Pedestrian traffic behavior prediction method, device, computer equipment, readable storage medium and program product

Abstract

The present application relates to a method, device, computer equipment, computer-readable storage medium and computer program product for predicting pedestrian traffic behavior. The method includes: based on traffic scene images, determining the traffic scene information and interference source identification results corresponding to sample pedestrians; the interference source identification results include interference source types and interference source dynamics information; based on pedestrian dynamic data, interference source dynamics information, and pedestrian psychological data, determining the traffic behavior evaluation results of sample pedestrians; based on the pedestrian static data and traffic behavior evaluation results of each sample pedestrian, and the traffic scene information and interference source identification results corresponding to each sample pedestrian, constructing a target training set; based on the target training set, training the target neural network to obtain a pedestrian traffic behavior prediction model; the pedestrian traffic behavior prediction model is used to predict pedestrian traffic behavior. The use of this method can achieve accurate prediction of pedestrian traffic behavior.

Description

Technical Field

[0001] The present application relates to the field of autonomous driving technology, and in particular to a method, apparatus, computer equipment, computer-readable storage medium, and computer program product for predicting pedestrian traffic behavior. Background Art

[0002] Self-driving cars are one of the most eye-catching fields in today's science and technology. With the continuous advancement of technology, self-driving cars are gradually becoming the mainstream of future transportation. However, pedestrian recognition technology for self-driving cars still faces many challenges and development opportunities. Among them, the development of pedestrian recognition and prediction technology is crucial to the safety of self-driving cars. Self-driving cars use sensors or cameras to perceive the trajectory, movement, posture and other characteristics of pedestrians on the roadside, and combine the self-driving vehicle's own dynamic information (such as speed, acceleration, etc.) to realize the recognition of specific pedestrians' intention to cross the street and trajectory prediction, so that self-driving vehicles can better avoid pedestrians and thus improve road safety.

[0003] Therefore, there is an urgent need for a pedestrian traffic behavior prediction method that can accurately predict pedestrian traffic behavior. Summary of the invention

[0004] Based on this, it is necessary to provide a pedestrian traffic behavior prediction method, device, computer equipment, computer-readable storage medium and computer program product that can accurately predict pedestrian traffic behavior in response to the above technical problems.

[0005] In a first aspect, the present application provides a method for predicting pedestrian traffic behavior, comprising:

[0006] Acquire pedestrian dynamic data and pedestrian psychological data of a sample pedestrian in a sample traffic scene, a traffic scene image of the sample traffic scene, and pedestrian static data of the sample pedestrian;

[0007] Based on the traffic scene image, determine the traffic scene information and interference source identification result corresponding to the sample pedestrian; the interference source identification result includes the interference source type and interference source dynamics information;

[0008] Determining a traffic behavior evaluation result of the sample pedestrian based on the pedestrian dynamic data, the interference source dynamic information, and the pedestrian psychological data;

[0009] Constructing a target training set based on the pedestrian static data and traffic behavior evaluation results of each of the sample pedestrians, as well as the traffic scene information and interference source identification results corresponding to each of the sample pedestrians;

[0010] Based on the target training set, a target neural network is trained to obtain a pedestrian traffic behavior prediction model; the pedestrian traffic behavior prediction model is used to perform pedestrian traffic behavior prediction.

[0011] In one embodiment, the pedestrian psychological data includes a crossing strategy and a strategy decision time, the traffic behavior evaluation result includes a traffic behavior type, and determining the traffic behavior evaluation result of the sample pedestrian based on the pedestrian dynamic data, the interference source dynamics information, and the pedestrian psychological data includes:

[0012] In the mapping relationship between the preset crossing strategy and the traffic behavior evaluation starting value, querying the traffic behavior evaluation starting value corresponding to the sample pedestrian according to the crossing strategy of the sample pedestrian;

[0013] Determine the traffic behavior evaluation floating value corresponding to the sample pedestrian based on the pedestrian dynamic data, the strategy decision time and the interference source dynamic information;

[0014] Adding the traffic behavior evaluation starting value and the traffic behavior evaluation floating value to obtain the traffic behavior evaluation value corresponding to the sample pedestrian;

[0015] In the preset mapping relationship between the traffic behavior evaluation value and the traffic behavior type, the traffic behavior type corresponding to the sample pedestrian is queried according to the traffic behavior evaluation value corresponding to the sample pedestrian.

[0016] In one embodiment, determining the traffic behavior evaluation floating value corresponding to the sample pedestrian based on the pedestrian dynamic data, the strategy decision time and the interference source dynamic information includes:

[0017] Determining an evaluation index value of a traffic behavior evaluation index of the sample pedestrian based on the pedestrian dynamic data, the strategy decision time and the interference source dynamic information;

[0018] In the mapping relationship between the preset road crossing strategy and the traffic behavior evaluation strategy, query the traffic behavior evaluation strategy corresponding to the sample pedestrian according to the road crossing strategy of the sample pedestrian;

[0019] The traffic behavior evaluation floating value corresponding to the sample pedestrian is determined according to the evaluation index value and the traffic behavior evaluation strategy corresponding to the sample pedestrian.

[0020] In one embodiment, the traffic behavior evaluation index includes the relative distance between a person and a vehicle and / or the time of conflict occurrence, and determining the traffic behavior evaluation floating value corresponding to the sample pedestrian according to the evaluation index value and the traffic behavior evaluation strategy corresponding to the sample pedestrian includes:

[0021] Determining the target group type of the sample pedestrian according to the pedestrian static data of the sample pedestrian;

[0022] Calculating the evaluation average of the traffic behavior evaluation index corresponding to the target group type according to the evaluation index value of the traffic behavior evaluation index of each sample pedestrian whose group type is the target group type;

[0023] In the preset mapping relationship among the traffic behavior evaluation index, the jaywalking strategy and the threshold multiple, querying the target threshold multiple corresponding to the sample pedestrian according to the traffic behavior evaluation index and the jaywalking strategy of the sample pedestrian;

[0024] The product of the evaluation average value and the target threshold multiple is used as the target threshold corresponding to the sample pedestrian;

[0025] According to the comparison result of the evaluation index value and the target threshold value, the traffic behavior evaluation floating value corresponding to the sample pedestrian is determined.

[0026] In one embodiment, constructing a target training set based on the pedestrian static data and traffic behavior evaluation results of each of the sample pedestrians, and the traffic scene information and interference source identification results corresponding to each of the sample pedestrians, includes:

[0027] For each group type, determining group characteristic information corresponding to the group type according to pedestrian dynamic data and pedestrian psychological data of each sample pedestrian of the group type;

[0028] A target training set is constructed based on the pedestrian static data and traffic behavior evaluation results of each of the sample pedestrians, the traffic scene information and interference source identification results corresponding to each of the sample pedestrians, and the group characteristic information corresponding to each of the group types.

[0029] In one embodiment, constructing a target training set based on the pedestrian static data and traffic behavior evaluation results of each of the sample pedestrians, and the traffic scene information and interference source identification results corresponding to each of the sample pedestrians, includes:

[0030] Based on the pedestrian dynamic data of the sample pedestrian, determining the traffic behavior trajectory diagram, speed time series diagram and joint angle change diagram of the sample pedestrian;

[0031] A target training set is constructed based on the pedestrian static data, traffic behavior evaluation results, traffic behavior trajectory diagram, speed time series diagram and joint angle change diagram of each of the sample pedestrians, as well as the traffic scene information and interference source identification results corresponding to each of the sample pedestrians.

[0032] In a second aspect, the present application also provides a pedestrian traffic behavior prediction device, comprising:

[0033] An acquisition module, used to acquire pedestrian dynamic data and pedestrian psychological data of a sample pedestrian in a sample traffic scene, a traffic scene image of the sample traffic scene, and pedestrian static data of the sample pedestrian;

[0034] A first determination module is used to determine the traffic scene information and interference source identification result corresponding to the sample pedestrian based on the traffic scene image; the interference source identification result includes the interference source type and interference source dynamics information;

[0035] A second determination module, configured to determine a traffic behavior evaluation result of the sample pedestrian based on the pedestrian dynamic data, the interference source dynamic information, and the pedestrian psychological data;

[0036] A construction module, used to construct a target training set based on the pedestrian static data and traffic behavior evaluation results of each of the sample pedestrians, as well as the traffic scene information and interference source identification results corresponding to each of the sample pedestrians;

[0037] The training module is used to train the target neural network based on the target training set to obtain a pedestrian traffic behavior prediction model; the pedestrian traffic behavior prediction model is used to perform pedestrian traffic behavior prediction.

[0038] In one embodiment, the pedestrian psychological data includes a road crossing strategy and a strategy decision time, the traffic behavior evaluation result includes a traffic behavior type, and the second determination module is specifically used to:

[0039] In the mapping relationship between the preset crossing strategy and the traffic behavior evaluation starting value, querying the traffic behavior evaluation starting value corresponding to the sample pedestrian according to the crossing strategy of the sample pedestrian;

[0040] Determine the traffic behavior evaluation floating value corresponding to the sample pedestrian based on the pedestrian dynamic data, the strategy decision time and the interference source dynamic information;

[0041] Adding the traffic behavior evaluation starting value and the traffic behavior evaluation floating value to obtain the traffic behavior evaluation value corresponding to the sample pedestrian;

[0042] In the preset mapping relationship between the traffic behavior evaluation value and the traffic behavior type, the traffic behavior type corresponding to the sample pedestrian is queried according to the traffic behavior evaluation value corresponding to the sample pedestrian.

[0043] In one embodiment, the second determining module is specifically configured to:

[0044] Determining an evaluation index value of a traffic behavior evaluation index of the sample pedestrian based on the pedestrian dynamic data, the strategy decision time and the interference source dynamic information;

[0045] In the mapping relationship between the preset road crossing strategy and the traffic behavior evaluation strategy, query the traffic behavior evaluation strategy corresponding to the sample pedestrian according to the road crossing strategy of the sample pedestrian;

[0046] The traffic behavior evaluation floating value corresponding to the sample pedestrian is determined according to the evaluation index value and the traffic behavior evaluation strategy corresponding to the sample pedestrian.

[0047] In one embodiment, the traffic behavior evaluation index includes the relative distance between the person and the vehicle and / or the time when the conflict occurs, and the second determination module is used to:

[0048] Determining the target group type of the sample pedestrian according to the pedestrian static data of the sample pedestrian;

[0049] Calculating the evaluation average of the traffic behavior evaluation index corresponding to the target group type according to the evaluation index value of the traffic behavior evaluation index of each sample pedestrian whose group type is the target group type;

[0050] In the preset mapping relationship among the traffic behavior evaluation index, the jaywalking strategy and the threshold multiple, querying the target threshold multiple corresponding to the sample pedestrian according to the traffic behavior evaluation index and the jaywalking strategy of the sample pedestrian;

[0051] The product of the evaluation average value and the target threshold multiple is used as the target threshold corresponding to the sample pedestrian;

[0052] According to the comparison result of the evaluation index value and the target threshold value, the traffic behavior evaluation floating value corresponding to the sample pedestrian is determined.

[0053] In one embodiment, the building block is specifically used for:

[0054] For each group type, determining group characteristic information corresponding to the group type according to pedestrian dynamic data and pedestrian psychological data of each sample pedestrian of the group type;

[0055] A target training set is constructed based on the pedestrian static data and traffic behavior evaluation results of each of the sample pedestrians, the traffic scene information and interference source identification results corresponding to each of the sample pedestrians, and the group characteristic information corresponding to each of the group types.

[0056] In one embodiment, the building block is specifically used for:

[0057] Based on the pedestrian dynamic data of the sample pedestrian, determining the traffic behavior trajectory diagram, speed time series diagram and joint angle change diagram of the sample pedestrian;

[0058] A target training set is constructed based on the pedestrian static data, traffic behavior evaluation results, traffic behavior trajectory diagram, speed time series diagram and joint angle change diagram of each of the sample pedestrians, as well as the traffic scene information and interference source identification results corresponding to each of the sample pedestrians.

[0059] In a third aspect, the present application further provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps described in the first aspect when executing the computer program.

[0060] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps described in the first aspect above.

[0061] In a fifth aspect, the present application further provides a computer program product, comprising a computer program, which implements the steps described in the first aspect when executed by a processor.

[0062] The above-mentioned pedestrian traffic behavior prediction method, device, computer equipment, computer-readable storage medium and computer program product obtain the pedestrian dynamic data and pedestrian psychological data of the sample pedestrian in the sample traffic scene, the traffic scene image of the sample traffic scene, and the pedestrian static data of the sample pedestrian; based on the traffic scene image, determine the traffic scene information and interference source identification result corresponding to the sample pedestrian; the interference source identification result includes the interference source type and interference source dynamic information; based on the pedestrian dynamic data, the interference source dynamic information, and the pedestrian psychological data, determine the traffic behavior evaluation result of the sample pedestrian; based on the pedestrian static data and traffic behavior evaluation result of each of the sample pedestrians, and the traffic scene information and interference source identification result corresponding to each of the sample pedestrians, construct a target training set; based on the target training set, train the target neural network to obtain a pedestrian traffic behavior prediction model; the pedestrian traffic behavior prediction model is used to perform pedestrian traffic behavior prediction. In this way, static data, dynamic data, psychological data and traffic scene images of sample pedestrians are collected when they automatically complete designated traffic behaviors such as crossing the road under actual road traffic conditions, and traffic scene images are identified to obtain traffic scene information and interference source identification results. The traffic behavior evaluation results of the sample pedestrians are determined based on the above information, and a target training set for training a pedestrian traffic behavior prediction model is constructed based on the traffic behavior evaluation results of each sample pedestrian and the above information. The trained pedestrian traffic behavior prediction model is used to predict pedestrian traffic behavior. Considering the different static information and action abilities of pedestrians, and the huge differences in the movement behaviors of each person in a complex road environment, the constructed training set is a pedestrian traffic behavior data set of different types of pedestrians, different traffic behaviors, and different traffic conflicts. It is rich and diverse, providing comprehensive and reliable technical support and data sources for optimizing pedestrian traffic behavior prediction technology, and can achieve accurate prediction of pedestrian traffic behavior. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related technologies, the drawings required for use in the embodiments of the present application or the related technical descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0064] Figure 1 A flowchart of a method for predicting pedestrian traffic behavior according to an embodiment of the present invention;

[0065] Figure 2 A schematic diagram of a flow chart of the steps of determining a sample pedestrian's traffic behavior evaluation result in one embodiment;

[0066] Figure 3A schematic diagram of a flow chart of a step of determining a floating value of a traffic behavior evaluation corresponding to a sample pedestrian in one embodiment;

[0067] Figure 4 A flow chart of a step of determining a floating value of a traffic behavior evaluation corresponding to a sample pedestrian according to an evaluation index value and a traffic behavior evaluation strategy corresponding to the sample pedestrian in an embodiment;

[0068] Figure 5 A schematic diagram of a process for constructing a target training set in one embodiment;

[0069] Figure 6 A schematic diagram of a process for constructing a target training set in another embodiment;

[0070] Figure 7 is a structural block diagram of a pedestrian traffic behavior prediction device in one embodiment;

[0071] Figure 8 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0072] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0073] In one embodiment, Figure 1 As shown, a method for predicting pedestrian traffic behavior is provided. This embodiment uses the method applied to a terminal as an example for illustration. It can be understood that the method can also be applied to a server, and can also be applied to a system including a terminal and a server, and is implemented through the interaction between the terminal and the server. Among them, the terminal can be but is not limited to various personal computers, laptops, smart phones, tablet computers, Internet of Things devices and portable wearable devices. The Internet of Things devices can be smart speakers, smart TVs, smart air conditioners, smart car-mounted devices, projection devices, etc. Portable wearable devices can be smart watches, smart bracelets, head-mounted devices, etc. The head-mounted device can be a virtual reality (VR) device, an augmented reality (AR) device, smart glasses, etc. The server can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides cloud computing services. In this embodiment, the method includes the following steps:

[0074] Step 101, obtaining pedestrian dynamic data and pedestrian psychological data of a sample pedestrian in a sample traffic scene, a traffic scene image of the sample traffic scene, and pedestrian static data of the sample pedestrian.

[0075] In an embodiment of the present application, the sample pedestrian is a pedestrian in a sample used to train a pedestrian traffic behavior prediction model. Different sample pedestrians correspond to different sample numbers. The sample traffic scene is a traffic scene in a sample used to train a pedestrian traffic behavior prediction model. The traffic scene is a scene of the entire process of a pedestrian's traffic behavior, including the traffic environment and the movement scene of each traffic target in the traffic environment. Traffic targets include sample pedestrians and interference sources. The traffic scene image is an image of the entire process of a pedestrian's traffic behavior, which may be a video and / or a picture. Pedestrian static data is data of the pedestrian itself that is not affected by the pedestrian's traffic behavior. Pedestrian static data information may include one or more of age, gender, weight, height, and arm span. Pedestrian dynamic data is the pedestrian's dynamic information, which may include dynamic information of the pedestrian's body parts, and may also include dynamic information of the pedestrian's main joints. The pedestrian dynamic data information may include one or more of the head rotation angle, the left hip flexion and extension angle, the right hip flexion and extension angle, the left knee flexion and extension angle, the right knee flexion and extension angle, the left ankle flexion and extension angle, the right ankle flexion and extension angle, the left shoulder flexion and extension angle, the right shoulder flexion and extension angle, the left shoulder adduction / abduction angle, the right shoulder adduction / abduction angle, the left shoulder internal rotation / external rotation angle, the right shoulder internal rotation / external rotation angle, the left elbow flexion and extension angle, the right elbow flexion and extension angle, the left elbow adduction / abduction angle, the right elbow adduction / abduction angle, the left elbow internal rotation / external rotation angle, the right elbow internal rotation / external rotation angle, the hip displacement speed (Z direction component), the hip displacement speed (X direction component), the hip relative position (Z direction component) and the hip relative position (X direction component). The pedestrian psychological data is the pedestrian's judgment of danger and decision-making information on whether to cross the road during the process of crossing the road. The pedestrian psychological data may include crossing the road strategy. The crossing strategy is the decision-making information of whether and how the pedestrian crosses the road during the crossing process. The crossing strategy can include normal crossing, rushing through, and yielding through.

[0076] The terminal obtains the static data of the sample pedestrian through the static data collector. The terminal obtains the dynamic data of the sample pedestrian in the sample traffic scene through the human motion capture device. The terminal obtains the psychological data of the sample pedestrian in the sample traffic scene through the psychological data collector. The terminal obtains the traffic scene image of the sample pedestrian in the sample traffic scene through the high-definition camera. Among them, the static data collector is a device that can collect static data of pedestrians. The sample traffic scene can be the process of sample pedestrians completing traffic behaviors such as crossing the road and walking along the side of the road under their own consciousness in the actual road traffic scene. The human motion capture device can be worn by pedestrians to collect the dynamic information of various parts of the pedestrian's body, and can also be used to collect the dynamic information of the main joints of the pedestrian. In this way, the use of human motion capture equipment to collect the movement change rules of the main joints of pedestrians can improve the accuracy of data collection, can depict the details of small movement changes, and can improve the accuracy of pedestrian traffic behavior prediction. The psychological data collector can be a device that can be held in the hands of pedestrians, and is used to record the sample pedestrians' first-hand crossing strategies based on their observation and judgment of road conditions and moving vehicles, the danger of crossing the road, and the movement method of crossing the road during the crossing process. It is also used to record the time when the crossing strategy is made, that is, the strategy decision time. The psychological data collector can have buttons corresponding to different crossing strategies, and the sample pedestrians record the crossing strategies they make by pressing buttons. High-definition cameras can be set up next to the data collection section to obtain traffic scene images of the sample pedestrians completing the entire data collection process.

[0077] In one example, the terminal obtains pedestrian dynamic data and pedestrian psychological data of a sample pedestrian in a sample traffic scene, a traffic scene image of the sample traffic scene, and pedestrian static data of the sample pedestrian from a pre-built pedestrian traffic behavior database.

[0078] In one example, the terminal assigns a unique and independent sample pedestrian number, such as P0001, to each collected object (sample pedestrian). Then, the terminal establishes a sample pedestrian data set based on the sample pedestrian number. Then, the terminal creates four data forms under each sample pedestrian data set, namely, a pedestrian static data table, a pedestrian dynamic data table, a pedestrian psychological data table, and a traffic scene data table. Then, the terminal imports the pedestrian dynamic data and pedestrian psychological data of the sample pedestrian in the sample traffic scene, the traffic scene image of the sample traffic scene, and the pedestrian static data of the sample pedestrian obtained by the static data collector, the human motion capture device, the psychological data collector and the high-definition camera into each data form through the communication protocol and the data interface protocol. Then, the terminal stores each data form in the pedestrian traffic behavior database to obtain a constructed pedestrian traffic behavior database.

[0079] Step 102: Determine the traffic scene information and interference source identification result corresponding to the sample pedestrian based on the traffic scene image.

[0080] The interference source identification result includes the interference source type and interference source dynamics information.

[0081] In an embodiment of the present application, the terminal uses a deep learning detection algorithm to determine the traffic scene information corresponding to the sample pedestrian based on the traffic scene image. At the same time, the terminal determines the interference source identification result corresponding to the sample pedestrian based on the traffic scene image and the preset target detection algorithm. Among them, the traffic scene information may include weather information, time information, and one or more of the section type and section width of the data collection section. Weather information includes but is not limited to sunny days, rainy days, snowy days and foggy (haze) days. Time information includes but is not limited to daytime, night (with street lights) and night (without street lights). Section types include but are not limited to expressways, trunk roads, secondary trunk roads, branch roads, sections without traffic lights and sections with traffic lights. The target detection algorithm can be YOLO v7. The interference source is a traffic target that has a traffic conflict with the sample pedestrian in the sample traffic scene. The interference source dynamics information may include one or more of the interference source motion trajectory, direction, speed, and relative distance parameters from the sample pedestrian. The interference source type can be represented by the interference source type number. For example, the interference source type and numbering rules are shown in Table 1. If no interference source is encountered during the process of the sample pedestrian completing the traffic behavior data collection, that is, the interference source identification result indicates that no interference source is detected, then the interference source type number is 00.

[0082] Table 1 Interference source types and numbering rules

[0083]

[0084] Step 103, determining the traffic behavior evaluation result of the sample pedestrian based on the pedestrian dynamic data, the interference source dynamic information, and the pedestrian psychological data.

[0085] In the embodiment of the present application, the traffic behavior evaluation result is used to characterize the traffic behavior characteristics of the sample pedestrians.

[0086] Step 104 , constructing a target training set based on the pedestrian static data and traffic behavior evaluation results of each sample pedestrian, as well as the traffic scene information and interference source identification results corresponding to each sample pedestrian.

[0087] In an embodiment of the present application, the terminal forms a target training set with the pedestrian static data and traffic behavior evaluation results of each sample pedestrian, as well as the traffic scene information and interference source identification results corresponding to each sample pedestrian. Specifically, for each sample pedestrian, the terminal forms the sample data corresponding to the sample pedestrian with the pedestrian static data and traffic behavior evaluation results of the sample pedestrian, as well as the traffic scene information and interference source identification results corresponding to the sample pedestrian. Then, the terminal forms a target training set with the sample data corresponding to each sample pedestrian.

[0088] Step 105: Based on the target training set, the target neural network is trained to obtain a pedestrian traffic behavior prediction model.

[0089] Among them, the pedestrian traffic behavior prediction model is used to predict pedestrian traffic behavior.

[0090] In the above-mentioned pedestrian traffic behavior prediction method, pedestrian static data, pedestrian dynamic data, pedestrian psychological data and traffic scene images are collected when sample pedestrians automatically and consciously complete designated traffic behaviors such as crossing the road under actual road traffic conditions, traffic scene images are identified, and traffic scene information and interference source identification results are obtained. The traffic behavior evaluation results of the sample pedestrians are determined based on the above-mentioned information, and a target training set for training a pedestrian traffic behavior prediction model is constructed based on the traffic behavior evaluation results of each sample pedestrian and the above-mentioned information. The trained pedestrian traffic behavior prediction model is used to predict pedestrian traffic behavior. Considering that pedestrian static information and action abilities are different, and each person's movement behavior in a complex road environment is very different, the constructed training set is a pedestrian traffic behavior data set of different types of pedestrians, different traffic behaviors, and different traffic conflicts. It is rich and diverse, and provides comprehensive and reliable technical support and data sources for optimizing pedestrian traffic behavior prediction technology, and can achieve accurate prediction of pedestrian traffic behavior. Moreover, this method more realistically reflects the behavior of pedestrians in real road environments, their first reactions when traffic conflicts occur, and the interaction between the traffic behaviors of pedestrians and other traffic participants on the road. The proposed data collection items are also important indicators for analyzing and studying the characteristics, changing patterns, and correlation between pedestrian and vehicle traffic conflicts of pedestrian traffic behaviors. They are more in line with actual conditions and can improve the accuracy of pedestrian traffic behavior prediction.

[0091] In an exemplary embodiment, Figure 2 As shown, the pedestrian psychological data includes the crossing strategy and the strategy decision time, the traffic behavior evaluation result includes the traffic behavior type, and step 103 includes steps 201 to 204. Among them:

[0092] Step 201 : In the mapping relationship between the preset jaywalking strategy and the traffic behavior evaluation starting value, according to the jaywalking strategy of the sample pedestrian, the traffic behavior evaluation starting value corresponding to the sample pedestrian is queried.

[0093] In the embodiment of the present application, the strategies for crossing the road include normal passing, rushing through, and letting through. The traffic behavior evaluation starting value corresponding to letting through is less than the traffic behavior evaluation starting value corresponding to normal passing, and the traffic behavior evaluation starting value corresponding to normal passing is less than the traffic behavior evaluation starting value corresponding to rushing through, that is, the traffic behavior evaluation starting values ​​corresponding to letting through, normal passing, and rushing through are increased. For example, the traffic behavior evaluation starting values ​​corresponding to letting through, normal passing, and rushing through can be 4, 5, and 6, respectively.

[0094] Step 202, based on the pedestrian dynamic data, strategy decision time and interference source dynamic information, determine the traffic behavior evaluation floating value corresponding to the sample pedestrian.

[0095] In the embodiment of the present application, the traffic behavior evaluation floating value includes positive numbers, negative numbers and 0.

[0096] Step 203 , adding the traffic behavior evaluation starting value and the traffic behavior evaluation floating value to obtain the traffic behavior evaluation value corresponding to the sample pedestrian.

[0097] In the embodiment of the present application, the traffic behavior evaluation value is used to characterize the traffic behavior characteristics of the sample pedestrians.

[0098] Step 204 , in the preset mapping relationship between traffic behavior evaluation values ​​and traffic behavior types, query the traffic behavior type corresponding to the sample pedestrian according to the traffic behavior evaluation value corresponding to the sample pedestrian.

[0099] In the embodiment of the present application, the traffic behavior types may include cautious, mature and impatient. The traffic behavior evaluation values ​​corresponding to the cautious, mature and impatient types are increased.

[0100] In the above pedestrian traffic behavior prediction method, different traffic behavior evaluation starting values ​​corresponding to different sample pedestrians are determined through different crossing strategies, and the traffic behavior evaluation floating value corresponding to the sample pedestrian is determined through the actual traffic scenes and various actual actions faced by the sample pedestrian when crossing the road. The sum of the traffic behavior evaluation starting value and the traffic behavior evaluation floating value is used as the traffic behavior evaluation value corresponding to the sample pedestrian, and then the traffic behavior type corresponding to the traffic behavior evaluation value is queried to obtain the traffic behavior evaluation result of the sample pedestrian. In this way, the pedestrian's crossing strategy, the actual traffic scene faced, and the actual action of crossing the road are comprehensively considered to comprehensively evaluate his traffic behavior, which is more in line with the actual situation, more comprehensive and accurate, and can improve the accuracy of the target training set, thereby improving the accuracy of pedestrian traffic behavior prediction.

[0101] In an exemplary embodiment, Figure 3 As shown, step 202 includes steps 301 to 303. Among them:

[0102] Step 301 : determining the evaluation index value of the traffic behavior evaluation index of the sample pedestrian based on the pedestrian dynamic data, the strategy decision time and the interference source dynamic information.

[0103] In the embodiment of the present application, the traffic behavior evaluation index may be one or more. The traffic behavior evaluation index includes head turn observation information, the number of waving gestures, the relative distance between the person and the vehicle determined by the strategy, and the time of conflict occurrence. The head turn observation information includes the number of head turn observations and the head turn observation duration. The number of head turn observations is the frequency (unit: times) of the sample pedestrian turning his head to observe when crossing the road, that is, F h The head turning observation duration is the duration (in seconds) that the sample pedestrian turns his head to observe while crossing the road, which can be the longest duration, that is, T h The head turning observation refers to the absolute value of the head rotation angle being greater than the preset angle threshold. For example, the angle threshold is 20°. The number of waving gestures refers to the frequency (in times) of the sample pedestrian raising his hand or waving his hand when crossing the road, that is, F d If there is no hand raising / waving gesture, F d The relative distance between pedestrians and vehicles determined by the strategy is the relative distance between pedestrians and vehicles at the time when the sample pedestrian makes the strategy of crossing the road (i.e., the strategy decision time) (unit: meter), i.e., S t The conflict time is the predicted time after which a collision between a person and a vehicle may occur (in seconds), i.e., T d .

[0104] Based on the dynamic data of the pedestrian, the terminal determines the evaluation index value of the sample pedestrian's head-turning observation information and the evaluation index value of the number of waving gestures. Specifically, the terminal determines the evaluation index value of the sample pedestrian's head-turning observation information based on the head rotation angle. Then, the terminal determines the evaluation index value of the sample pedestrian's number of waving gestures based on the left shoulder joint flexion and extension angle, the right shoulder joint flexion and extension angle, the left shoulder joint adduction / abduction angle, the right shoulder joint adduction / abduction angle, the left shoulder joint internal rotation / external rotation angle, the right shoulder joint internal rotation / external rotation angle, the left elbow joint flexion and extension angle, the right elbow joint flexion and extension angle, the left elbow joint adduction / abduction angle, the right elbow joint adduction / abduction angle, the left elbow joint internal rotation / external rotation angle, and the right elbow joint internal rotation / external rotation angle.

[0105] The terminal determines the evaluation index value of the strategy-determined relative distance between the pedestrian and the vehicle based on the strategy-determined time and the interference source dynamics information. Then, the terminal calculates the evaluation index value of the conflict occurrence time of the sample pedestrian based on the evaluation index value of the strategy-determined relative distance between the pedestrian and the vehicle and the interference source dynamics information. Specifically, the terminal calculates the evaluation index value of the strategy-determined relative distance between the pedestrian and the vehicle S t The speed V of the interference source t The ratio of is taken as the conflict occurrence time T of the sample pedestrians. dThe evaluation index value of .

[0106] Step 302 , in the preset mapping relationship between the jaywalking strategy and the traffic behavior evaluation strategy, query the traffic behavior evaluation strategy corresponding to the sample pedestrian according to the jaywalking strategy of the sample pedestrian.

[0107] Step 303: Determine the traffic behavior evaluation floating value corresponding to the sample pedestrian according to the evaluation index value and the traffic behavior evaluation strategy corresponding to the sample pedestrian.

[0108] In an embodiment of the present application, different crossing strategies correspond to different traffic behavior evaluation strategies. The traffic behavior evaluation strategy includes a traffic behavior evaluation threshold. The terminal compares the evaluation index value of the traffic behavior evaluation index of the sample pedestrian with the traffic behavior evaluation threshold corresponding to the traffic behavior evaluation index to obtain a comparison result of the traffic behavior evaluation index. Then, the terminal queries the traffic behavior evaluation floating value corresponding to the sample pedestrian according to the comparison results of the traffic behavior evaluation indicators corresponding to the sample pedestrian in the mapping relationship between the comparison results of the preset traffic behavior evaluation indicators and the traffic behavior evaluation floating value contained in the traffic behavior evaluation strategy corresponding to the sample pedestrian.

[0109] In the above pedestrian traffic behavior prediction method, the evaluation index value of the sample pedestrian's traffic behavior evaluation index is determined based on the pedestrian's dynamic data, strategy decision time and interference source dynamics information; in the preset mapping relationship between the crossing strategy and the traffic behavior evaluation strategy, the traffic behavior evaluation strategy corresponding to the sample pedestrian is queried according to the crossing strategy of the sample pedestrian; according to the evaluation index value and the traffic behavior evaluation strategy corresponding to the sample pedestrian, the traffic behavior evaluation floating value corresponding to the sample pedestrian is determined. In this way, the evaluation index value of the sample pedestrian's traffic behavior evaluation index is calculated, and different traffic behavior evaluation strategies are adopted for sample pedestrians who adopt different crossing strategies, and the traffic behavior evaluation floating value corresponding to the sample pedestrian is comprehensively determined, which is more in line with the actual situation, and its traffic behavior is evaluated more comprehensively and accurately, which can improve the accuracy of the target training set, and thus improve the accuracy of pedestrian traffic behavior prediction.

[0110] In an exemplary embodiment, Figure 4 As shown, the traffic behavior evaluation index includes the relative distance between people and vehicles and / or the time when the conflict occurs, and step 303 includes steps 401 to 405. Among them:

[0111] Step 401: Determine the target group type of the sample pedestrians according to the static data of the sample pedestrians.

[0112] In the embodiment of the present application, the terminal queries the target group type of the sample pedestrian according to the gender and age of the sample pedestrian in the preset mapping relationship between gender, age and group type. The target group type is the group type of the sample pedestrian. For example, the mapping relationship between the gender, age and group type of the sample pedestrian is shown in Table 2.

[0113] Table 2 Mapping relationship between gender, age and group type of sample pedestrians

[0114]

[0115] Step 402 , based on the evaluation index values ​​of the traffic behavior evaluation indexes of the sample pedestrians whose group type is the target group type, the evaluation average value of the traffic behavior evaluation index corresponding to the target group type is calculated.

[0116] In an embodiment of the present application, for each traffic behavior evaluation index, the terminal uses the ratio of the sum of the evaluation index values ​​of the traffic behavior evaluation index of each sample pedestrian whose group type is the target group type to the sum of the evaluation index values ​​of each sample pedestrian of the target group type as the evaluation average value of the traffic behavior evaluation index corresponding to the target group type.

[0117] Step 403, in the preset mapping relationship among traffic behavior evaluation index, jaywalking strategy and threshold multiple, query the target threshold multiple corresponding to the sample pedestrian according to the traffic behavior evaluation index and jaywalking strategy of the sample pedestrian.

[0118] In the embodiment of the present application, for the same traffic behavior evaluation index, the threshold multiples corresponding to yielding, normal passing and rushing through are increased. The target threshold multiple is a multiple of the reference value of the traffic behavior evaluation threshold corresponding to the sample pedestrian. The traffic behavior evaluation threshold can be one or more, and accordingly, the target threshold multiple can be one or more.

[0119] Step 404: The product of the evaluation average value and the target threshold multiple is used as the target threshold corresponding to the sample pedestrian.

[0120] In the embodiment of the present application, the target threshold is the traffic behavior evaluation threshold corresponding to the sample pedestrian.

[0121] Step 405: Determine the traffic behavior evaluation floating value corresponding to the sample pedestrian according to the comparison result between the evaluation index value and the target threshold value.

[0122] In the embodiment of the present application, for each traffic behavior evaluation index, the terminal compares the evaluation index value of the traffic behavior evaluation index with the target threshold value of the traffic behavior evaluation index to obtain the comparison result of the traffic behavior evaluation index. Then, the terminal queries the traffic behavior evaluation floating value corresponding to the sample pedestrian according to the comparison result of each traffic behavior evaluation index corresponding to the sample pedestrian in the mapping relationship between the comparison result of each traffic behavior evaluation index and the traffic behavior evaluation floating value contained in the traffic behavior evaluation strategy corresponding to the sample pedestrian.

[0123] In the above-mentioned pedestrian traffic behavior prediction method, the evaluation average value of the traffic behavior evaluation index of the group type to which the sample pedestrian belongs is used as the benchmark of the target threshold corresponding to the sample pedestrian, and different threshold multiples corresponding to different crossing strategies are used. The product of the evaluation average value and the target threshold multiple is used as the target threshold corresponding to the sample pedestrian. According to the comparison result of the evaluation index value and the target threshold, the traffic behavior evaluation floating value corresponding to the sample pedestrian is determined. The differences of the sample pedestrians themselves, the group characteristics of the sample pedestrians and the influence of different crossing strategies on the threshold are comprehensively considered, which is more in line with the actual situation, and their traffic behavior is evaluated more comprehensively and accurately, which can improve the accuracy of the target training set, and thus improve the accuracy of pedestrian traffic behavior prediction.

[0124] In one embodiment, the strategies for crossing the road include yielding, passing normally, and rushing through. The starting values ​​of traffic behavior evaluation corresponding to yielding, passing normally, and rushing through are 4, 5, and 6, respectively. Traffic behavior evaluation indicators include turning head observation information, the number of waving gestures, the relative distance between people and vehicles determined by the strategy, and the time when the conflict occurs. The mapping relationship between the crossing road strategy and the traffic behavior evaluation strategy is shown in Table 3-5. The strategy corresponding to the target group type determines the average evaluation value of the relative distance between people and vehicles. is the evaluation average value of the conflict occurrence time corresponding to the target group type. The mapping relationship between the preset traffic behavior evaluation index, the crossing strategy and the threshold multiple is as follows: the strategy determines the relative distance between people and vehicles, and the threshold multiple corresponding to giving way is 1.2; the conflict occurrence time, the threshold multiple corresponding to giving way is 1.2; the strategy determines the relative distance between people and vehicles, the first threshold multiple and the second threshold multiple corresponding to normal passing are 0.7 and 1.3 respectively; the conflict occurrence time, the first threshold multiple and the second threshold multiple corresponding to normal passing are 0.7 and 1.3 respectively; the strategy determines the relative distance between people and vehicles, the first threshold multiple and the second threshold multiple corresponding to rushing through are 0.8 and 1.4 respectively; the conflict occurrence time, the first threshold multiple and the second threshold multiple corresponding to rushing through are 0.8 and 1.4 respectively. Among them, the threshold multiple includes the first threshold multiple and the second threshold multiple, and the first threshold multiple is less than the second threshold multiple; the traffic behavior evaluation threshold includes the first threshold multiple and the second threshold multiple, the first threshold multiple is the multiple of the benchmark value of the first threshold corresponding to the sample pedestrian, and the second threshold multiple is the multiple of the benchmark value of the second threshold corresponding to the sample pedestrian. The mapping relationship between the traffic behavior evaluation value and the traffic behavior type is: the traffic behavior type corresponding to the traffic behavior evaluation value of 0-3 is cautious, the traffic behavior type corresponding to the traffic behavior evaluation value of 4-6 is mature, and the traffic behavior type corresponding to the traffic behavior evaluation value of 7-10 is impatient.

[0125] Table 3 Mapping relationship between yield and traffic behavior evaluation strategy

[0126]

[0127] Table 4 Mapping relationship between normal passing and traffic behavior evaluation strategy

[0128]

[0129] Table 5 Mapping relationship between rushing through and traffic behavior evaluation strategy

[0130]

[0131] In an exemplary embodiment, Figure 5 As shown, step 104 includes step 501 to step 502. Among them:

[0132] Step 501 : for each group type, determine the group characteristic information corresponding to the group type according to the pedestrian dynamic data and pedestrian psychological data of each sample pedestrian of the group type.

[0133] In an embodiment of the present application, the group characteristic information includes but is not limited to the total sample volume of sample pedestrians of the group type, the average speed of pedestrians, and the proportion of pedestrians of each traffic behavior type. The terminal determines the group type of each sample pedestrian based on the pedestrian static data of each sample pedestrian. It can be understood that the specific process of the terminal determining the group type of each sample pedestrian is similar to the specific process of step 401. In the case where the group characteristic information includes the proportion of pedestrians of each traffic behavior type, the terminal determines the traffic behavior type corresponding to each sample pedestrian of the group type. Then, for each traffic behavior type, the terminal uses the ratio of the number of sample pedestrians of the traffic behavior type of the group type to the total sample volume of sample pedestrians of the group type as the proportion of pedestrians of the traffic behavior type of the group type. It can be understood that the specific process of the terminal determining the traffic behavior type corresponding to each sample pedestrian of the group type is similar to the specific process of steps 201-204.

[0134] Step 502 , construct a target training set based on the pedestrian static data and traffic behavior evaluation results of each sample pedestrian, the traffic scene information and interference source identification results corresponding to each sample pedestrian, and the group characteristic information corresponding to each group type.

[0135] In the embodiment of the present application, for each sample pedestrian, the terminal uses the pedestrian static data and traffic behavior evaluation results of the sample pedestrian, as well as the traffic scene information and interference source identification results corresponding to the sample pedestrian, to form the sample data corresponding to the sample pedestrian. Then, the terminal constructs a target training set based on the sample data corresponding to each sample pedestrian and the group characteristic information corresponding to each group type. For example, the terminal uses the sample data corresponding to each sample pedestrian and the group characteristic information corresponding to each group type to form the target training set.

[0136] In one embodiment, for each group type, the terminal generates a group characteristic chart corresponding to the group type based on the group characteristic information corresponding to the group type. Then, the terminal forms a target training set with the sample data corresponding to each sample pedestrian, the group characteristic information and the group characteristic chart corresponding to each group type. In one example, the terminal can also display the group characteristic chart corresponding to each group type.

[0137] In the above pedestrian traffic behavior prediction method, for each group type, the group characteristic information corresponding to the group type is determined based on the pedestrian dynamic data and pedestrian psychological data of each sample pedestrian of the group type; based on the pedestrian static data and traffic behavior evaluation results of each sample pedestrian, the traffic scene information and interference source identification results corresponding to each sample pedestrian, and the group characteristic information corresponding to each group type, a target training set is constructed. In this way, through classification query and data statistics, the target training set is constructed, which is not only based on the various information of each sample pedestrian, the traffic behavior evaluation results and the traffic scene information, but also based on the group characteristic portrait corresponding to each group type, which is more in line with the actual situation, can optimize the autonomous driving vehicle pedestrian detection and pedestrian traffic behavior intention recognition technology, and improve the accuracy and efficiency of pedestrian traffic behavior prediction.

[0138] In an exemplary embodiment, Figure 6 As shown, step 104 includes step 601 to step 602. Among them:

[0139] Step 601, based on the pedestrian dynamic data of the sample pedestrian, determine the traffic behavior trajectory diagram, speed time series diagram and joint angle change diagram of the sample pedestrian.

[0140] In an embodiment of the present application, the terminal generates a traffic behavior trajectory diagram, a speed time series diagram, and a joint angle change diagram of the sample pedestrian based on the pedestrian dynamic data of the sample pedestrian.

[0141] Step 602, construct a target training set based on the pedestrian static data, traffic behavior evaluation results, traffic behavior trajectory diagram, speed time series diagram and joint angle change diagram of each sample pedestrian, as well as the traffic scene information and interference source identification results corresponding to each sample pedestrian.

[0142] In the embodiment of the present application, for each sample pedestrian, the terminal uses the pedestrian static data, traffic behavior evaluation results, traffic behavior trajectory diagram, speed time series diagram and joint angle change diagram of the sample pedestrian, as well as the traffic scene information and interference source identification results corresponding to the sample pedestrian to form the sample data corresponding to the sample pedestrian. Then, the terminal constructs a target training set based on the sample data corresponding to each sample pedestrian. For example, the terminal uses the sample data corresponding to each sample pedestrian to form the target training set.

[0143] In one embodiment, the terminal can also display the traffic behavior trajectory diagram, speed time series diagram and joint angle change diagram of each sample pedestrian.

[0144] In the above pedestrian traffic behavior prediction method, based on the dynamic data of the sample pedestrians, the traffic behavior trajectory diagram, speed sequence diagram and joint angle change map of the sample pedestrians are determined; based on the pedestrian static data, traffic behavior evaluation results, traffic behavior trajectory diagram, speed sequence diagram and joint angle change map of each sample pedestrian, as well as the traffic scene information and interference source identification results corresponding to each sample pedestrian, a target training set is constructed. In this way, through data query and export, the target training set is constructed, which is not only based on the various information, traffic behavior evaluation results and traffic scene information of each sample pedestrian, but also based on the two-dimensional sample pedestrian traffic behavior trajectory diagram, sample pedestrian speed sequence diagram, and the joint angle change map of each joint of the sample pedestrian collected. These sample pedestrian traffic behavior feature maps can improve data accuracy, be more intuitive, and more in line with actual conditions, and can optimize the autonomous driving vehicle pedestrian detection and pedestrian traffic behavior intention recognition technology, and improve the accuracy and efficiency of pedestrian traffic behavior prediction.

[0145] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0146] Based on the same inventive concept, the embodiment of the present application also provides a pedestrian traffic behavior prediction device for implementing the pedestrian traffic behavior prediction method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more pedestrian traffic behavior prediction device embodiments provided below can refer to the limitations of the pedestrian traffic behavior prediction method above, and will not be repeated here.

[0147] In an exemplary embodiment, Figure 7 As shown, a pedestrian traffic behavior prediction device 700 is provided, comprising: an acquisition module 710, a first determination module 720, a second determination module 730, a construction module 740 and a training module 750, wherein:

[0148] An acquisition module 710, for acquiring pedestrian dynamic data and pedestrian psychological data of a sample pedestrian in a sample traffic scene, a traffic scene image of the sample traffic scene, and pedestrian static data of the sample pedestrian;

[0149] A first determination module 720 is used to determine the traffic scene information and interference source identification result corresponding to the sample pedestrian based on the traffic scene image; the interference source identification result includes the interference source type and interference source dynamics information;

[0150] A second determination module 730, configured to determine a traffic behavior evaluation result of the sample pedestrian based on the pedestrian dynamic data, the interference source dynamic information, and the pedestrian psychological data;

[0151] A construction module 740 is used to construct a target training set based on the pedestrian static data and traffic behavior evaluation results of each of the sample pedestrians, as well as the traffic scene information and interference source identification results corresponding to each of the sample pedestrians;

[0152] The training module 750 is used to train the target neural network based on the target training set to obtain a pedestrian traffic behavior prediction model; the pedestrian traffic behavior prediction model is used to perform pedestrian traffic behavior prediction.

[0153] Optionally, the pedestrian psychological data includes a road crossing strategy and a strategy decision time, the traffic behavior evaluation result includes a traffic behavior type, and the second determination module 730 is specifically used to:

[0154] In the mapping relationship between the preset crossing strategy and the traffic behavior evaluation starting value, querying the traffic behavior evaluation starting value corresponding to the sample pedestrian according to the crossing strategy of the sample pedestrian;

[0155] Determine the traffic behavior evaluation floating value corresponding to the sample pedestrian based on the pedestrian dynamic data, the strategy decision time and the interference source dynamic information;

[0156] Adding the traffic behavior evaluation starting value and the traffic behavior evaluation floating value to obtain the traffic behavior evaluation value corresponding to the sample pedestrian;

[0157] In the preset mapping relationship between the traffic behavior evaluation value and the traffic behavior type, the traffic behavior type corresponding to the sample pedestrian is queried according to the traffic behavior evaluation value corresponding to the sample pedestrian.

[0158] Optionally, the second determining module 730 is specifically configured to:

[0159] Determining an evaluation index value of a traffic behavior evaluation index of the sample pedestrian based on the pedestrian dynamic data, the strategy decision time and the interference source dynamic information;

[0160] In the mapping relationship between the preset road crossing strategy and the traffic behavior evaluation strategy, query the traffic behavior evaluation strategy corresponding to the sample pedestrian according to the road crossing strategy of the sample pedestrian;

[0161] The traffic behavior evaluation floating value corresponding to the sample pedestrian is determined according to the evaluation index value and the traffic behavior evaluation strategy corresponding to the sample pedestrian.

[0162] Optionally, the traffic behavior evaluation index includes the relative distance between the person and the vehicle and / or the time when the conflict occurs. The second determination module 730 is used to:

[0163] Determining the target group type of the sample pedestrian according to the pedestrian static data of the sample pedestrian;

[0164] Calculating the evaluation average of the traffic behavior evaluation index corresponding to the target group type according to the evaluation index value of the traffic behavior evaluation index of each sample pedestrian whose group type is the target group type;

[0165] In the preset mapping relationship among the traffic behavior evaluation index, the jaywalking strategy and the threshold multiple, querying the target threshold multiple corresponding to the sample pedestrian according to the traffic behavior evaluation index and the jaywalking strategy of the sample pedestrian;

[0166] The product of the evaluation average value and the target threshold multiple is used as the target threshold corresponding to the sample pedestrian;

[0167] According to the comparison result of the evaluation index value and the target threshold value, the traffic behavior evaluation floating value corresponding to the sample pedestrian is determined.

[0168] Optionally, the construction module 740 is specifically used to:

[0169] For each group type, determining group characteristic information corresponding to the group type according to pedestrian dynamic data and pedestrian psychological data of each sample pedestrian of the group type;

[0170] A target training set is constructed based on the pedestrian static data and traffic behavior evaluation results of each of the sample pedestrians, the traffic scene information and interference source identification results corresponding to each of the sample pedestrians, and the group characteristic information corresponding to each of the group types.

[0171] Optionally, the construction module 740 is specifically used to:

[0172] Based on the pedestrian dynamic data of the sample pedestrian, determining the traffic behavior trajectory diagram, speed time series diagram and joint angle change diagram of the sample pedestrian;

[0173] A target training set is constructed based on the pedestrian static data, traffic behavior evaluation results, traffic behavior trajectory diagram, speed time series diagram and joint angle change diagram of each of the sample pedestrians, as well as the traffic scene information and interference source identification results corresponding to each of the sample pedestrians.

[0174] Each module in the above pedestrian traffic behavior prediction device can be implemented in whole or in part by software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute the operations corresponding to each module.

[0175] In an exemplary embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as shown in FIG. Figure 8 As shown. The computer device includes a processor, a memory, an input / output interface, a communication interface, a display unit and an input device. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface, the display unit and the input device are connected to the system bus through the input / output interface. The processor of the computer device is used 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 the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be implemented through WIFI, a mobile cellular network, near field communication (NFC) or other technologies. When the computer program is executed by the processor, a pedestrian traffic behavior prediction method is implemented. The display unit of the computer device is used to form a visually visible picture, which can be a display screen, a projection device or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad set on the computer device shell, or an external keyboard, touchpad or mouse.

[0176] Those skilled in the art will understand that Figure 8 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0177] In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above-mentioned method embodiments when executing the computer program.

[0178] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0179] In one embodiment, a computer program product is provided, including a computer program, which implements the steps in the above method embodiments when executed by a processor.

[0180] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0181] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, an artificial intelligence (AI) processor, etc., but are not limited to this.

[0182] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0183] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A pedestrian traffic behavior prediction method, characterized in that: The method comprises: Acquire pedestrian dynamic data and pedestrian psychological data of a sample pedestrian in a sample traffic scene, a traffic scene image of the sample traffic scene, and pedestrian static data of the sample pedestrian; Based on the traffic scene image, determine the traffic scene information and interference source identification result corresponding to the sample pedestrian; the interference source identification result includes the interference source type and interference source dynamics information; Determining a traffic behavior evaluation result of the sample pedestrian based on the pedestrian dynamic data, the interference source dynamic information, and the pedestrian psychological data; Constructing a target training set based on the pedestrian static data and traffic behavior evaluation results of each of the sample pedestrians, as well as the traffic scene information and interference source identification results corresponding to each of the sample pedestrians; Based on the target training set, a target neural network is trained to obtain a pedestrian traffic behavior prediction model; the pedestrian traffic behavior prediction model is used to perform pedestrian traffic behavior prediction; Wherein, the pedestrian psychological data includes a crossing strategy and a strategy decision time, the traffic behavior evaluation result includes a traffic behavior type, and the traffic behavior evaluation result of the sample pedestrian is determined based on the pedestrian dynamic data, the interference source dynamic information, and the pedestrian psychological data, including: in the preset mapping relationship between the crossing strategy and the traffic behavior evaluation starting value, according to the crossing strategy of the sample pedestrian, query the traffic behavior evaluation starting value corresponding to the sample pedestrian; based on the pedestrian dynamic data, the strategy decision time and the interference source dynamic information, determine the traffic behavior evaluation floating value corresponding to the sample pedestrian; add the traffic behavior evaluation starting value and the traffic behavior evaluation floating value to obtain the traffic behavior evaluation value corresponding to the sample pedestrian; in the preset mapping relationship between the traffic behavior evaluation value and the traffic behavior type, query the traffic behavior type corresponding to the sample pedestrian according to the traffic behavior evaluation value corresponding to the sample pedestrian; Wherein, the determining of the traffic behavior evaluation floating value corresponding to the sample pedestrian based on the pedestrian dynamic data, the strategy decision time and the interference source dynamic information includes: determining the evaluation index value of the traffic behavior evaluation index of the sample pedestrian based on the pedestrian dynamic data, the strategy decision time and the interference source dynamic information; in the preset mapping relationship between the crossing strategy and the traffic behavior evaluation strategy, querying the traffic behavior evaluation strategy corresponding to the sample pedestrian according to the crossing strategy of the sample pedestrian; determining the traffic behavior evaluation floating value corresponding to the sample pedestrian according to the evaluation index value and the traffic behavior evaluation strategy corresponding to the sample pedestrian; Among them, the traffic behavior evaluation index includes the relative distance between people and vehicles and / or the time of conflict occurrence, and the traffic behavior evaluation floating value corresponding to the sample pedestrian is determined according to the evaluation index value and the traffic behavior evaluation strategy corresponding to the sample pedestrian, including: determining the target group type of the sample pedestrian according to the pedestrian static data of the sample pedestrian; calculating the evaluation average value of the traffic behavior evaluation index corresponding to the target group type according to the evaluation index value of the traffic behavior evaluation index of each sample pedestrian whose group type is the target group type; in the mapping relationship between the preset traffic behavior evaluation index, the crossing strategy and the threshold multiple, querying the target threshold multiple corresponding to the sample pedestrian according to the traffic behavior evaluation index and the crossing strategy of the sample pedestrian; taking the product of the evaluation average value and the target threshold multiple as the target threshold corresponding to the sample pedestrian; and determining the traffic behavior evaluation floating value corresponding to the sample pedestrian according to the comparison result of the evaluation index value and the target threshold.

2. The method according to claim 1, characterized in that The target training set is constructed based on the pedestrian static data and traffic behavior evaluation results of each of the sample pedestrians, as well as the traffic scene information and interference source identification results corresponding to each of the sample pedestrians, including: For each group type, determining group characteristic information corresponding to the group type according to pedestrian dynamic data and pedestrian psychological data of each sample pedestrian of the group type; A target training set is constructed based on the pedestrian static data and traffic behavior evaluation results of each of the sample pedestrians, the traffic scene information and interference source identification results corresponding to each of the sample pedestrians, and the group characteristic information corresponding to each of the group types.

3. The method according to claim 1, characterized in that The target training set is constructed based on the pedestrian static data and traffic behavior evaluation results of each of the sample pedestrians, as well as the traffic scene information and interference source identification results corresponding to each of the sample pedestrians, including: Based on the pedestrian dynamic data of the sample pedestrian, determining the traffic behavior trajectory diagram, speed time series diagram and joint angle change diagram of the sample pedestrian; A target training set is constructed based on the pedestrian static data, traffic behavior evaluation results, traffic behavior trajectory diagram, speed time series diagram and joint angle change diagram of each of the sample pedestrians, as well as the traffic scene information and interference source identification results corresponding to each of the sample pedestrians.

4. The method according to claim 1, characterized in that The determining, based on the traffic scene image, traffic scene information and interference source identification results corresponding to the sample pedestrian includes: Using a deep learning detection algorithm, based on the traffic scene image, determine the traffic scene information corresponding to the sample pedestrian; Based on the traffic scene image and a preset target detection algorithm, an interference source recognition result corresponding to the sample pedestrian is determined.

5. A pedestrian traffic behavior prediction device, characterized in that: The device comprises: An acquisition module, used to acquire pedestrian dynamic data and pedestrian psychological data of a sample pedestrian in a sample traffic scene, a traffic scene image of the sample traffic scene, and pedestrian static data of the sample pedestrian; A first determination module is used to determine the traffic scene information and interference source identification result corresponding to the sample pedestrian based on the traffic scene image; the interference source identification result includes the interference source type and interference source dynamics information; A second determination module, configured to determine a traffic behavior evaluation result of the sample pedestrian based on the pedestrian dynamic data, the interference source dynamic information, and the pedestrian psychological data; A construction module, used to construct a target training set based on the pedestrian static data and traffic behavior evaluation results of each of the sample pedestrians, as well as the traffic scene information and interference source identification results corresponding to each of the sample pedestrians; A training module, used to train a target neural network based on the target training set to obtain a pedestrian traffic behavior prediction model; the pedestrian traffic behavior prediction model is used to perform pedestrian traffic behavior prediction; Among them, the pedestrian psychological data includes a road crossing strategy and a strategy decision time, the traffic behavior evaluation result includes a traffic behavior type, and the second determination module is specifically used to: in the preset mapping relationship between the road crossing strategy and the traffic behavior evaluation starting value, according to the road crossing strategy of the sample pedestrian, query the traffic behavior evaluation starting value corresponding to the sample pedestrian; based on the pedestrian dynamic data, the strategy decision time and the interference source dynamics information, determine the traffic behavior evaluation floating value corresponding to the sample pedestrian; add the traffic behavior evaluation starting value and the traffic behavior evaluation floating value to obtain the traffic behavior evaluation value corresponding to the sample pedestrian; in the preset mapping relationship between the traffic behavior evaluation value and the traffic behavior type, according to the traffic behavior evaluation value corresponding to the sample pedestrian, query the traffic behavior type corresponding to the sample pedestrian; Wherein, the second determination module is specifically used to: determine the evaluation index value of the traffic behavior evaluation index of the sample pedestrian based on the pedestrian dynamic data, the strategy decision time and the interference source dynamic information; in the preset mapping relationship between the crossing strategy and the traffic behavior evaluation strategy, query the traffic behavior evaluation strategy corresponding to the sample pedestrian according to the crossing strategy of the sample pedestrian; determine the traffic behavior evaluation floating value corresponding to the sample pedestrian according to the evaluation index value and the traffic behavior evaluation strategy corresponding to the sample pedestrian; Among them, the traffic behavior evaluation index includes the relative distance between people and vehicles and / or the time of conflict occurrence. The second determination module is specifically used to: determine the target group type of the sample pedestrian according to the pedestrian static data of the sample pedestrian; calculate the evaluation average value of the traffic behavior evaluation index corresponding to the target group type according to the evaluation index value of the traffic behavior evaluation index of each sample pedestrian whose group type is the target group type; in the preset mapping relationship between traffic behavior evaluation index, crossing strategy and threshold multiple, query the target threshold multiple corresponding to the sample pedestrian according to the traffic behavior evaluation index and crossing strategy of the sample pedestrian; use the product of the evaluation average value and the target threshold multiple as the target threshold corresponding to the sample pedestrian; and determine the traffic behavior evaluation floating value corresponding to the sample pedestrian according to the comparison result of the evaluation index value and the target threshold.

6. The device according to claim 5, characterized in that The building blocks are specifically used for: For each group type, determining group characteristic information corresponding to the group type according to pedestrian dynamic data and pedestrian psychological data of each sample pedestrian of the group type; A target training set is constructed based on the pedestrian static data and traffic behavior evaluation results of each of the sample pedestrians, the traffic scene information and interference source identification results corresponding to each of the sample pedestrians, and the group characteristic information corresponding to each of the group types.

7. The device according to claim 5, characterized in that The building blocks are specifically used for: Based on the pedestrian dynamic data of the sample pedestrian, determining the traffic behavior trajectory diagram, speed time series diagram and joint angle change diagram of the sample pedestrian; A target training set is constructed based on the pedestrian static data, traffic behavior evaluation results, traffic behavior trajectory diagram, speed time series diagram and joint angle change diagram of each of the sample pedestrians, as well as the traffic scene information and interference source identification results corresponding to each of the sample pedestrians.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 4 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 4 are implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 4 are implemented.

Images