sample motion object acquired at the sample acquisition time;

The standard lane change probability acquisition module is used for acquiring standard lane change probability corresponding to the first sample motion feature, wherein the standard lane change probability is determined according to a lane change time interval between the object lane change time corresponding to the first sample motion feature and the sample acquisition time, and the standard lane change probability and the lane change time interval form a negative correlation;

the prediction lane change probability obtaining module is used for inputting the first sample motion characteristic into a lane change prediction regression model to be trained to obtain a prediction lane change probability corresponding to the first sample motion characteristic;

the model loss value obtaining module is used for obtaining a model loss value according to the probability difference between the predicted lane change probability and the standard lane change probability; and

And the trained lane change prediction regression model obtaining module is used for utilizing the model loss value to adjust model parameters in the lane change prediction regression model to obtain the trained lane change prediction regression model so as to conduct lane change prediction according to the trained lane change prediction regression model.

A computer device comprising a memory and one or more processors, the memory having stored therein computer-readable instructions that, when executed by the processor, cause the one or more processors to perform the steps of:

One or more computer storage media storing computer-readable instructions that, when executed by one or more processors, cause the one or more processors to perform the steps of:

A lane change prediction method, comprising:

acquiring a target motion characteristic corresponding to a target motion object;

inputting the target motion characteristics into a trained lane change prediction regression model to obtain target lane change probability corresponding to the target motion object; the trained lane change prediction regression model is obtained through training according to a first sample motion feature and a standard lane change probability corresponding to the first sample motion feature, the first sample motion feature is obtained according to motion data of a sample moving object acquired at a sample acquisition time, the standard lane change probability is determined according to a lane change time interval between an object lane change time corresponding to the first sample motion feature and the sample acquisition time, and the standard lane change probability and the lane change time interval form a negative correlation; and

And determining a target lane change result corresponding to the target moving object according to the target lane change probability.

A lane change prediction apparatus comprising:

the target motion characteristic acquisition module is used for acquiring target motion characteristics corresponding to the target motion object;

the target lane change probability obtaining module is used for inputting the target motion characteristics into a trained lane change prediction regression model to obtain target lane change probability corresponding to the target motion object; the trained lane change prediction regression model is obtained through training according to a first sample motion feature and a standard lane change probability corresponding to the first sample motion feature, the first sample motion feature is obtained according to motion data of a sample moving object acquired at a sample acquisition time, the standard lane change probability is determined according to a lane change time interval between an object lane change time corresponding to the first sample motion feature and the sample acquisition time, and the standard lane change probability and the lane change time interval form a negative correlation; and

And the target lane change result determining module is used for determining a target lane change result corresponding to the target moving object according to the target lane change probability.

The details of one or more embodiments of the application are set forth in the accompanying drawings and the description below. Other features and advantages of the application will be apparent from the description and drawings, and from the claims.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are needed in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.

FIG. 1 is an application scenario diagram of a lane change predictive regression model training method in accordance with one or more embodiments;

FIG. 2 is a flow diagram of a method of training a lane change predictive regression model in accordance with one or more embodiments;

FIG. 3 is a flow diagram of steps for obtaining a target time interval threshold and a target lane change probability generation factor in accordance with one or more embodiments;

FIG. 4 is a flow diagram of a lane change prediction method in accordance with one or more embodiments;

FIG. 5 is a block diagram of a lane change predictive regression model training apparatus in accordance with one or more embodiments;

FIG. 6 is a block diagram of a lane change prediction apparatus in accordance with one or more embodiments;

FIG. 7 is a block diagram of a computer device in accordance with one or more embodiments;

FIG. 8 is a block diagram of a computer device in accordance with one or more embodiments.

Detailed Description

In order to make the technical solution and advantages of the present application more apparent, the present application will be further described in detail with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the present application.

The lane change prediction regression model training method provided by the application can be applied to an application environment shown in fig. 1. The terminal 102 communicates with the server 104 via a network. The server 104 obtains a first sample motion feature, the first sample motion feature is obtained according to motion data of a sample motion object acquired at a sample acquisition time, obtains a standard lane change probability corresponding to the first sample motion feature, the standard lane change probability is determined according to a lane change time interval between an object lane change time corresponding to the first sample motion feature and the sample acquisition time, the standard lane change probability and the lane change time interval form a negative correlation, the first sample motion feature is input into a lane change prediction regression model to be trained, the predicted lane change probability corresponding to the first sample motion feature is obtained, a model loss value is obtained according to a probability difference between the predicted lane change probability and the standard lane change probability, model parameters in the lane change prediction regression model are adjusted by using the model loss value, and a trained lane change prediction regression model is obtained so as to conduct lane change prediction according to the trained lane change prediction regression model. The server 104 may perform lane change prediction using the trained lane change prediction regression model, control the motion according to the result of the lane change prediction, the server 104 may also transmit the trained lane change prediction regression model to the terminal 102, the terminal 102 may perform lane change prediction using the trained lane change prediction regression model, and the server or the terminal may control the terminal to perform the motion according to the result of the lane change prediction. The terminal 102 may be, but is not limited to, an autonomous car and a mobile robot, and the server 104 may be implemented as a stand-alone server or as a server cluster of multiple servers.

In some embodiments, as shown in fig. 2, a lane-changing prediction regression model training method is provided, and the method is applied to the server 104 in fig. 1 for illustration, and includes the following steps:

s202, acquiring a first sample motion characteristic, wherein the first sample motion characteristic is obtained according to motion data of a sample moving object acquired at a sample acquisition time.

Specifically, a moving object refers to an object in a moving state. The object may be a living object, for example, may comprise at least one of a human or an animal, or may be an inanimate object, for example, may be at least one of a vehicle, an aircraft, or a bicycle. The sample moving object may be any moving object, or may be a designated moving object.

The motion data may be data related to motion of the sample moving object, which is acquired by the sample acquisition time, may include point cloud (point cloud) data acquired by a point cloud acquisition device during motion of the sample moving object, and may also include at least one of image data acquired by an image acquisition device. A point cloud refers to a set of three-dimensional data points in a three-dimensional coordinate system, for example, may be a set of three-dimensional data points corresponding to a surface of an object in the three-dimensional coordinate system, and the point cloud may represent an outer surface shape of an object. Three-dimensional data points refer to points in three-dimensional space. The three-dimensional data points may also include at least one of RGB colors, gray values, or time. The point cloud may be obtained by laser radar scanning. The point cloud acquisition device may be any device capable of acquiring point cloud data, and may be, but not limited to, a laser radar, for example, a laser radar arranged on top of an autonomous vehicle. The image capture device may be any device that can capture image data, and may be, but is not limited to, a camera. The laser radar is an active sensor, and after the laser beam is emitted to the surface of an object, the laser beam is bounced, and a bounced laser signal is collected to obtain the point cloud of the object. The motion data may be pre-stored in the server. The motion data may be acquired by a point cloud acquisition device or an image acquisition device installed in the sample moving object, or may be acquired by a point cloud acquisition device or an image acquisition device in the surrounding environment of the sample moving object, for example, may be acquired by a device installed in an object whose distance from the sample moving object is smaller than a distance threshold.

The motion features refer to motion-related features calculated from motion data. The motion characteristics may be calculated according to motion characteristic related data extracted from motion data, for example, the motion characteristics may be motion characteristic related data, or may be calculated according to motion characteristic related data corresponding to motion data at different moments. For example, the server may select at least two point cloud frames from the point cloud data, extract motion feature related data from each selected point cloud frame, perform data fitting on the motion feature related data to obtain a fitting result, and use the fitting result as a motion feature, for example, may calculate a statistical value of the motion feature related data to obtain the motion feature. The statistical value may include at least one of a mean or a variance. The motion feature related data may be extracted from the motion data by a neural network model. For example, the server may input the motion data into a trained motion feature related data recognition model, which may process the motion data, such as a convolution process, to obtain motion feature related data.

The motion characteristic related data may comprise at least one of a position, a velocity, an acceleration, a distance, a direction of motion or a relative distance of the moving object, for example may comprise at least one of a position, a velocity, an acceleration or a direction of motion of the moving object in a world coordinate system. The world coordinate system may be a three-dimensional coordinate system, and the positions in the world coordinate system may be represented by (x, y, z), where x, y, and z are coordinate axes perpendicular to each other and intersecting each other. The speed may include, for example, at least one of a horizontal speed or a vertical speed of the moving object with respect to the center line of the roadway, the horizontal speed referring to a speed in a direction parallel to the center line of the roadway, and the vertical speed referring to a speed in a direction perpendicular to the center line of the roadway. The direction of movement may be, for example, the vehicle orientation. The relative distance may be a distance between the sample moving object and a reference object, which may be an object in a stationary state, for example, at least one of a road boundary or an object provided on a road, for example, a tree. The reference may also be a virtual object, such as a route, which refers to the path of flight of the aircraft, or a position in a three-dimensional coordinate system, such as the origin of the three-dimensional coordinate system. Road boundaries may also be referred to as lane boundaries. That is, the relative distance may be a distance of the moving object from the road boundary, and the distance of the moving object from the road boundary may be referred to as a road boundary distance. The road boundary distance may include at least one of a road left boundary distance or a road right boundary distance. The road left boundary distance refers to the distance between the sample moving object and the road left boundary, and the road right boundary distance refers to the distance between the sample moving object and the road right boundary.

The motion feature may be determined according to a sorting result of the motion feature related data, the sorting result of the motion feature related data may be used as a motion feature, for example, the relative distances at different moments may be sorted according to a time sequence to obtain a distance sorting result, the distance sorting result may be used as a motion feature, or the sorting result of the motion feature related data may be calculated to obtain a motion feature, for example, data in the distance sorting result may be normalized to obtain a motion feature, or the data in the distance sorting result may be statistically calculated or data fitted to obtain a motion feature, or the sorting result of the motion feature related data may be spliced to obtain a splicing result, for example, the distance sorting result may be spliced with a velocity sorting result to obtain a motion feature, the velocity sorting result may be sorted according to a time sequence, for example, the horizontal velocity at different moments may be sorted according to a time sequence, for example, the distance sorting result is (s 1, s2, s 3), the velocity sorting result is (v 1, v2, v 3), and the distance sorting result and the velocity sorting result may be spliced to be expressed as (s 1, s2, v 3). Of course, statistical calculation or normalization processing can be performed on the data in the splicing result to obtain the motion characteristics. The number of relative distances in the distance ranking result may or may not be consistent with the number of speeds in the speed ranking result. The point cloud frames corresponding to the relative distances in the distance sorting result may or may not be identical to the point cloud frames corresponding to the speed in the speed sorting result.

The sample motion features refer to motion features of a sample motion object, and the first sample motion features can be any sample motion features or designated sample motion features and are used for training a lane change prediction regression model to obtain a trained lane change prediction regression model. The lane change predictive regression model is a regression model for predicting the probability of occurrence of a lane change. The input of the lane change predictive regression model may be a motion feature and the output may be a probability of a lane change occurring. The lane change prediction regression model may be an existing regression model or a custom regression model, for example, may be a four-layer neural network. Regression models may also be referred to as regressors and regressive neural networks. The sample collection time refers to the time when the motion data is collected, and may be a time point or a time period, for example, (t 1, t 2), where t1 represents the start time of the sample collection time, and t2 represents the end time of the sample collection time, and the duration of the time period may be 5 minutes, for example.

In some embodiments, the server may obtain a moving point cloud of the sample moving object in the moving process from the point cloud collecting device, obtain a first number of point cloud frames from the point cloud frames included in the obtained moving point cloud, determine motion feature related data of the sample moving object in each point cloud frame in the first number of point cloud frames, sort the motion feature related data corresponding to each point cloud frame according to the time sequence of the point cloud frames, obtain a sorting result, obtain a sample motion feature according to the sorting result, for example, calculate a statistical value of data in the sorting result, obtain a sample motion feature, or perform data fitting on the data in the sorting result, obtain a sample motion feature according to the result of the data fitting, for example, may use the result of the data fitting as the sample motion feature, or perform normalization processing on the result of the data fitting to obtain the sample motion feature. The point cloud frames in the first number of point cloud frames may be continuous or discontinuous. The first number may be preset, or may be set as needed, for example, may be 10.

S204, obtaining standard lane change probability corresponding to the first sample motion feature, wherein the standard lane change probability is determined according to a lane change time interval between the object lane change time corresponding to the first sample motion feature and the sample acquisition time, and the standard lane change probability and the lane change time interval form a negative correlation.

Specifically, the object lane change time refers to the time when the sample moving object is lane-changed. The object lane change time corresponding to the first sample motion feature refers to the time when the first lane change of the sample motion object occurs after the sample acquisition time. For example, if the sample moving object changes lanes at time t2+a1 after the sample acquisition time (t 1, t 2), t2+a1 is the object lane change time. The lane change time interval refers to the time interval between the lane change time of the object corresponding to the first sample motion feature and the sample acquisition time. The standard lane change probability corresponding to the first sample motion feature refers to the true lane change probability of the first sample motion feature. For any lane change interval, the standard lane change probability may be inversely related to the lane change interval. Of course, when the lane change time interval satisfies the time interval condition, the standard lane change probability and the lane change time interval may be in a negative correlation. The time interval condition may include the lane change time interval being less than or equal to a target time interval threshold. The target time interval threshold may be preset, or may be determined by training a lane change prediction regression model, for example, may be determined by cross-validation. When the standard lane change probability and the lane change time interval form a negative correlation, the standard lane change probability can be calculated according to the square value of the lane change time interval, for example, the standard lane change probability can be obtained according to the ratio of the square value of the lane change time interval to a specific value, and the specific value can be preset or determined by training a lane change prediction regression model.

In some embodiments, the server may obtain a fixed lane change probability as the standard lane change probability when the lane change time interval is greater than the target time interval threshold. The fixed lane change probability may be set as required, or may be preset, for example, may be 0.

S206, inputting the first sample motion characteristic into a lane change prediction regression model to be trained, and obtaining the predicted lane change probability corresponding to the first sample motion characteristic.

Specifically, the lane change predictive regression model to be trained refers to a lane change predictive regression model that needs to be trained. The server may use the first sample motion feature as an input of a lane change prediction regression model to be trained, and the lane change prediction regression model to be trained may calculate, for example, perform convolution calculation, the first sample motion feature to obtain a predicted lane change probability corresponding to the first sample motion feature. The predicted lane change probability is the lane change probability output by the lane change prediction regression model.

S208, obtaining a model loss value according to the probability difference between the predicted lane change probability and the standard lane change probability.

Specifically, the probability difference refers to a difference between the predicted lane change probability and the standard lane change probability. The model loss value is positively correlated with the probability difference. The model loss value may be, for example, any one of the square of the probability difference or a multiple of the square of the probability difference, for example, a multiple of one-half.

In some embodiments, the model loss value may be determined from a magnitude relationship between the probability difference and the difference threshold. For example, when the probability difference is less than the difference threshold, a model loss value may be determined from the square of the probability difference, the model loss value being in positive correlation with the square of the probability difference. When the probability difference is greater than or equal to the difference threshold, a model loss value can be calculated according to the probability difference and the difference threshold. The difference threshold may be preset, or may be set as needed.

S210, adjusting model parameters in the lane change prediction regression model by using the model loss value to obtain a trained lane change prediction regression model, so as to conduct lane change prediction according to the trained lane change prediction regression model.

In particular, the model parameters refer to variable parameters inside the lane-changing predictive regression model, which may also be referred to as neural network weights (weights) for the neural network model. The trained lane-changing predictive regression model may be trained one or more times. For example, the server may adjust model parameters in the lane change prediction regression model toward a direction in which the loss value becomes smaller, and may perform multiple iterative training to obtain a trained lane change prediction regression model.

In some embodiments, the server may perform back propagation according to the model loss value, and update model parameters of the lane change prediction regression model along the gradient descent direction in the back propagation process, to obtain a trained lane change prediction regression model. The reverse direction means that the update of the parameters is opposite to the direction of the prediction of the lane change, and the update of the parameters is counter-propagating, so that a descending gradient can be obtained according to the model loss value, the last layer of the lane change prediction regression model is started, and the gradient update of the model parameters is started according to the descending gradient until the first layer of the lane change prediction regression model is reached. The gradient descent method may be any one of a random gradient descent method or a batch gradient descent method. It can be understood that the training of the model may be iterated multiple times, that is, the trained lane change prediction regression model may be obtained by iterating the training, and stopping the training when the model convergence condition is satisfied, where the model convergence condition may be that the model loss value is smaller than a preset loss value, or that the change of the model parameter is smaller than a preset parameter change value.

In some embodiments, the server may transmit the trained lane change prediction regression model to the terminal, where the terminal uses the trained lane change prediction regression model to perform lane change prediction on the moving object in the surrounding environment, thereby controlling the movement of the terminal and implementing obstacle avoidance.

In the lane change prediction regression model training method, the first sample motion characteristic is obtained according to the motion data of the sample moving object acquired at the sample acquisition time, the standard lane change probability corresponding to the first sample motion characteristic is obtained, the first sample motion characteristic is input into the lane change prediction regression model to be trained, the predicted lane change probability corresponding to the first sample motion characteristic is obtained, the model loss value is obtained according to the probability difference between the predicted lane change probability and the standard lane change probability, the model parameters in the lane change prediction regression model are adjusted by the model loss value, the trained lane change prediction regression model is obtained, lane change prediction is carried out according to the trained lane change prediction regression model, and as the standard lane change probability is determined according to the lane change time interval between the object lane change time corresponding to the first sample motion characteristic and the sample acquisition time, the standard lane change probability and the lane change time interval form a negative correlation relationship, thereby improving the accuracy of the trained lane change prediction regression model.

In some embodiments, obtaining the standard lane change probability corresponding to the first sample motion feature comprises: when the lane change time interval is smaller than or equal to a target time interval threshold value, acquiring a target lane change probability generating factor; and calculating according to the lane change time interval and the target lane change probability generation factor to obtain the standard lane change probability corresponding to the first sample motion characteristic.

Specifically, the time interval threshold may be a period of time, for example, may be 10 seconds. The target time interval threshold may be selected from candidate time interval thresholds. The candidate time interval threshold may be a time interval threshold stored in the server in advance, or may be a time interval threshold acquired or generated by the server. The lane change probability generation factor is used to generate a lane change probability. The lane change probability refers to the probability that a moving object is lane changed. The target lane change probability generating factor may be selected from candidate lane change probability generating factors. The candidate lane change probability generation factor may be a lane change probability generation factor stored in the server in advance, or may be a lane change probability generation factor acquired or generated by the server. The target lane-change probability generating factor may be determined by training a lane-change predictive regression model. For example, the server may generate candidate lane change probability according to the candidate lane change probability generation factor and the lane change time interval, use the candidate lane change probability as a probability label of the second sample motion feature, train the lane change prediction regression model by using the second sample motion feature and the corresponding probability label, and determine the target lane change probability generation factor according to the accuracy of the trained lane change prediction regression model. The second sample motion feature may be any sample motion feature or may be a specified sample motion feature, for example, a sample motion feature different from the first sample motion feature. Probability labels may also be referred to as regression truths.

In some embodiments, the server may calculate the lane change time interval and the target lane change probability generating factor by using a lane change probability calculation formula, so as to obtain a standard lane change probability corresponding to the first sample motion feature. In the lane change probability calculation formula, when the value of the independent variable is greater than or equal to the value, the dependent variable is reduced along with the increase of the independent variable, and the value of the dependent variable is greater than 0 and less than 1. The server can replace the independent variable in the lane change probability calculation formula with the lane change time interval, replace the constant in the lane change probability calculation formula with the target lane change probability generation factor, and calculate the standard lane change probability corresponding to the first sample motion characteristic. The lane change probability calculation formula may be, for example, a gaussian function. The server can calculate the lane change time interval and the target lane change probability generating factor by using a Gaussian function to obtain the standard lane change probability corresponding to the first sample motion characteristic. For example, the server may replace the argument of the gaussian function with the lane change time interval, replace the variance in the gaussian function with the target lane change probability generating factor, set the mean value of the gaussian function to 0, and calculate the standard lane change probability corresponding to the first sample motion feature.

In the above embodiment, when the lane change time interval is smaller than or equal to the target time interval threshold, the target lane change probability generating factor is obtained, and calculation is performed according to the lane change time interval and the target lane change probability generating factor, so as to obtain the standard lane change probability corresponding to the first sample motion feature, thereby ensuring that the lane change time interval corresponding to the standard lane change probability is not too large, ensuring the effectiveness of the lane change time interval, and improving the accuracy of the standard lane change probability.

In some embodiments, calculating according to the lane change time interval and the target lane change probability generating factor, the obtaining the standard lane change probability corresponding to the first sample motion feature includes: square operation is carried out on the lane change time interval to obtain an interval square value; calculating the ratio of the square value of the interval to the target lane change probability generating factor to obtain a target ratio; and taking the first numerical value as a base number, and taking the negative number of the target ratio as an index to perform index calculation to obtain the standard lane change probability corresponding to the first sample motion characteristic.

Specifically, the square value of the interval is the square of the lane change time interval, e.g. the lane change time interval is d, the square value of the interval is d ² . The target ratio refers to the ratio of the square value of the interval to the target lane change probability generation factor, and can be d ² and/C, C represents a target lane change probability generation factor. The first value may be any positive number greater than 1, for example, a base number e≡2.71828 of natural logarithm.

In some embodiments, the server may determine the exponent from the target ratio using the first value as a base, and perform an exponential calculation to obtain the standard lane change probability. For example, the server may perform an exponential calculation with the first value as a base and the negative value of the target ratio as an index, to obtain a standard lane change probability corresponding to the first sample motion feature. The server may perform an exponential calculation with the first value as a base number and a ratio of the negative value of the target ratio to the second value as an index, to obtain a standard lane change probability. The second value may be any positive number, for example 2. The standard lane change probability can be expressed, for example, as formula (1), where c ² Representing the target lane change probability generating factor. f (d) represents the standard lane change probability.

In the above embodiment, the square operation is performed on the lane change time interval to obtain the interval square value, the ratio of the interval square value to the target lane change probability generating factor is calculated to obtain the target ratio, the first numerical value is used as the base number, the negative number of the target ratio is used as the index to perform the index calculation, and the standard lane change probability corresponding to the first sample motion feature is obtained, so that the value of the standard lane change probability can be ensured to be between 0 and 1, and the accuracy of the standard lane change probability is improved.

In some embodiments, as shown in fig. 3, the step of obtaining the target time interval threshold and the target lane change probability generation factor includes:

s302, at least two candidate parameter combinations are determined, wherein the candidate parameter combinations comprise a candidate time interval threshold and a candidate lane change probability generation factor.

Specifically, the server may arbitrarily select one candidate time interval threshold from the plurality of candidate time interval thresholds as the candidate time interval threshold in the candidate parameter combination, and may arbitrarily select one candidate lane change probability generation factor from the plurality of candidate lane change probability generation factors as the candidate time interval threshold in the candidate parameter combination. If the number of candidate time interval thresholds is N, the number of candidate lane change probability generating factors is M, and the number of candidate parameter combinations may be nxm. For example, the candidate time interval thresholds are r1 and r2, and the candidate lane change probability generation factors are w1 and w2, and the candidate parameter combinations may include (r 1, w 1), (r 1, w 2), (r 2, w 1), and (r 2, w 2).

S304, acquiring a second sample motion feature, and determining a lane change probability corresponding to the second sample motion feature according to the candidate parameter combination, wherein the lane change probability is used as a probability label corresponding to the second sample motion feature.

In particular, the second sample motion feature may be any sample motion feature, or may be a specified sample motion feature, for example, a sample motion feature that is different from the first sample motion feature. There may be a plurality of second sample motion features. Probability labels refer to labels of the second sample motion feature, which may also be referred to as regression truth values. The probability label is similar to the standard lane change probability in the obtaining mode, the probability label is determined according to the lane change time interval corresponding to the second sample motion feature, when the lane change time interval is smaller than or equal to the candidate time interval threshold value in the candidate parameter combination, the probability label corresponding to the second sample motion feature is obtained through calculation according to the lane change time interval and the candidate lane change probability generation factor in the candidate parameter combination. When the lane change time interval is greater than the candidate time interval threshold in the candidate parameter combination, the probability label corresponding to the second sample motion feature is determined to be a fixed probability, and the fixed probability can be set according to the requirement or can be preset, for example, can be 0.

In some embodiments, the server may obtain a set of candidate sample motion features, select a second number of candidate sample motion features from the set of candidate sample motion features, and obtain a second set of sample motion features. And forming a third sample motion feature set by the candidate sample motion features except the second sample motion feature set in the candidate sample motion feature set. The second number may be set as desired, e.g. determined from the number of candidate sample motion features in the set of candidate sample motion features (denoted as the total number of features), e.g. a fifth of the total number of features.

S306, training the lane change prediction regression model based on the second sample motion feature and the probability label corresponding to the second sample motion feature to obtain a trained lane change prediction regression model corresponding to the candidate parameter combination.

Specifically, the server may input the second sample motion feature into the lane change prediction regression model to obtain a predicted lane change probability corresponding to the second sample motion feature, and the server may obtain a model loss value according to a probability difference between the predicted lane change probability and the probability label, and adjust model parameters in the lane change prediction regression model by using the model loss value to obtain the trained lane change prediction regression model. For example, each second sample motion feature in the second sample motion feature set may be input into a lane change predictive regression model to obtain a trained lane change predictive regression model.

In some embodiments, the server may obtain a plurality of second sample motion feature sets from the candidate sample motion feature sets, the second sample motion features in different second sample motion feature sets being different. And determining probability labels of the second sample motion features in each second sample motion feature set by using the same candidate parameter combination. And training the lane change prediction regression model by utilizing each second sample motion feature set to obtain a trained lane change prediction regression model obtained by different second sample motion feature sets under the same candidate parameter combination.

S308, determining model accuracy of the trained lane change prediction regression model corresponding to the candidate parameter combination, selecting a target parameter combination meeting accuracy conditions from at least two candidate parameter combinations based on the model accuracy, taking a candidate time interval threshold in the target parameter combination as a target time interval threshold, and taking a candidate lane change probability generation factor in the target parameter combination as a target lane change probability generation factor.

In particular, the candidate parameter combination may further comprise a candidate lane change probability threshold. The server can input the third sample motion feature in the third sample motion feature set into the trained lane change prediction regression model to obtain the predicted lane change probability of the third sample motion feature output by the lane change prediction regression model, compare the predicted lane change probability with a candidate lane change probability threshold, determine that the predicted lane change result of the third sample motion feature is lane change when the predicted lane change probability is greater than the candidate lane change probability threshold, and determine that the predicted lane change result of the third sample motion feature is unchanged when the predicted lane change probability is less than or equal to the candidate lane change probability threshold. The predicted lane change result is a predicted lane change result. The server can acquire a standard lane change result of the third sample motion feature, compares the predicted lane change result with the standard lane change result, and determines that the trained lane change prediction regression model is correct for the predicted result of the third sample motion feature when the comparison is consistent, otherwise, the predicted result is incorrect. The standard lane change result is the actual lane change result of the third sample motion feature. The server can determine the model accuracy of the trained lane change prediction regression model according to the prediction results corresponding to the motion features of the third samples, for example, the number of correct prediction results can be divided by the number of all prediction results to obtain the model accuracy. The accuracy condition may include an accuracy greater than at least one of an accuracy threshold or a model accuracy maximum. The accuracy threshold may be set as desired, or may be preset, for example, may be 95%. The target parameter combination is a candidate parameter combination meeting an accuracy condition in the acquired at least two candidate parameter combinations.

In some embodiments, the server may determine the accuracies of the trained lane-changing prediction regression models obtained by different second sample motion feature sets under the same candidate parameter combination, and calculate an average value of the accuracies as the model accuracies of the trained lane-changing prediction regression models corresponding to the candidate parameter combination. The server may determine model accuracies corresponding to each candidate parameter combination in the at least two candidate parameter combinations, and select a candidate parameter combination satisfying an accuracy condition from the model accuracies corresponding to each candidate parameter combination as a target parameter combination, for example, select a candidate parameter combination corresponding to the maximum model accuracy as the target parameter combination. The server may take the candidate lane change probability threshold in the target parameter combination as the target lane change probability threshold.

In some embodiments, the server may use the trained lane change prediction regression model corresponding to the target parameter combination as the lane change prediction regression model to be trained.

In the above embodiment, the model accuracy of the trained lane change prediction regression model corresponding to the candidate parameter combination is determined, the target parameter combination satisfying the accuracy condition is selected from at least two candidate parameter combinations based on the model accuracy, the candidate time interval threshold in the target parameter combination is used as the target time interval threshold, and the candidate lane change probability generation factor in the target parameter combination is used as the target lane change probability generation factor, so that the target time interval threshold and the target lane change probability generation factor which can enable the accuracy of the lane change prediction regression model to be high are obtained.

In some embodiments, the first sample motion feature comprises a distance change feature, and the step of determining the distance change feature comprises: acquiring a target distance sequence corresponding to a sample moving object in sample acquisition time, wherein the target distance sequence is ordered according to the movement time; determining a distance change trend according to the size relation of the distances in the target distance sequence; and determining the distance change characteristics corresponding to the sample moving object according to the distance change trend.

In particular, a distance sequence refers to a sequence of relative distances, each relative distance in the distance sequence being ordered in terms of motion time. The earlier the movement time is in the distance sequence, the earlier the ordering of the relative distances may be, for example, (S1, S2, S3, S4, S5, S6), where S1 to S6 each represent a relative distance. The motion time refers to the time in the motion process of the sample moving object, and can be a time point or a time period. The sample moving object may correspond to a plurality of distance sequences, for example, may correspond to at least one of a left boundary distance sequence or a right boundary distance sequence. The left boundary distance sequence refers to a sequence obtained by sequencing left boundary distances of roads according to time sequence, and the right boundary distance sequence refers to a sequence obtained by sequencing right boundary distances of roads according to time sequence. The distance sequence may also be represented by a vector. The sample moving object may correspond to a plurality of distance sequences, for example, the server may obtain a plurality of distance sequences corresponding to the sample moving object according to the motion data acquired at the sample acquisition time, and the target distance sequence may be a distance sequence selected from the respective distance sequences corresponding to the sample moving object. The distance sequence can be selected in a random selection mode or a preset selection mode, for example, the distance sequence can be selected according to the distance change trend corresponding to the distance sequence, and the distance sequence meeting the distance change trend condition is used as a target distance sequence. Of course, each distance sequence of the sample moving object may be respectively taken as the target distance sequence. There may be multiple target distance sequences. The distance change trend refers to a law of distance change over time in a distance sequence, and may include any one of a gradually smaller change trend or a gradually larger change trend. The distance change trend condition may be, for example, a gradually decreasing change trend. For example, when the distance change trend of the left boundary distance sequence is a gradually decreasing change trend, the left boundary distance sequence is set as the target distance sequence, and when the distance change trend of the left boundary distance sequence is a gradually increasing change trend, the right boundary distance sequence is set as the target distance sequence.

The distance change feature may be a feature of a distance change trend, may be obtained by calculating according to the distance change trend, for example, may determine the speed of the distance change according to the distance change trend, so as to obtain the distance change feature. The degree of the distance change may include at least one of a degree of the distance increase or a degree of the distance decrease. The distance change feature can also be determined according to an average distance change degree, wherein the average distance change degree refers to the size of the distance change in unit time, and the distance change degree can be calculated according to the target distance sequence and a time sequence corresponding to the target distance sequence. The time sequence corresponding to the target distance sequence comprises time corresponding to each distance in the target distance sequence, the time in the time sequence is ordered according to the sequence from front to back, and the earlier the time is, the earlier the ordering is. The time sequence may be, for example, (t 1, t2, t3, t4, t5, t 6). Wherein t1 occurs before t 6. For example, the distance difference may be obtained by a difference between the distance of the start position and the distance of the end position in the target distance sequence, the time difference may be obtained from the difference between the time of the start position and the end position in the time sequence, and the average distance change degree may be (s 6-s 1)/(t 6-t 1) for example, obtained from the ratio of the distance difference to the time difference. The distance change feature may be represented by a vector, which may be referred to as a distance feature vector, for example. The distance change feature may also be a target distance sequence, i.e. the target distance sequence may be taken as the distance change feature.

In some embodiments, the server may calculate the target distance sequence and the corresponding time sequence using a change feature calculation method to obtain the distance transformation feature. The change characteristic calculation method may be, for example, formulas (2) and (3). Wherein x represents the distance in the target distance sequence and y represents the time corresponding to the distance in the target distance sequence. a represents a distance change feature, which can also be understood as a rate of change (slope) of the relative distance over time.Mean value of y,/-, is shown>The average value of x is shown.

In the above embodiment, in the sample collection time, the target distance sequence corresponding to the sample moving object is obtained, the target distance sequence is ordered according to the movement time, the distance change trend is determined according to the magnitude relation of the distances in the target distance sequence, the distance change feature corresponding to the sample moving object is determined according to the distance change trend, and the accuracy of the first sample moving feature is improved.

In some embodiments, as shown in fig. 4, a lane change prediction method is provided, and the method is applied to the terminal 102 in fig. 1 for illustration, and includes the following steps:

s402, obtaining target motion characteristics corresponding to the target motion object.

Specifically, the target moving object may be a moving object existing in the surrounding environment of the terminal. The surrounding environment of the terminal may be, for example, an area surrounded by a circle with a fixed radius as a center of the circle. The fixed radius can be set according to the requirement or can be preset. The target motion feature refers to a motion feature of the current time corresponding to the target motion object. The current time is a time period, the ending time of the current time is the current time, the starting time of the current time is the historical time of which the time interval with the current time is the first time interval. The first time interval may be preset or may be determined as needed.

S404, inputting the target motion characteristics into a trained lane change prediction regression model to obtain target lane change probability corresponding to a target motion object; the trained lane change prediction regression model is obtained through training according to a first sample motion feature and a standard lane change probability corresponding to the first sample motion feature, the first sample motion feature is obtained according to motion data of a sample moving object acquired at a sample acquisition time, the standard lane change probability is determined according to a lane change time interval between the object lane change time corresponding to the first sample motion feature and the sample acquisition time, and the standard lane change probability and the lane change time interval form a negative correlation.

Specifically, the trained lane change prediction regression model may be trained by the lane change prediction regression model training method set forth above. The first time interval may coincide with the duration of the sample acquisition time.

S406, determining a target lane change result corresponding to the target moving object according to the target lane change probability.

Specifically, the target lane change result may include any one of a lane change or a no-lane change, and may further include predicted lane change time information. The predicted lane change time information is used to predict the time at which the lane change occurs, and may be within 5 seconds of the future, for example. The target lane change result is that the lane change indicates that the target moving object is about to change lanes, for example, lane change may occur within 5 seconds in the future. The lane change may be denoted by 1 and the constant lane may be denoted by 0.

In some embodiments, the terminal may control the movement of the terminal according to the target lane change result to avoid collision with the target moving object. The terminal may decelerate to travel when, for example, the target lane change result of the moving object on the road in front of the terminal is lane change.

In the lane change prediction method, the target motion feature corresponding to the target motion object is acquired, the target motion feature is input into the trained lane change prediction regression model, the target lane change probability corresponding to the target motion object is obtained, the target lane change result corresponding to the target motion object is determined according to the target lane change probability, and the trained lane change prediction regression model is trained according to the first sample motion feature and the standard lane change probability corresponding to the first sample motion feature, wherein the first sample motion feature is obtained according to the motion data of the sample motion object acquired at the sample acquisition time, the standard lane change probability is determined according to the lane change time interval between the object lane change time corresponding to the first sample motion feature and the sample acquisition time, and the standard lane change probability and the lane change time interval form a negative correlation relationship, so that the accuracy of the trained lane change prediction regression model is improved, and the accuracy of lane change prediction is improved.

In some embodiments, determining a target lane change result corresponding to the target moving object according to the target lane change probability includes: comparing the target lane change probability with a target lane change probability threshold to obtain a comparison result; and determining a target lane change result corresponding to the target moving object according to the comparison result.

Specifically, the target lane change probability threshold may be preset, or may be determined by training a lane change prediction regression model, for example, may be 0.9. The comparison result can be any one of the target lane change probability being larger than the target lane change probability threshold and the target lane change probability being smaller than or equal to the target lane change probability threshold. When the comparison result is that the target lane change probability is larger than the target lane change probability threshold, the terminal can determine that the target lane change result is lane change, otherwise, the terminal can determine that the target lane change result is unchanged.

In the embodiment, the target lane change probability is compared with the target lane change probability threshold to obtain the comparison result, and the target lane change result corresponding to the target moving object is determined according to the comparison result, so that the accuracy of the target lane change result is improved.

It should be understood that, although the steps in the flowcharts of fig. 2-4 are shown in order as indicated by the arrows, these steps are not necessarily performed in order as indicated by the arrows. The steps are not strictly limited to the order of execution unless explicitly recited herein, and the steps may be executed in other orders. Moreover, at least some of the steps in fig. 2-4 may include multiple sub-steps or stages that are not necessarily performed at the same time, but may be performed at different times, nor do the order in which the sub-steps or stages are performed necessarily occur sequentially, but may be performed alternately or alternately with at least a portion of the sub-steps or stages of other steps or steps.

In some embodiments, as shown in fig. 5, there is provided a lane change prediction regression model training apparatus, including: a first sample motion feature acquisition module 502, a standard lane change probability acquisition module 504, a predicted lane change probability acquisition module 506, a model loss value acquisition module 508, and a trained lane change prediction regression model acquisition module 510, wherein:

a first sample motion feature acquiring module 502, configured to acquire a first sample motion feature, where the first sample motion feature is obtained according to motion data of a sample motion object acquired at a sample acquisition time.

The standard lane change probability obtaining module 504 is configured to obtain a standard lane change probability corresponding to the first sample motion feature, where the standard lane change probability is determined according to a lane change time interval between the object lane change time corresponding to the first sample motion feature and the sample acquisition time, and the standard lane change probability and the lane change time interval form a negative correlation.

The predicted lane change probability obtaining module 506 is configured to input the first sample motion feature into a lane change prediction regression model to be trained, and obtain a predicted lane change probability corresponding to the first sample motion feature.

The model loss value obtaining module 508 is configured to obtain a model loss value according to a probability difference between the predicted lane change probability and the standard lane change probability.

The trained lane change prediction regression model obtaining module 510 is configured to adjust model parameters in the lane change prediction regression model by using the model loss value to obtain a trained lane change prediction regression model, so as to perform lane change prediction according to the trained lane change prediction regression model.

In some embodiments, the standard lane change probability acquisition module 504 includes:

and the target lane change probability generation factor acquisition unit is used for acquiring the target lane change probability generation factor when the lane change time interval is smaller than or equal to the target time interval threshold value.

The standard lane change probability obtaining unit is used for calculating according to the lane change time interval and the target lane change probability generating factor to obtain the standard lane change probability corresponding to the first sample motion characteristic.

In some embodiments, the standard lane change probability obtaining unit is further configured to square the lane change time interval to obtain an interval square value; calculating the ratio of the square value of the interval to the target lane change probability generating factor to obtain a target ratio; and taking the first numerical value as a base number, and taking the negative number of the target ratio as an index to perform index calculation to obtain the standard lane change probability corresponding to the first sample motion characteristic.

In some embodiments, the lane change predictive regression model training apparatus further includes a target time interval threshold obtaining module, the target time interval threshold obtaining module including:

And the candidate parameter combination determining unit is used for determining at least two candidate parameter combinations, wherein the candidate parameter combinations comprise a candidate time interval threshold value and a candidate lane change probability generating factor.

The probability tag obtaining unit is used for obtaining the second sample motion feature, determining the lane change probability corresponding to the second sample motion feature according to the candidate parameter combination, and taking the lane change probability as the probability tag corresponding to the second sample motion feature.

The trained lane change prediction regression model obtaining unit is used for training the lane change prediction regression model based on the second sample motion characteristics and probability labels corresponding to the second sample motion characteristics to obtain a trained lane change prediction regression model corresponding to the candidate parameter combination.

The target time interval threshold obtaining unit is used for determining the model accuracy of the trained lane change prediction regression model corresponding to the candidate parameter combination, selecting a target parameter combination meeting accuracy conditions from at least two candidate parameter combinations based on the model accuracy, taking the candidate time interval threshold in the target parameter combination as the target time interval threshold, and taking the candidate lane change probability generating factor in the target parameter combination as the target lane change probability generating factor.

In some embodiments, the first sample motion feature comprises a distance change feature, the lane change predictive regression model training apparatus further comprises a distance change feature determination module comprising:

the target distance sequence acquisition unit is used for acquiring a target distance sequence corresponding to the sample moving object in the sample acquisition time, and the target distance sequences are ordered according to the movement time.

And the distance change trend determining unit is used for determining the distance change trend according to the magnitude relation of the distances in the target distance sequence.

And the distance change feature determining unit is used for determining the distance change feature corresponding to the sample moving object according to the distance change trend.

For specific limitations on the lane change prediction regression model training apparatus, reference may be made to the above limitations on the lane change prediction regression model training method, and no further description is given here. The above-mentioned various modules in the lane change prediction regression model training apparatus may be implemented in whole or in part by software, hardware, and a combination thereof. The above modules may be embedded in hardware or may be independent of a processor in the computer device, or may be stored in software in a memory in the computer device, so that the processor may call and execute operations corresponding to the above modules.

In some embodiments, as shown in fig. 6, there is provided a lane change prediction apparatus, including: a target motion feature acquisition module 602, a target lane change probability obtaining module 604, and a target lane change result determining module 606, wherein:

the target motion feature acquisition module 602 is configured to acquire a target motion feature corresponding to a target motion object.

The target lane change probability obtaining module 604 is configured to input a target motion feature into a trained lane change prediction regression model, so as to obtain a target lane change probability corresponding to a target motion object; the trained lane change prediction regression model is obtained through training according to a first sample motion feature and a standard lane change probability corresponding to the first sample motion feature, the first sample motion feature is obtained according to motion data of a sample moving object acquired at a sample acquisition time, the standard lane change probability is determined according to a lane change time interval between the object lane change time corresponding to the first sample motion feature and the sample acquisition time, and the standard lane change probability and the lane change time interval form a negative correlation.

The target lane change result determining module 606 is configured to determine a target lane change result corresponding to the target moving object according to the target lane change probability.

In some embodiments, the target lane change result determination module 606 includes:

the comparison result obtaining unit is used for comparing the target lane change probability with a target lane change probability threshold value to obtain a comparison result.

And the target lane change result obtaining unit is used for determining a target lane change result corresponding to the target moving object according to the comparison result.

For specific limitations of the lane change prediction apparatus, reference may be made to the above limitation of the lane change prediction method, and no further description is given here. The various modules in the lane change prediction apparatus described above may be implemented in whole or in part by software, hardware, or a combination thereof. The above modules may be embedded in hardware or may be independent of a processor in the computer device, or may be stored in software in a memory in the computer device, so that the processor may call and execute operations corresponding to the above modules.

In some embodiments, a computer device is provided, which may be a server, the internal structure of which may be as shown in fig. 7. The computer device includes a processor, a memory, a network interface, and a database connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, computer readable instructions, and a database. The internal memory provides an environment for the execution of an operating system and computer-readable instructions in a non-volatile storage medium. The database of the computer equipment is used for storing data such as sample motion characteristics, standard lane change probability, predicted lane change probability, model loss value, lane change prediction regression model and the like. The network interface of the computer device is used for communicating with an external terminal through a network connection. The computer readable instructions, when executed by a processor, implement a lane change predictive regression model training method.

In some embodiments, a computer device is provided, which may be a terminal, and the internal structure of which may be as shown in fig. 8. The computer device includes a processor, a memory, a communication interface, a display screen, and an input device connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The communication interface of the computer device is used for carrying out wired or wireless communication with an external terminal, and the wireless mode can be realized through WIFI, an operator network, NFC (near field communication) or other technologies. The computer program is executed by a processor to implement a network communication method. The display screen of the computer equipment can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer equipment can be a touch layer covered on the display screen, can also be keys, a track ball or a touch pad arranged on the shell of the computer equipment, and can also be an external keyboard, a touch pad or a mouse and the like.

It will be appreciated by those skilled in the art that the structures shown in fig. 7 and 8 are block diagrams of only some of the structures associated with the present application and are not intended to limit the computer device to which the present application may be applied, and that a particular computer device may include more or fewer components than shown, or may combine certain components, or have a different arrangement of components.

A computer device comprising a memory and one or more processors, the memory having stored therein computer readable instructions that, when executed by the processors, cause the one or more processors to perform the steps of the lane change predictive regression model training method described above.

A computer device comprising a memory and one or more processors, the memory having stored therein computer readable instructions that, when executed by the processors, cause the one or more processors to perform the steps of the lane change prediction method described above.

One or more computer storage media storing computer-readable instructions that, when executed by one or more processors, cause the one or more processors to perform the steps of the lane change predictive regression model training method described above.

One or more computer storage media storing computer-readable instructions that, when executed by one or more processors, cause the one or more processors to perform the steps of the lane change prediction method described above.

The computer storage medium is a readable storage medium, and the readable storage medium may be nonvolatile or volatile.

Those skilled in the art will appreciate that implementing all or part of the processes of the methods of the embodiments described above may be accomplished by instructing the associated hardware by computer readable instructions stored on a non-transitory computer readable storage medium, which when executed may comprise processes of embodiments of the methods described above. Any reference to memory, storage, database, or other medium used in the various embodiments provided herein may include non-volatile and/or volatile memory. The nonvolatile memory can include Read Only Memory (ROM), programmable ROM (PROM), electrically Programmable ROM (EPROM), electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double Data Rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous Link DRAM (SLDRAM), memory bus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), among others.

The technical features of the above embodiments may be arbitrarily combined, and all possible combinations of the technical features in the above embodiments are not described for brevity of description, however, as long as there is no contradiction between the combinations of the technical features, they should be considered as the scope of the description.

The above examples merely represent a few embodiments of the present application, which are described in more detail and are not to be construed as limiting the scope of the invention. It should be noted that it would be apparent to those skilled in the art that various modifications and improvements could be made without departing from the spirit of the present application, which would be within the scope of the present application. Accordingly, the scope of protection of the present application is to be determined by the claims appended hereto.

Claims

1. A lane-changing predictive regression model training method comprises the following steps:

And adjusting model parameters in the lane change prediction regression model by using the model loss value to obtain a trained lane change prediction regression model, so as to perform lane change prediction according to the trained lane change prediction regression model.

2. The method of claim 1, wherein the obtaining the standard lane change probability corresponding to the first sample motion feature comprises:

when the lane change time interval is smaller than or equal to a target time interval threshold value, a target lane change probability generating factor is obtained; and

And calculating according to the lane change time interval and the target lane change probability generating factor to obtain the standard lane change probability corresponding to the first sample motion characteristic.

3. The method of claim 2, wherein the calculating according to the lane change time interval and the target lane change probability generating factor to obtain the standard lane change probability corresponding to the first sample motion feature comprises:

Square operation is carried out on the lane change time interval to obtain an interval square value;

calculating the ratio of the interval square value to the target lane change probability generation factor to obtain a target ratio; and

And taking the first numerical value as a base number, and taking the negative number of the target ratio as an index to perform index calculation to obtain the standard lane change probability corresponding to the first sample motion characteristic.

4. The method of claim 3, wherein the step of deriving the target time interval threshold and the target lane change probability generation factor comprises:

determining at least two candidate parameter combinations, wherein the candidate parameter combinations comprise candidate time interval thresholds and candidate lane change probability generating factors;

acquiring a second sample motion feature, and determining the probability of a channel corresponding to the second sample motion feature according to the candidate parameter combination, wherein the probability of the channel corresponds to the second sample motion feature and is used as a probability label corresponding to the second sample motion feature;

training the lane change prediction regression model based on the second sample motion feature and a probability label corresponding to the second sample motion feature to obtain a trained lane change prediction regression model corresponding to the candidate parameter combination; and

Determining model accuracy of a trained lane change prediction regression model corresponding to the candidate parameter combination, selecting a target parameter combination meeting accuracy conditions from the at least two candidate parameter combinations based on the model accuracy, taking a candidate time interval threshold in the target parameter combination as the target time interval threshold, and taking a candidate lane change probability generation factor in the target parameter combination as the target lane change probability generation factor.

5. The method of claim 1, wherein the first sample motion feature comprises a distance change feature, the step of determining the distance change feature comprising:

acquiring a target distance sequence corresponding to the sample moving object in the sample acquisition time, wherein the target distance sequence is ordered according to the movement time;

determining a distance change trend according to the size relation of the distances in the target distance sequence; and

And determining the distance change characteristics corresponding to the sample moving object according to the distance change trend.

6. A lane change prediction method, comprising:

inputting the target motion characteristics into a trained lane change prediction regression model to obtain target lane change probability corresponding to the target motion object; the trained lane change prediction regression model is obtained by inputting a first sample motion characteristic into a lane change prediction regression model to be trained to obtain a predicted lane change probability corresponding to the first sample motion characteristic, obtaining a model loss value according to a probability difference between the predicted lane change probability and a standard lane change probability, and adjusting model parameters in the lane change prediction regression model by using the model loss value; the first sample motion characteristic is obtained according to motion data of a sample motion object acquired at a sample acquisition time, the standard lane change probability is determined according to a lane change time interval between the object lane change time corresponding to the first sample motion characteristic and the sample acquisition time, and the standard lane change probability and the lane change time interval form a negative correlation; and

7. The method of claim 6, wherein the determining, according to the target lane change probability, a target lane change result corresponding to the target moving object comprises:

comparing the target lane change probability with a target lane change probability threshold to obtain a comparison result; and

And determining a target lane change result corresponding to the target moving object according to the comparison result.

8. A lane-changing predictive regression model training apparatus comprising:

And the trained lane change prediction regression model obtaining module is used for utilizing the model loss value to adjust model parameters in the lane change prediction regression model to obtain a trained lane change prediction regression model so as to conduct lane change prediction according to the trained lane change prediction regression model.

9. A lane change prediction apparatus comprising:

the target lane change probability obtaining module is used for inputting the target motion characteristics into a trained lane change prediction regression model to obtain target lane change probability corresponding to the target motion object; the trained lane change prediction regression model is obtained by inputting a first sample motion characteristic into a lane change prediction regression model to be trained to obtain a predicted lane change probability corresponding to the first sample motion characteristic, obtaining a model loss value according to a probability difference between the predicted lane change probability and a standard lane change probability, and adjusting model parameters in the lane change prediction regression model by using the model loss value; the first sample motion characteristic is obtained according to motion data of a sample motion object acquired at a sample acquisition time, the standard lane change probability is determined according to a lane change time interval between the object lane change time corresponding to the first sample motion characteristic and the sample acquisition time, and the standard lane change probability and the lane change time interval form a negative correlation; and

10. A computer device comprising a memory and one or more processors, the memory having stored therein computer-readable instructions that, when executed by the one or more processors, cause the one or more processors to perform the steps of the method of any of claims 1 to 5 or 6 to 7.