Information processing device, information processing method, and information processing program

Abstract

Even when road traffic conditions change, it is possible to simulate the decisions of traffic participants without generating a pre-trained model for each traffic condition. [Solution] The information processing device acquires a basic model that has been pre-trained on the decisions made by a target person in a basic situation where a pedestrian attempting to cross a crosswalk at an intersection is being intersected by one vehicle while going straight, and by one vehicle after turning right and intersecting with the pedestrian is also entering the intersection. Multiple traffic participants in an intersection with traffic conditions different from the basic situation are divided into a first group that enters the intersection by going straight and a second group that enters the intersection by turning right or left. From each group, one traffic participant is selected who will reach the point where they intersect with the target person the fastest from their current location. Movement status information indicating the position and speed of the selected traffic participant is acquired, and the movement status information of the selected traffic participant is input into the basic model to estimate the decisions made by the target person.

Description

[Technical Field]

[0001] The present invention relates to an information processing device, an information processing method, and an information processing program, and more particularly to an information processing device, an information processing method, and an information processing program suitable for estimating the judgment of traffic participants. [Background technology]

[0002] Conventionally, simulations have been conducted to simulate the decisions made by traffic participants in response to road traffic conditions. To perform such simulations, a trained model is required, which has been trained using the movement status of other traffic participants and the decisions of the target of the simulation corresponding to these movement statuses as training data for each traffic condition (see, for example, Patent Document 1).

[0003] Because various traffic conditions exist in the real world, it is necessary to reproduce various traffic conditions in simulations of traffic participant decision-making. While it was desirable to generate pre-trained models for each traffic condition, collecting training data and conducting training under various traffic conditions was extremely difficult from the standpoint of cost and effort. [Prior art documents] [Patent Documents]

[0004] [Patent Document 1] Japanese Patent Publication No. 2023-57652 [Overview of the Initiative] [Problems that the invention aims to solve]

[0005] Therefore, the present invention aims to provide an information processing device, an information processing method, and an information processing program that can simulate the decisions of traffic participants without generating a pre-trained model for each traffic situation, even when road traffic conditions change. [Means for solving the problem]

[0006] That is, the information processing apparatus according to the first aspect is an information processing apparatus that targets a traffic participant who attempts to cross an intersection without a traffic signal as a simulation target and estimates the judgment of the target according to the traffic situation at the intersection. The basic model acquisition unit acquires a basic model that has learned in advance the judgment made by the target in the basic situation where a pedestrian attempting to cross the crosswalk at the intersection, one vehicle intersecting while going straight, and one vehicle intersecting after turning right enter the intersection. The grouping unit divides a plurality of traffic participants at an intersection with a traffic situation different from the basic situation into a first group that enters the intersection by going straight and a second group that enters the intersection by turning right or left. The selection unit selects one traffic participant with the fastest arrival from the current location to the intersection point with the target from each of the first group and the second group. The moving state information acquisition unit acquires moving state information indicating the moving state of the position and moving speed of the selected traffic participant. The estimation unit estimates the judgment made by the target by inputting the moving state information of the selected traffic participant into the basic model. It is characterized by comprising these components.

[0007] In the second aspect, in the information processing apparatus according to the first aspect, the estimation unit may estimate, as the judgment made by the target, a judgment that the target can pass through the crosswalk, a judgment to wait for the passage of the traffic participant, and any one before the target makes a judgment, calculate the probability of each judgment made by the target, and estimate the judgment with the highest probability as the judgment made by the target.

[0008] In the third aspect, in the information processing apparatus according to the second aspect, it may further comprise a speed calculation unit that calculates the moving speed or acceleration of the target according to the probability.

[0009] In the fourth aspect, in the information processing apparatus according to the first aspect, when the moving direction of the selected traffic participant is opposite to the moving direction of the traffic participant in the basic situation, it may further comprise a moving state information conversion unit that converts the moving state information of the selected traffic participant so that the position of the selected traffic participant moves to a position that is a point target centered on the center of the intersection.

[0010] In the fifth aspect, in the information processing apparatus according to the first aspect, the target person may be a pedestrian or a driver driving a vehicle.

[0011] The information processing method according to the sixth aspect is an information processing method for simulating a traffic participant who intends to cross an intersection without a traffic signal as a target person and estimating a determination of the target person according to the traffic situation of the intersection. The computer obtains a basic model that has learned in advance the determination made by the target person in a basic situation where a pedestrian who intends to cross a crosswalk of the intersection, a vehicle that intersects while going straight, and a vehicle that intersects after turning right enter the intersection. This is a basic model acquisition step. A grouping step of dividing a plurality of traffic participants at an intersection with a traffic situation different from the basic situation into a first group that enters the intersection by going straight and a second group that enters the intersection by turning right or left. A selection step of selecting one traffic participant who reaches the point of intersection with the target person from the current location earliest from each of the first group and the second group. A moving state information acquisition step of acquiring moving state information indicating the moving state of the position and moving speed of the selected traffic participant. An estimation step of estimating the determination made by the target person by inputting the moving state information of the selected traffic participant into the basic model. It is characterized by executing the above steps.

[0012] The seventh aspect of the information processing program is an information processing program that simulates traffic participants attempting to cross an intersection without traffic signals and estimates the judgment of the participants according to the traffic conditions at the intersection, and is characterized by realizing the following functions: a basic model acquisition function that acquires a basic model that has been pre-learned by a computer, which is a basic model that acquires [Effects of the Invention]

[0013] The information processing device according to the present invention is an information processing device that simulates traffic participants attempting to cross an intersection without traffic signals and estimates the judgment of the participants according to the traffic conditions at the intersection, and includes a basic model acquisition unit that acquires a basic model that has been pre-learned of the judgment made by a participant in a basic situation in which a pedestrian attempting to cross a crosswalk at an intersection is about to be crossed by one vehicle that is going straight and another vehicle that is about to be crossed by one vehicle that is going right and then turns right and enters the intersection, and a first group of traffic participants entering the intersection by going straight and a second group of traffic participants entering the intersection by turning right or left, which are different from the basic situation, The system is characterized by comprising: a grouping unit that divides the traffic into groups; a selection unit that selects one traffic participant from each of the first and second groups who will reach the point where they intersect with the target person from their current location in the shortest time; a movement state information acquisition unit that acquires movement state information indicating the position and speed of the selected traffic participants; and an estimation unit that estimates the decisions made by the target person by inputting the movement state information of the selected traffic participants into a basic model. Therefore, even if the road traffic conditions change, it is possible to simulate the decisions of traffic participants without generating a trained model for each traffic condition.

[0014] Furthermore, the information processing method and information processing program according to the present invention, like the information processing apparatus according to the present invention, can simulate the decisions of traffic participants without generating a pre-trained model for each traffic condition, even when road traffic conditions change. [Brief explanation of the drawing]

[0015] [Figure 1] Figure 1 is a diagram illustrating the basic situation according to the first embodiment. [Figure 2] Figure 2 shows the processing flow for constructing the basic model according to the first embodiment. [Figure 3] Figure 3 is a diagram illustrating the integration of individual decisions in the information processing device according to the first embodiment. [Figure 4]Figure 4 is a diagram illustrating an example of an integrated table for individual decisions in the information processing device according to the first embodiment. [Figure 5] Figure 5 is a diagram illustrating an example of a traffic situation that is the target of simulation by the information processing device according to the first embodiment. [Figure 6] Figure 6 is a diagram illustrating the overview of the processing flow for performing a simulation of the information processing device according to the first embodiment. [Figure 7] Figure 7 is a diagram illustrating the TTCP according to the first embodiment. [Figure 8] Figure 8 is a diagram illustrating the data conversion of the information processing apparatus according to the first embodiment. [Figure 9] Figure 9 is a diagram illustrating the hardware configuration of the information processing device according to the first embodiment. [Figure 10] Figure 10 is a diagram illustrating the functional configuration of the information processing apparatus according to the first embodiment. [Figure 11] Figure 11 is a diagram illustrating the basic situation according to the second embodiment. [Figure 12] Figure 12 is a diagram illustrating the traffic conditions that are the subject of the simulation according to the second embodiment. [Figure 13] Figure 13 is an example of a flowchart of an information processing program according to the third embodiment. [Figure 14] Figure 14 is an example of a flowchart of an information processing program according to the fourth embodiment. [Modes for carrying out the invention]

[0016] An example of an information processing device 100 according to the first embodiment of this disclosure will be described with reference to Figures 1 to 10. The information processing device 100 is a so-called computer, also known as a server, personal computer (hereinafter referred to as PC), notebook PC, tablet PC, or smartphone. The information processing device 100 according to the first embodiment uses a traffic participant attempting to cross an intersection without traffic signals as the subject of the simulation and estimates the decision of this subject. The target of the simulation by the information processing device 100 according to the first embodiment is a pedestrian 24a attempting to cross a crosswalk 23 at an intersection 20 without traffic signals. The target of the simulation by the information processing device 100 according to the second embodiment, described later, is the driver of a vehicle 74 that is proceeding straight through the intersection 20 without traffic signals.

[0017] (Regarding the basic situation and basic model according to the first embodiment) The basic situation and basic model according to the first embodiment will be described with reference to Figures 1 and 2. Figure 1 is a diagram illustrating the basic situation according to the first embodiment, and Figure 2 is a diagram showing the processing flow for constructing the basic model according to the first embodiment. The basic situation refers to the fundamental traffic conditions that arise at intersection 1, which has no traffic signals, and involves one of each of the traffic participants at intersection 1: a pedestrian 5, a vehicle going straight (CarS) 6, and a vehicle turning right (CarR) 7. Specifically, the basic situation refers to a situation where there is a pedestrian 5 attempting to cross a crosswalk and one vehicle going straight (CarS) that intersects with it, and a pedestrian 5 attempting to cross a crosswalk and one vehicle turning right (CarR) that intersects with it after it has turned off the road. The basic model is a machine learning model that associates the decision of a traffic participant to cross a crosswalk in a basic situation with the movement status (position, speed, etc.) of other traffic participants. By inputting the relative movement status (position, speed, etc.) of a traffic participant attempting to cross a crosswalk in a basic situation with other traffic participants into the basic model, the decision of a traffic participant attempting to cross a crosswalk can be estimated. The information processing device 100 according to the first embodiment estimates the judgment of traffic participants attempting to cross a pedestrian crossing in various traffic situations different from the basic situation occurring at an intersection 1 without traffic signals, by using an extended basic model. The traffic participants may include pedestrians 5, straight-ahead vehicles (CarS) 6, right-turn vehicles (CarR) 7, and left-turners.

[0018] As shown in FIG. 1, the intersection 1 without a traffic signal is a place where the one-way one-lane priority road 2 extending horizontally and the non-priority road 3 extending vertically intersect, and no traffic signal is installed there. Generally, at intersections, there is particularly a lot of traffic where traffic participants such as pedestrians and vehicles cross the road, and it is a place where traffic accidents are likely to occur. Especially at intersections without traffic signals, crossing is carried out based on the judgment of traffic participants, so it is possible to find important significance in simulating the judgment of traffic participants in such traffic situations.

[0019] As shown in FIG. 1, the position coordinates at the intersection 1 are represented in a Cartesian coordinate system with the vertical axis as the x-axis and the horizontal axis as the y-axis, and the origin 0 is at the center of the upper end of the crosswalk 4. As shown in FIG. 1, the position coordinates of the pedestrian 5 at the intersection 1 are (X P (t), Y P (t)), and the traveling direction speed is represented by v P (t). The position coordinates of the straight-ahead vehicle (CarS) 6 are (X S (t), Y S (t)), and the traveling direction speed is represented by V S (t). The position coordinates of the right-turn vehicle (CarR) 7 are (X R (t), Y R (t)), and the traveling direction speed is represented by V R (t). These position coordinates and traveling direction speeds are represented with time as an explanatory variable.

[0020] The "measurement data of human behavior" 10 shown in FIG. 2 was obtained by constructing the intersection 1 without a traffic signal in a virtual space and conducting a crosswalk crossing observation experiment using a multi-player simulator. The multi-player in this embodiment refers to the pedestrian 5 and the drivers of the straight-ahead vehicle (CarS) 6 and the right-turn vehicle (CarR) 7. In other words, in the crosswalk observation experiment, actual people operate a simulator to participate in the traffic situation at intersection 1 constructed in a virtual space, acting as traffic participants such as pedestrian 5, straight-going vehicle (CarS) 6, and right-turning vehicle (CarR) 7. The "human behavior measurement data" 10 obtained in this crosswalk observation experiment is the position coordinate (X) of pedestrian 5 at intersection 1 as described above. P (t), Y P (t)) and forward direction speed (traveling speed) v P (t), position coordinates (X S (t), Y S (t)) and speed in the forward direction V S (t), and the position coordinates (X) of the right-turning vehicle (CarR) 7. R (t), Y R (t)) and speed in the forward direction V R (t)

[0021] The multiplayer simulator consists of a pedestrian simulator for pedestrian 5 and driving simulators for straight-moving vehicles (CarS) 6 and right-turning vehicles (CarR) 7. The pedestrian simulator is composed of VR goggles, controllers, and trackers (none of which are shown). By having the subject (pedestrian 5) wear these VR devices, they can move freely around intersection 1 in a virtual space constructed with Unity®, etc., by walking in the real world. Therefore, it is possible to obtain experimental data that is close to actual walking. The pedestrian simulator can acquire the face angle, position, speed, and decision results of pedestrian 5. The pedestrian's judgment is obtained by the button operation of the controller held by the subject (Pedestrian 5). The driving simulator (not shown) consists of a steering wheel, seat, brakes, and accelerator used in a real car. Furthermore, by using simulation applications that reproduce the behavior of automobiles, such as CarSim (registered trademark), it is possible to reproduce the same behavior as a real car. By using the driving simulator, it is possible to obtain vehicle position, speed, angle, pedal pressure, and the driver's judgment. This experiment was conducted multiple times under various conditions. In addition, in this experiment, in order to observe judgments and movements that are not strictly defined by rules, participants were instructed to "drive and walk as you normally would, rather than complying with the law."

[0022] In the "data preprocessing" 11 shown in Figure 2, training data for machine learning the basic model is extracted and preprocessed. The training data is a dataset consisting of combinations of input and output variables, with "human behavior measurement data" 10 as input variables and the subject's judgment corresponding to these input variables as output variables. The output variables include Go (judgment that it is possible to pass the crosswalk), Wait (judgment to wait for the vehicle to pass), and Undecided (before the decision of Go / Wait). The "human behavior measurement data" 10 input variables are obtained by normalizing them. Normalization means adjusting the units or scale of the data to a common standard in order to facilitate comparison and analysis, for example, scaling the given data so that the minimum value is 0 and the maximum value is 1.

[0023] In the "preprocessed data" 12 shown in Figure 2, the data acquired in "data preprocessing" 11 is stored in the storage unit 100d of the information processing device 100 (described later) as training data used to construct the basic model.

[0024] In "Basic Model Construction" 13, the basic model is formulated, and the necessary parameters (the coefficient vector η described later) are calculated using the preprocessed data accumulated in "Preprocessed Data" 12. In "Basic Model Construction" 13, the "Individual Judgment Model 14 for Straight-Going Vehicle (CarS) 6", the "Individual Judgment Model 15 for Right-Turning Vehicle (CarR) 7", and the "Integrated Table of Individual Judgments 16" are constructed. In other words, the subject makes individual judgments regarding the straight-going vehicle (CarS) 6 and the right-turning vehicle (CarR) 7. Specifically, the subject's judgment regarding the straight-going vehicle (CarS) 6 (Individual Judgment D1(t)) is estimated using the "Individual Judgment Model 14 for Straight-Going Vehicle (CarS) 6" based on the movement status information (X1(t)) of the straight-going vehicle (CarS) 6. The subject's judgment regarding the right-turning vehicle (CarR) 7 (Individual Judgment D2(t)) is estimated using the "Individual Judgment Model 15 for Right-Turning Vehicle (CarR) 7" based on the movement status information (X2(t)) of the right-turning vehicle (CarR) 7. The estimation of the subject's final decision D(t) (Go, Wait, Undecided) regarding whether or not to cross the pedestrian crossing is performed by integrating the subject's individual decision D1(t) regarding the vehicle going straight (CarS) 6 and the individual decision (D2(t)) regarding the vehicle turning right (CarR) 7. This integration is performed using equation (3) below. Specifically, the probability of each decision in the final decision D(t) (Go, Wait, Undecided) is calculated using equation (3) described below, and the decision with the highest probability is estimated to be the subject's final decision D(t). The "Individual Decision Model 14 for Straight-Going Vehicle (CarS) 6" and the "Individual Decision Model 15 for Right-Turning Vehicle (CarR) 7" are constructed using the logistic regression model described below, and the probability of the individual decision (Go, Wait, or Undecided) made by the subject (pedestrian 5) at time t is described by the following equations (1a, 1b). The "Individual Decision Model 14 for Straight-Going Vehicle (CarS) 6" and the "Individual Decision Model 15 for Right-Turning Vehicle (CarR) 7" estimate the individual decision (D1(t), D2(t)) made by the subject (pedestrian 5) at time t as the one with the highest probability of the individual decision (Go, Wait, or Undecided) calculated by equations (1a, 1b). The "Integrated Individual Judgment Table 16," as shown in Figure 4, is a table in which a weight coefficient for integrating judgments (see Equation 3) is set for each combination of individual judgments and final judgments. This weight coefficient is used to calculate the probability of the final judgment by integrating the individual judgments for the straight-going vehicle (CarS) 6 and the right-turning vehicle (CarR) 7. The weight coefficients for integrating judgments in the Integrated Individual Judgment Table are set based on data obtained from cross-pedestrian crossing observation experiments using a multiplayer simulator.

[0025] Referring to Figure 3, we will explain the construction of a basic model of the subject in a basic situation, which is the basic model. Figure 3 is a diagram for explaining the integration of individual decisions of the information processing device 100 according to the first embodiment. Integration of individual judgments refers to integrating the judgments of individuals that were individually estimated for vehicles going straight (CarS) and vehicles turning right (CarR) to arrive at the individual's final judgment. As shown in Figure 3, X1(t)52 represents the relative movement state of the subject (pedestrian 5) relative to the current (time t) of the moving vehicle (CarS) 6, and D1(t)53 represents the subject's judgment toward the moving vehicle (CarS) 6, estimated by the basic model represented by the logistic regression model described later. As shown in Figure 3, X2(t)54 represents the relative movement state of the target person (pedestrian 5) of the right-turning vehicle (CarR) 7 at the present time (time t), and D2(t)55 represents the target person's judgment towards the right-turning vehicle (CarR) 7, estimated by the basic model represented by the logistic regression model described later.

[0026] In the first embodiment, a logistic regression model is used to express the probability of a decision made by the subject, pedestrian 5. A logistic regression model expresses the probability of an event occurring using n-dimensional explanatory variables φ(t)∈R n Its coefficient vector η∈R n This is a model expressed using a sigmoid function. In the first embodiment, the probability of the decision D(t) (Go, Wait, or Undecided) made by the subject (pedestrian 5) at time t is described by the following equations (1a, 1b). However, D(t) is a random variable of the form d(t)∈(Go, Wait, Undecided), and η d(t) represents the coefficient vector when the decision is d(t). A random variable is a variable whose value is determined randomly and probabilistically by the outcome of a trial, and in this embodiment, it refers to one of Go, Wait, or Undecided. Formula (1a) is used to calculate the probability that the subject will decide to go or wait. Equation (1b) is used to calculate the probability that the subject will determine the answer as "Undecided". The explanatory variable φ(t) was defined as shown in equation (2) below. The first component of equation (2) represents the constant term. Furthermore, the coefficient vector η for each decision is obtained using maximum likelihood estimation based on the obtained preprocessed data.

[0027]

number

[0028] As shown in Figure 3, D(t)56 represents the subject's (pedestrian 5's) final decision, which is one of Go, Wait, or Undecided. Equation (3) is used to calculate the probability of the subject's (pedestrian 5's) final decision. Of the subject's (pedestrian 5's) final decisions of Go, Wait, and Undecided, the decision with the highest probability calculated by equation (3) is adopted as the subject's (pedestrian 5's) final decision. The weight coefficient for decision integration in equation (3) is a weight coefficient for the probability of each decision, corresponding to the decision that has the highest probability for each of the individual decisions described above. The information processing device 100 multiplies the probabilities of G, Wait, and Undecided obtained for each individual decision by the integration weight coefficient shown in Figure 4, and the probability that is highest as a result of the multiplication is taken as the final decision of the subject (pedestrian 5).

[0029] Figure 4 is a diagram illustrating an example of an integrated table for individual decisions of the information processing device 100 according to the first embodiment. The PGGG shown in Figure 4 is a weighting coefficient for integrating judgments when the subject's judgment D1(t) for a straight-going vehicle (CarS) 6 is Go, the subject's judgment D2(t) for a right-turning vehicle (CarR) 7 is Go, and the subject's final judgment is Go. For example, it is 0.9. PGGW, shown in Figure 4, is a weight coefficient for integrating judgments when the subject's judgment D1(t) for a straight-going vehicle (CarS) 6 is Go, the subject's judgment D2(t) for a right-turning vehicle (CarR) 7 is Go, and the subject's final judgment is Wait. For example, it is 0.05. The PGGU shown in Figure 4 is a weight coefficient for integrating decisions when the subject's decision D1(t) for a straight-going vehicle (CarS) 6 is Go, the subject's decision D2(t) for a right-turning vehicle (CarR) 7 is Go, and the subject's final decision is Undecided. For example, it is 0.05.

[0030] The PGWG shown in Figure 4 is a weight coefficient for integrating decisions when the subject's decision D1(t) for a straight-going vehicle (CarS) 6 is Go, the subject's decision D2(t) for a right-turning vehicle (CarR) 7 is Wait, and the subject's final decision is Go. For example, it is 0.01. The PGWW shown in Figure 4 is a weight coefficient for integrating judgments when the subject's judgment D1(t) for a straight-going vehicle (CarS) 6 is Go, the subject's judgment D2(t) for a right-turning vehicle (CarR) 7 is Wait, and the subject's final judgment is Wait. For example, it is 0.98. The PGWU shown in Figure 4 is a weight coefficient for integrating decisions when the subject's decision D1(t) for a straight-going vehicle (CarS) 6 is Go, the subject's decision D2(t) for a right-turning vehicle (CarR) 7 is Wait, and the subject's final decision is Undecided. For example, it is 0.01.

[0031] The PGUG shown in Figure 4 is a weight coefficient for integrating decisions when the subject's decision D1(t) for a straight-going vehicle (CarS) 6 is Go, the subject's decision D2(t) for a right-turning vehicle (CarR) 7 is Undecided, and the subject's final decision is Go. For example, it is 0.01. The PGUW shown in Figure 4 is a weight coefficient for integrating decisions when the subject's decision D1(t) for a straight-going vehicle (CarS) 6 is Go, the subject's decision D2(t) for a right-turning vehicle (CarR) 7 is Undecided, and the subject's final decision is Wait. For example, it is 0.01. The PGUU shown in Figure 4 is a weight coefficient for integrating decisions when the subject's decision D1(t) for a straight-going vehicle (CarS) 6 is Go, the subject's decision D2(t) for a right-turning vehicle (CarR) 7 is Undecided, and the subject's final decision is Undecided. For example, it is 0.98. The weight coefficients for decision integration in the integrated table shown in Figure 4 are set based on data obtained from cross-pedestrian crossing observation experiments using a multiplayer simulator.

[0032]

number

[0033] (Regarding the traffic conditions to be simulated and the implementation of the simulation) Referring to Figures 5 and 6, the traffic conditions that are the target of the simulation of the information processing device 100 according to the first embodiment and the implementation of said simulation will be explained. Figure 5 is a diagram illustrating an example of a traffic situation that is the target of simulation by the information processing device 100 according to the first embodiment, and Figure 6 is a diagram illustrating an overview of the processing flow for implementing the simulation by the information processing device 100 according to the first embodiment. The traffic conditions simulated by the information processing device 100 differ from the basic conditions described above. They include situations where there are multiple vehicles (straight-going vehicles: CarS) crossing the road with pedestrians attempting to cross a crosswalk, and multiple vehicles (right-turning vehicles: CarR) crossing the road after turning right with pedestrians attempting to cross a crosswalk. The intersection 20 that the information processing device 100 simulates is, as shown in Figure 5, a place where a horizontal road 21 with one lane in each direction extending to the left and right intersects with a vertical road 22 with one lane in each direction extending up and down, and where no traffic signals are installed. The lateral road 21 comprises a left-facing road 21a on which vehicles travel to the left, and a right-facing road 21b on which vehicles travel to the right. The longitudinal road 22 comprises an upward road 22a on which vehicles travel upwards and a downward road 22b on which vehicles travel downwards. As shown in Figure 5, the lateral road 21 has a pedestrian crossing 23 for crossing it. Assume that at pedestrian crossing 23, there are pedestrians 24 (24a, 24b) and 25 (25a, 25b) attempting to cross downwards.

[0034] The traffic conditions that the information processing device 100 according to the first embodiment will simulate are, as an example among various traffic conditions, the traffic conditions shown below. That is, there are two pedestrians 24 (24a, 24b) attempting to cross downwards, and two pedestrians 25 (25a, 25b) attempting to cross upwards. In the first embodiment, the target of the simulation by the information processing device 100 is a pedestrian 24a attempting to cross downwards. Therefore, the relative position coordinates of other traffic participants are calculated based on the pedestrian 24a. Also, a left-facing vehicle traveling on left-facing road 21a (CarS R There are two 26s (26a, 26b), and a right-facing straight vehicle (CarS) is traveling on the right-facing road 21b. L Assume there are two units of )27 (27a, 27b). Also, a vehicle making a right turn while traveling on the downward road 22b (CarR U )Assume there are two vehicles with a 28 (28a, 28b), and the vehicle turning right with an upward-facing direction (CarR D Assume there are two units of 29 (29a and 29b). In the traffic conditions being simulated, at the start of the simulation, pedestrians 24a, 24b, 25a, 25b, vehicles turning right downwards 28a, 28b, and vehicles turning right upwards 29a, 29b will stop and wait at their initial positions. After that, they will begin to cross. Vehicles going straight to the left 26a, 26b and vehicles going straight to the right 27a, 27b will accelerate sufficiently to the legal speed limit before reaching their initial positions and then begin to cross. The traffic conditions used in the above simulation are just one example and are not limited to this; simulations will be conducted using a variety of traffic conditions.

[0035] As shown in Figure 6, the information processing device 100 acquires the "individual judgment model for straight-moving vehicles (CarS)" 14, which was constructed in "basic model construction" 13, the "individual judgment model for right-turning vehicles (CarR)" 15, and the "integrated table of individual judgments" 16. The "individual judgment model for straight-moving vehicles (CarS)" 14, the "individual judgment model for right-turning vehicles (CarR)" 15, and the "integrated table of individual judgments" 16 are used in "behavior estimation" 39, which will be described later.

[0036] In the "Selection of Traffic Participants" 36 shown in Figure 6, traffic participants who will influence the subject's judgment and decision-making are selected. Specifically, in the "Selection of Traffic Participants" 36, other traffic participants are grouped based on differences in how they interact with the subject, and one traffic participant is selected from each group to influence the subject's judgment and decision-making. The difference in how you cross paths with the subject lies in whether you cross paths with the subject while walking straight down the road, or after turning (turning right) down the road. In the first embodiment, the group that crosses paths with the subject while driving straight on the road is a left-turning straight-ahead vehicle (CarS R )26 and vehicles going straight to the right (CarS L )27 includes, that is, left-facing straight-ahead vehicles 26a, 26b and right-facing straight-ahead vehicles 27a, 27b. In the first embodiment, the group that crosses paths with the subject after turning (right) on the road is a vehicle turning right with a downward direction (CarR U )28 and vehicles turning right upwards (CarR D )29 includes, that is, vehicles turning right with a downward direction 28a, 28b and vehicles turning right with an upward direction 29a, 29b. Therefore, in the case of a group that crosses paths with the subject while traveling straight on the road, one vehicle will be selected from among the left-facing straight-ahead vehicles 26a and 26b, and the right-facing straight-ahead vehicles 27a and 27b. Additionally, for groups that cross paths with the target vehicle after turning right, one vehicle will be selected from vehicles 28a and 28b (downward-facing right turn) and vehicles 29a and 29b (upward-facing right turn).

[0037] The selection method is based on TTCP (Time To Cross Point), selecting the participant with the smallest TTCP from each group. TTCP refers to the time it takes for each participant to reach the point where they cross paths with the target participant from their current location. TTCP will be explained with reference to Figure 7. Figure 7 is a diagram illustrating TTCP according to the first embodiment. As shown in Figure 7, TTCP refers to the time it takes for a vehicle 6 traveling straight to reach point 51 where it intersects with the target person (pedestrian 5). In the first embodiment, a vehicle 27b heading straight to the right is selected from the group that crosses the target vehicle while it is traveling straight on the road, and a vehicle 29a heading right is selected from the group that crosses the target vehicle after it has turned (turned right) on the road.

[0038] In the "Data Conversion" 37 shown in Figure 6, the movement status (position, speed) of the selected vehicle (selected vehicle) is converted into a format that can be used by the basic model. If the direction of travel of the selected vehicle is different (reverse) from when the basic model was constructed, the data is converted so that it moves point-symmetrically around the center of the intersection 53 (Figure 8) to match the direction of travel of the vehicle when the basic model was constructed. Referring to Figure 8, the data conversion of the information processing device 100 according to the first embodiment will be explained. Figure 8 is a diagram illustrating the data conversion of the information processing device 100 according to the first embodiment. The upper part of Figure 8 shows the traffic conditions being simulated, and the lower part of Figure 8 shows the data conversion process. In "Selection of Traffic Participants" 36, vehicles 27b traveling straight to the right shall be selected from the group that intersects the subject while traveling straight on the road, and vehicles 29a traveling right with an upward turn shall be selected from the group that intersects the subject after turning right on the road. The direction of movement of the selected vehicle 29a, which is turning right upwards, is the same as the direction of movement of the vehicle when the basic model was constructed. However, the direction of movement of the selected vehicle 27b, which is going straight to the right, is the opposite of the direction of movement of the vehicle when the basic model was constructed. Therefore, it is necessary to convert the movement state (position, speed, etc.) of the vehicle going straight to the right 27b into a format that the basic model can use. The movement state (position, speed, etc.) of the vehicle going straight to the right 27b is converted into a data state (vehicle 27c going straight to the right after data conversion) such that it is point-symmetric with respect to the center of the intersection 53 (see Figure 8).

[0039] In the "Behavior Estimation" 39 shown in Figure 6, the final decision of the subject is estimated based on the movement status of the vehicle turning right upwards 29a and the vehicle going straight to the right after data conversion, which were selected using the basic model. Specifically, in "Behavior Estimation" 39, the "Individual Decision Model for Straight-Going Vehicles (CarS)" 14, the "Individual Decision Model for Right-Straight-Going Vehicles (CarR)" 15, and the "Integrated Table of Individual Decisions" 16 are used to calculate the probability of the subject's decision (Go, Wait, Undecided) based on the movement status of the selected vehicles: the upward-turning right-turning vehicle 29a and the right-straight-going vehicle 27c after data conversion. The decision with the highest probability is then estimated to be the decision made by the subject. The behavior estimation method calculates the probability of a final decision based on the movement status information of the subject (pedestrian 24a) and the selected vehicles (27c, 29a), and estimates that the decision with the highest probability was the one made by the subject. Specifically, the individual decision model 14 for vehicles going straight (CarS) is used to estimate the subject's (pedestrian 24a's) decision D1(t) regarding vehicle 27c, and the individual decision model 15 for vehicles turning right (CarR) is used to estimate the subject's (pedestrian 24a's) decision D2(t) regarding vehicle 29a. Then, using equation (3), the decisions D1(t) and D2(t) are combined to calculate the probability of each of the final decisions D(t) = (Go, Wait, Undecided), and the decision with the highest probability is estimated to be the final decision made by the subject (pedestrian 24a).

[0040] The speed of the subject's (pedestrian 24a) movement (movement speed v) P (t)) is calculated according to the probability of the final decision. Pedestrian 24a has a small range of movement speed and is expected to accelerate quickly to the target speed. Therefore, in this embodiment, the movement of pedestrian 24a is considered to be at a movement speed v P This was expressed as (t). And the movement speed v of pedestrian 24a. P (t) was defined as the probability-weighted mean sum based on the decision model, as shown in equation (4). However, v- P d This represents the average speed when the probability of each decision state d is 100%.

[0041]

number

[0042] (Regarding the hardware configuration of the information processing device 100) Referring to Figure 9, the hardware configuration of the information processing device 100 according to the first embodiment will be described. Figure 9 is a diagram illustrating the hardware configuration of the information processing device 100 according to the first embodiment. The information processing device 100 includes a communication unit 100a, a ROM 100b, a RAM 100c, a storage unit 100d, a processing unit 100e, and an input / output interface 100f, among other components. Furthermore, the information processing device 100 includes an input device 100g and an output device 100h, which perform data input and output via an input / output interface 100f as external devices.

[0043] The communication unit 100a is equipped with a function for bidirectional communication with other information processing devices. When the communication unit 100a performs bidirectional communication with other information processing devices, it may do so via the information communication network 150, or it may connect directly to the other information processing devices to perform bidirectional communication. The communication unit 100a may use wired communication or wireless communication for communication with other information processing devices.

[0044] ROM100b can be used as a recording device and stores the BIOS (Basic Input Output System), which is necessary for controlling the operation of each functional part of the information processing device 100, as well as various data used by the BIOS. The BIOS is a program that manages the basic input / output functions of the information processing unit 100. It is the first program to run when the information processing unit 100 is powered on, and it controls hardware such as the communication unit 100a, ROM 100b, RAM 100c, storage unit 100d, processing unit 100e, and input / output interface 100f, preparing the OS (Operating System) to start up.

[0045] RAM100c is used in the configuration of the main memory accessed by the processing unit 100e, and is also used to temporarily store various data acquired or generated by the information processing device 100 before storing them in the storage unit 100d.

[0046] The storage unit 100d is implemented using an HDD (Hard Disk Drive), SSD (Solid State Drive), online storage, etc., and stores the OS, the information processing programs described later, other application software, and various data used by these programs. The storage unit 100d also stores various data acquired or generated by the information processing device 100.

[0047] The processing unit 100e includes a CPU (Central Processing Unit), an MPU (Micro Processing Unit), a GPU (Graphics Processing Unit), etc., and is realized by logic circuits and dedicated circuits formed by integrated circuits (IC (Integrated circuit) chips, LSI (Large-Scale Integration)), etc. The input / output interface 100f is an interface for sending and receiving data to and from external devices such as the input device 100g and the output device 100h. The input device 100g includes a keyboard, mouse, etc., and accepts user input for operation of the information processing device 100. The output device 100h includes a monitor, printer, etc., and displays data generated by the information processing device 100 to the user.

[0048] (Functional configuration of the information processing device 100) Next, an example of the functional configuration of the information processing device 100 will be described with reference to Figure 10. Figure 10 is a diagram illustrating the functional configuration of the information processing device 100 according to the first embodiment. The information processing device 100 loads the information processing program, described later, stored in the storage unit 100d, into the main memory, which is composed of RAM 100c or the like. The processing unit 100e accesses the main memory into which the information processing program has been loaded and executes the information processing program. The information processing device 100, by executing an information processing program, has functional units in the processing unit 100e such as a basic model acquisition unit 40, a grouping unit 41, a selection unit 42, a movement state information acquisition unit 43, a movement state information conversion unit 44, an estimation unit 45, and a speed calculation unit 46.

[0049] The basic model acquisition unit 40 acquires a basic model that has been pre-learned of the decisions made by the subject in the basic situation in which a pedestrian 5 is about to cross the crosswalk 4 at intersection 1, a vehicle going straight (vehicle going straight 6) that intersects with the pedestrian 5 while it is going straight, and a vehicle turning right (vehicle turning right 7) that intersects with the pedestrian 5 after it turns right, all of which are about to enter intersection 1. The basic model is constructed using the basic model construction method described in step 13 above. The basic model includes an individual judgment model 14 for vehicles going straight (CarS), an individual judgment model 15 for vehicles going straight to the right (CarR), and an integrated table 16 for individual judgments. The basic model acquisition unit 40 may also be performed by acquiring an individual judgment model 14 for a straight-moving vehicle (CarS) 33, acquiring an individual judgment model 15 for a right-turning vehicle (CarR) 34, and acquiring an integrated table 16 of individual judgments 35.

[0050] The grouping unit 41 divides the multiple traffic participants (26a, 26b, 27a, 27b, 28a, 28b, 29a, 29b) at intersection 20 with traffic conditions different from the basic conditions into a first group that enters intersection 20 by going straight and a second group that enters the intersection by turning right or left. As shown in Figure 5, the straight-going vehicles 26a, 26b, 27a, and 27b are included in the first group, and the right-turning vehicles 28a, 28b, 29a, and 29b are included in the second group. The grouping section 41 may be carried out in the "Selection of traffic participants" 36 described above. Furthermore, the grouping unit 41 groups vehicles when the simulation target is a pedestrian, and groups both vehicles and pedestrians when the simulation target is a vehicle driver. In other words, if the subjects of the simulation are pedestrians, the grouping unit 41 groups vehicles entering the intersection 20 by going straight into the first group, and vehicles entering the intersection 20 by turning right or left into the second group. Furthermore, if the subject of the simulation is a vehicle driver, the grouping unit 41 groups vehicles and pedestrians entering the intersection 20 by going straight into the first group, and vehicles entering the intersection 20 by turning right into the second group.

[0051] The selection unit 42 selects one participant from each of the first and second groups who will reach the point where they intersect with the target person from their current location in the shortest time. The selection unit 42 may also be performed in the "Selection of Traffic Participants" 36 described above, in which case the straight-going vehicle 27b and the right-turning vehicle 29a will be selected.

[0052] The movement status information acquisition unit 43 acquires movement status information indicating the location and movement speed of the selected traffic participants. The position coordinates included in the movement status information of the selected vehicle, straight-moving vehicle 27b, are (X S (t), Y S (t)) is expressed as, and the velocity in the direction of travel is V S It is represented by (t). The position coordinates included in the movement status information of the selected vehicle, right-turning vehicle 29a, are (X R (t), Y R (t)) is expressed as, and the velocity in the direction of travel is V R It is represented by (t). t is the current time in the simulation.

[0053] The movement status information conversion unit 44 converts the movement status information of the selected traffic participant so that the position of the selected traffic participant moves to a point symmetrical position centered on the center 53 of the intersection 20 when the direction of movement of the selected traffic participant is opposite to the direction of movement of the traffic participant in the basic situation. The movement state information conversion unit 44 may also be performed in the "selection of traffic participants" 36 described above, in which case, as described above, the movement state information of the straight-ahead vehicle 27b is converted so that the straight-ahead vehicle 27b moves to the straight-ahead vehicle 27c.

[0054] The estimation unit 45 estimates the decisions made by selected traffic participants by inputting their movement status information into the basic model. The estimation unit 45 estimates that the subject will make one of the following decisions: a decision to go across the crosswalk, a decision to wait for traffic participants to pass, or an undecided decision. It then calculates the probability of each decision and estimates the decision with the highest probability as the decision the subject will make. The estimation unit 45 may be performed in the "behavior estimation" 39 described above.

[0055] The speed calculation unit 46 calculates the subject's movement speed or acceleration according to the probability. The speed calculation unit 46 calculates the speed of the subject's (pedestrian's 24a) movement (movement speed v P (t)) may be calculated according to the probability of the final decision. Pedestrian 24a has a small range of movement speed and is expected to accelerate quickly to the target speed. Therefore, in this embodiment, the movement of pedestrian 24a is considered to be at a movement speed v P This was expressed as (t). And the movement speed v of pedestrian 24a. P (t) was defined as the probability-weighted mean sum based on the decision model, as shown in equation (4). However, v- P d This represents the average speed when the probability of each decision state d is 100%. The speed calculation unit 46 may be performed in the "action estimation" 39 described above. If the subject of the simulation is the driver of a vehicle, the speed calculation unit 46 may calculate the vehicle's speed or acceleration.

[0056] (Regarding the information processing device 100 according to the second embodiment) The information processing device 100 according to the second embodiment will be described with reference to Figures 11 and 12. Figure 11 is a diagram illustrating the basic situation according to the second embodiment, and Figure 12 is a diagram illustrating the traffic situation that is the subject of the simulation according to the second embodiment. The information processing device 100 according to the second embodiment differs from the information processing device 100 according to the first embodiment in that the subject of the simulation is the driver of the straight-moving vehicle 26a instead of a pedestrian (see Figure 12). In describing the information processing device 100 according to the second embodiment, only the differences from the first embodiment will be described in detail, and the parts that are the same as the first embodiment will not be described. The subject of the simulation in the basic situation of intersection 1 is the driver of straight-ahead vehicle 6, and the position coordinates of straight-ahead vehicle 6 are (X S (t), Y S (t)) is denoted as such, and the velocity in the direction of travel is V S Let it be represented by (t). In the traffic situation shown in Figure 12, it is assumed that in "Selection of Traffic Participants" 36, a vehicle going straight 27b and a vehicle turning right 29a were set, and in "Data Conversion" 37, the vehicle going straight 27b was converted to a vehicle going straight 27c (see Figure 8). The individual judgment D1(t) of the driver of the straight-going vehicle 26a towards the straight-going vehicle 27c is estimated using the individual judgment model 14 for the straight-going vehicle (CarS). The individual judgment D2(t) of the driver of the straight-going vehicle 26a regarding the right-turning vehicle 29a is estimated using the individual judgment model 15 for the right-turning vehicle (CarR). The final decision of the driver of the vehicle 26a, which is the subject of the investigation, is estimated by equation (3).

[0057] (Regarding information processing methods and information processing programs) Next, with reference to Figure 13, an information processing program according to a third embodiment of the present invention will be described along with an information processing method. Figure 13 is an example of a flowchart of the information processing program according to this embodiment. The information processing method is executed by the processing unit 10e of the information processing device 10 based on the information processing program. The information processing program includes a basic model acquisition step S40, a grouping step S41, a selection step S42, a movement state information acquisition step S43, and an estimation step S45, among others. The information processing program implements basic model acquisition, grouping, selection, movement status information acquisition, and estimation functions in the processing unit 10e of the information processing device 10. These functions are executed in the order shown in the flowchart of Figure 13, but the order can be changed as appropriate. Since each function overlaps with the descriptions of the various functional units of the information processing device 10 mentioned above, detailed explanations are omitted.

[0058] The basic model acquisition function acquires a basic model that has been pre-trained on the decisions made by the subject in basic situations where a pedestrian is about to cross a crosswalk at an intersection, one vehicle is passing through while going straight, and one vehicle is passing through after turning right, both of which are about to enter the intersection (S40: Basic Model Acquisition Step).

[0059] The grouping function divides multiple traffic participants at an intersection with traffic conditions different from the basic situation into two groups: a first group that enters the intersection by going straight and a second group that enters the intersection by turning right or left (S41: Grouping step).

[0060] The selection function selects one traffic participant from each of the first and second groups who will reach the point where they intersect with the target person from their current location in the shortest time (S42: Selection step).

[0061] The movement status information acquisition function acquires movement status information indicating the position and speed of the selected vehicle (S43: Movement status information acquisition step).

[0062] The estimation function estimates the decision made by the subject by inputting the movement status information of the selected vehicle into the basic model (S45: estimation step).

[0063] (Regarding other embodiments of information processing methods and information processing programs) Next, with reference to Figure 14, the information processing program according to the fourth embodiment will be described along with the information processing method according to the other embodiments. Figure 14 is an example of a flowchart of the information processing program according to the other embodiments. The flowchart of the information processing program according to another embodiment shown in Figure 14 differs from the flowchart of the information processing program shown in Figure 13 in that it includes the addition of a movement state information conversion step S44 and a speed calculation step S46. Information processing methods according to other embodiments are executed by the processing unit 100e of the information processing device 100 based on the information processing program according to another embodiment shown in Figure 14. The information processing program according to another embodiment shown in Figure 14 includes a basic model acquisition step S40, a grouping step S41, a selection step S42, a movement state information acquisition step S43, a movement state information conversion step S44, an estimation step S45, and a speed calculation step S46, among others.

[0064] The information processing program according to another embodiment shown in Figure 14 implements functions such as basic model acquisition, grouping, selection, movement state information acquisition, movement state information conversion, estimation, and speed calculation in the processing unit 100e of the information processing device 100. These functions are executed in the order shown in the flowchart of Figure 14, but the order can be changed as appropriate. Below, we will describe only the differences between the information processing method and information processing program shown in Figure 13 and the information processing method and information processing program shown in Figure 13, which are based on other embodiments of the information processing method and information processing program shown in Figure 14. Furthermore, since each function overlaps with the descriptions of the various functional units of the information processing device 100 mentioned above, detailed explanations will be omitted.

[0065] The movement state information conversion function converts the movement state information of the selected vehicle so that, when the direction of movement of the selected vehicle is opposite to the direction of movement of the vehicle in the basic state, the position of the selected vehicle moves to a point symmetrical position centered on the center of the intersection (S44: Movement state information conversion step).

[0066] The speed calculation function calculates the subject's movement speed or acceleration based on probability (S46: Speed ​​calculation step).

[0067] According to the embodiment described above, the basic model constructed under the basic conditions of intersection 1 can be used to estimate pedestrian decisions under various traffic conditions at intersection 20. Therefore, it is not necessary to construct a machine learning model to estimate pedestrian decisions for each traffic condition, and the effort and cost of acquiring training data and conducting training can be reduced.

[0068] Furthermore, according to the embodiment described above, using the basic model constructed under the basic conditions of intersection 1, it is possible to estimate the speed of movement when it is determined that a pedestrian can pass through the crosswalk 23 under various traffic conditions at intersection 20.

[0069] Furthermore, according to the embodiment described above, the basic model constructed under the basic conditions of intersection 1 can be used to estimate the judgment of drivers of vehicles going straight at intersection 20 under various traffic conditions at intersection 20. Therefore, it is not necessary to construct a machine learning model to estimate the judgment of drivers of vehicles going straight for each traffic condition, and the effort and cost of acquiring training data and conducting training can be reduced.

[0070] Furthermore, according to the above-described embodiment, the training data necessary for constructing the basic model can be acquired using a multiplayer simulator, and thus can be acquired and stored as reproducible data.

[0071] Furthermore, according to the above-described embodiment, a machine learning model can be provided that can estimate the judgment of pedestrians on a crosswalk in various traffic conditions at the intersection 20.

[0072] Furthermore, according to the above-described embodiment, it is possible to provide a machine learning model that can estimate the judgment of drivers of vehicles going straight in various traffic conditions at intersection 20.

[0073] It should be noted that the present invention is not limited to the information processing apparatus 10, information processing method, and information processing program according to the above-described embodiment, and can be implemented by various other modifications or applications without departing from the spirit of the invention as described in the claims. Furthermore, although the above embodiment uses the term "data," the term "data" can be replaced with "information," and the term "information" can be replaced with "data." Furthermore, the vehicles in the above-described embodiment may include not only four-wheeled vehicles but also two-wheeled vehicles and mopeds. [Explanation of Symbols]

[0074] 1 Intersection 2 Priority road 3 Non-priority road 4 Pedestrian crossing 5 Pedestrians 6. Straight-going vehicles 7 Right turn vehicle Measurement data of the behavior of 10 people 11. Data Preprocessing 12 Preprocessed Data 13. Building the Basic Model 14. Individual decision model for vehicles traveling straight (CarS) 15 Individual judgment model for vehicles going straight right (CarR) 16. Integrated Table for Individual Decisions 20 Intersections 21 Transverse roads 21a Left-facing road 21b Right-facing road 22 Longitudinal roads 22a Uphill road 22b Downhill road 23 Pedestrian crossing 24. Pedestrian attempting to cross while facing downwards 24a Pedestrian 24b Pedestrian 25. Pedestrian attempting to cross while facing upwards 25a Pedestrian 25b Pedestrian 26. Left-facing vehicles going straight (CarS R ) 26a Left-facing vehicles going straight 26b Left-facing vehicles going straight 27 Right-facing vehicles going straight (CarS L ) 27a Right-facing vehicle going straight 27b Right-facing vehicle going straight 27c Right-facing straight vehicle after data conversion 28. Vehicles turning right with a downward turn (CarR U ) 28a Vehicle turning right with a downward turn 28b Vehicles turning right with a downward direction 29 Upward-facing right-turning vehicles (CarR D ) 29a Upward-facing right-turning vehicle 29b Upward turning vehicle 33. Acquisition of an individual judgment model for straight-moving vehicles (CarS) 34. Acquisition of an individual judgment model for vehicles traveling straight to the right (CarR). 35. Obtaining the integrated table for individual decisions. 36. Selection of transportation participants 37 Data Conversion 38 Converted data 39. Behavioral Estimation 40 Basic Model Acquisition Section 41 Grouping section 42 Selection Department 43. Unit for acquiring movement status information 44 Movement status information conversion unit 45 Estimation part 46 Speed ​​calculation section 51 points 52. Movement status of a vehicle traveling straight at time t. 53 Intersection Center 65 Straight-ahead vehicles 74 Straight-ahead vehicles 100 Information Processing Devices 100a Communications Department 100b ROM (Read Only Memory) 100c RAM(Randum Access Memory) 100d storage section 100e Processing Unit 100f Input / Output Interface 100g input device 100h output device 150 Information and Communication Networks

Claims

1. An information processing device that simulates traffic participants attempting to cross an intersection without traffic signals, and estimates the decisions of the participants according to the traffic conditions at the intersection, A basic model acquisition unit acquires a basic model that has been pre-learned of the judgments made by the subject in the basic situation in which a pedestrian attempting to cross the crosswalk at the intersection is about to enter the intersection, one vehicle that is going straight and crossing with the pedestrian, and one vehicle that is turning right and crossing with the pedestrian. A grouping unit that divides multiple traffic participants at the intersection in traffic conditions different from the basic conditions into a first group that enters the intersection by going straight and a second group that enters the intersection by turning right or left, A selection unit selects one traffic participant from each of the first and second groups who will reach the point where they intersect with the target person from their current location in the shortest time. A movement status information acquisition unit acquires movement status information indicating the movement status of the selected traffic participant, including the location and speed of movement. An estimation unit that estimates the decisions made by the selected traffic participant by inputting the movement status information of the selected traffic participant into the basic model, An information processing device characterized by comprising:

2. The estimation unit estimates one of the following as the judgment made by the subject: that the subject is able to pass the pedestrian crossing, that the subject will wait for the traffic participants to pass, or that the subject has not yet made a judgment. The information processing device according to claim 1, characterized in that it calculates the probability of each of the judgments made by the subject, and estimates the judgment with the highest probability as the judgment made by the subject.

3. The information processing apparatus according to claim 2, further comprising a speed calculation unit that calculates the movement speed or acceleration of the subject according to the aforementioned probability.

4. The information processing apparatus according to claim 1, further comprising a movement state information conversion unit that converts the movement state information of the selected traffic participant so that, when the direction of movement of the selected traffic participant is opposite to the direction of movement of the traffic participant in the basic situation, the position of the selected traffic participant moves to a point symmetrical position centered on the center of the intersection.

5. The information processing device according to claim 1, characterized in that the subject is a pedestrian or a driver operating a vehicle.

6. An information processing method that simulates traffic participants attempting to cross an intersection without traffic signals, and estimates the decisions of the participants according to the traffic conditions at the intersection, Computers A basic model acquisition step involves acquiring a basic model that has been pre-trained on the judgments made by the subject in a basic situation where a pedestrian is about to cross a crosswalk at the intersection, one vehicle is passing by while going straight, and another vehicle is passing by the pedestrian after turning right, and both are about to enter the intersection. A grouping step that divides multiple traffic participants at the intersection in traffic conditions different from the basic conditions into a first group that enters the intersection by going straight and a second group that enters the intersection by turning right or left, A selection step in which one traffic participant is selected from each of the first group and the second group who will reach the point where they intersect with the subject person from their current location in the shortest time, A movement status information acquisition step involves acquiring movement status information that indicates the movement status of the selected traffic participant, including its position and speed. An estimation step in which the decision made by the subject is estimated by inputting the movement status information of the selected traffic participant into the basic model, An information processing method characterized by performing the following.

7. An information processing program that simulates traffic participants attempting to cross an intersection without traffic signals, and estimates the decisions of the participants according to the traffic conditions at the intersection, On the computer, A basic model acquisition function that acquires a basic model that has been pre-learned of the judgments made by the subject in the basic situation in which a pedestrian attempting to cross the crosswalk at the intersection is about to enter the intersection, one vehicle that is going straight and crossing with the pedestrian, and one vehicle that is turning right and crossing with the pedestrian, are about to enter the intersection. A grouping function that divides multiple traffic participants at the intersection with traffic conditions different from the basic conditions into a first group that enters the intersection by going straight and a second group that enters the intersection by turning right or left, A selection function that selects one traffic participant from each of the first and second groups who will reach the point where they intersect with the target person from their current location in the shortest time, A function for acquiring movement status information that acquires movement status information indicating the location and movement speed of the selected traffic participant, An estimation function that estimates the decisions made by the selected traffic participants by inputting their movement status information into the basic model, An information processing program characterized by achieving this.

Images