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

Abstract

This invention provides an information processing device that enables the modeling of individual behavior from both external and internal perspectives. [Solution] The information processing device is an information processing device that generates a judgment model of the behavior of traffic participants when crossing a traffic environment and executes a simulation, and comprises: an acquisition unit that acquires characteristic information that characterizes traffic participants, which is the premise of the judgment model; a first generation unit that, when simulating the behavior of traffic participants when crossing a traffic environment, generates a behavior model of traffic participants as a reaction to the understanding of the surrounding situation as an external factor when crossing a traffic environment, based on the characteristic information; a second generation unit that, when simulating the behavior of traffic participants when crossing a traffic environment, generates a cognitive model of traffic participants as a reaction to the understanding of the surrounding situation as an internal factor when crossing a traffic environment, based on the characteristic information; and a third generation unit that generates a judgment model of the behavioral judgment of traffic participants when crossing a traffic environment, based on the behavior model and the cognitive model.

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 that model individual behavior from both external factors (surrounding circumstances) and internal factors (human behavioral psychology). [Background technology]

[0002] There is a traffic flow simulation device that simulates the flow of vehicles traveling on roads. This simulation device considers the characteristics of the vehicles and the individual characteristics of the drivers of those vehicles when simulating traffic flow. The individual characteristics of the drivers represent tendencies regarding the drivers' perception and judgment, and have parameters such as the degree to which they recognize stop signs, the degree to which they pass stop signs, the accuracy of their judgment, their desired speed, the degree to which they press the accelerator, and the stopping position relative to the stop line. In other words, the simulation device adjusts the likelihood of human error by the drivers by setting each parameter of the drivers' individual characteristics, thereby changing the movement of the vehicles.

[0003] Traffic flow is a concept that views the movement of numerous vehicles on a road as a flow, and is suitable for analyzing phenomena such as congestion. For example, it can be used to simulate various traffic situations by reproducing the movement of vehicles on a computer. For example, the information processing apparatus, information processing method, and information processing program disclosed in Patent Document 1 are said to be capable of estimating relevant parameters about a person's behavior.

[0004] While it is desirable to model individual behavior from both external factors (surrounding circumstances) and internal factors (people's behavioral psychology), and to represent differences in personality with parameters to express diverse models of people, allowing for variations in combinations with others, the information processing device, information processing method, and information processing program disclosed in Patent Document 1 do not take these points into consideration and cannot be said to adequately meet social needs. [Prior art documents] [Patent Documents]

[0005] [Patent Document 1] Japanese Patent Publication No. 2023-42027 [Overview of the project] [Problems that the invention aims to solve]

[0006] Therefore, the present invention aims to provide an information processing device, an information processing method, and an information processing program that can express differences in personality and attributes using parameters, and allow for variations in combinations with others to represent various behaviors of the masses, thereby enabling the modeling of individual behavior from both external factors (surrounding circumstances) and internal factors (people's behavioral psychology). [Means for solving the problem]

[0007] In other words, the information processing device according to the first embodiment is an information processing device that generates a judgment model of the behavior of traffic participants when crossing a traffic environment and performs a simulation, comprising: an acquisition unit that acquires characteristic information that characterizes traffic participants, which is the premise of the judgment model; a first generation unit that, when simulating the behavior of traffic participants when crossing a traffic environment, generates a behavior model of traffic participants as a reaction to the understanding of the surrounding situation as an external factor when crossing a traffic environment, based on the characteristic information; a second generation unit that, when simulating the behavior of traffic participants when crossing a traffic environment, generates a cognitive model of traffic participants as a reaction to the understanding of the surrounding situation as an internal factor when crossing a traffic environment, based on the characteristic information; and a third generation unit that generates a judgment model of the behavioral judgments made by traffic participants when crossing a traffic environment, based on the behavior model and the cognitive model.

[0008] The second embodiment is an information processing device according to the first embodiment, in which the decision model is a decision model that integrates individual decisions and motion for multiple vehicles and a decision model that integrates individual decisions and control, and may be generated by integrating the decision of whether to cross the pedestrian crossing or wait for the first vehicle and the decision of whether to cross the pedestrian crossing or wait for the second vehicle which is different from the first vehicle, based on the visual behavior of the traffic participant.

[0009] A third embodiment is an information processing device according to the second embodiment, wherein the decision model includes a route decision model that indicates the route taken by a traffic participant when crossing the traffic environment, and the participant proceeds straight if there are no obstacles on the route when crossing the traffic environment, and takes evasive action if a collision with an obstacle is expected on the route, and the decision is made by comparing the straight-ahead action and the evasive action.

[0010] In the fourth embodiment, the information processing device according to the first embodiment may be configured such that the decision model is generated by clustering the behavioral tendencies of traffic participants using inverse reinforcement learning with characteristic information of traffic participants.

[0011] A fifth aspect is that, in the information processing device according to the first aspect, traffic participants may include pedestrians and cyclists.

[0012] A sixth embodiment is an information processing device according to the first embodiment, wherein the third generation unit generates a judgment model for the behavioral judgments made by a traffic participant when crossing a traffic environment, based on a behavioral model and cognitive model when the traffic participant is a cyclist.

[0013] The seventh embodiment is an information processing device according to the sixth embodiment, wherein the surrounding circumstances include pedestrians, bicycles and automobiles, and the acquisition unit acquires cognitive information, judgment information, and behavioral information of humans regarding perception as a reaction to pedestrians, judgment information regarding judgments and behavioral information regarding actions, cognitive information, judgment information, and behavioral information regarding perception as a reaction to bicycles, and cognitive information, judgment information, and behavioral information regarding perception as a reaction to automobiles, and the behavioral model, judgment model, and cognitive model may change depending on the combination of surrounding circumstances.

[0014] The eighth aspect is an information processing device according to the first aspect, wherein the acquisition unit may acquire, as attribute information, information relating to at least one of the results of a questionnaire or test previously conducted on the person regarding the person's age, the number of years of experience driving a vehicle and the frequency of driving, the person's personality, and the person's athletic ability.

[0015] The information processing method relating to the ninth aspect is an information processing method that generates a judgment model of the behavior of traffic participants when crossing a traffic environment and performs a simulation, wherein the computer performs an acquisition step of acquiring characteristic information that characterizes traffic participants, which is the premise of the judgment model; a first generation step of generating a behavior model of traffic participants as a reaction to the understanding of the surrounding situation as an external factor when crossing a traffic environment, based on the characteristic information when simulating the behavior of traffic participants when crossing a traffic environment; a second generation step of generating a cognitive model of traffic participants as a reaction to the understanding of the surrounding situation as an internal factor when crossing a traffic environment, based on the characteristic information when simulating the behavior of traffic participants when crossing a traffic environment; and a third generation step of generating a judgment model of the behavioral judgments made by traffic participants when crossing a traffic environment, based on the behavior model and the cognitive model.

[0016] The information processing program according to the tenth embodiment is an information processing program that generates a judgment model of the behavior of traffic participants when crossing a traffic environment and executes a simulation, and causes a computer to execute an acquisition function that acquires characteristic information that characterizes traffic participants, which is the premise of the judgment model; a first generation function that, when simulating the behavior of traffic participants when crossing a traffic environment, generates a behavior model of traffic participants as a reaction to the understanding of the surrounding situation as an external factor when crossing a traffic environment, based on the characteristic information; a second generation function that, when simulating the behavior of traffic participants when crossing a traffic environment, generates a cognitive model of traffic participants as a reaction to the understanding of the surrounding situation as an internal factor when crossing a traffic environment, based on the characteristic information; and a third generation function that generates the judgment model of the behavioral judgment of traffic participants when crossing a traffic environment, based on the behavior model and the cognitive model. [Effects of the Invention]

[0017] The information processing device according to the present invention is an information processing device that generates a judgment model of the behavior of traffic participants when crossing a traffic environment and performs a simulation, and is characterized by comprising: an acquisition unit that acquires characteristic information that characterizes traffic participants, which is the premise of the judgment model; a first generation unit that, when simulating the behavior of traffic participants when crossing a traffic environment, generates a behavior model of traffic participants as a reaction to the understanding of the surrounding situation as an external factor when crossing a traffic environment, based on the characteristic information; a second generation unit that, when simulating the behavior of traffic participants when crossing a traffic environment, generates a cognitive model of traffic participants as a reaction to the understanding of the surrounding situation as an internal factor when crossing a traffic environment, based on the characteristic information; and a third generation unit that generates a judgment model of the behavioral judgment of traffic participants when crossing a traffic environment, based on the behavior model and the cognitive model. As such, it is possible to represent differences in individuality and differences in attributes with parameters and to represent various mass behaviors by allowing variations in combinations with others, and it is possible to model individual behavior from both external factors (surrounding situation) and internal factors (human behavioral psychology).

[0018] In addition, similar to the information processing apparatus according to the present invention described above, the information processing method and the information processing program according to the present invention can express differences in personality and attributes as parameters, and can have variations in combinations with others, thereby making it possible to express various behaviors of the general public. It becomes possible to model individual behaviors from both aspects of external factors (surrounding situations) and internal factors (human behavior psychology).

Brief Description of the Drawings

[0019] [Figure 1] FIG. 1 is a diagram for explaining the other-person reaction type individual behavior model according to the present embodiment. [Figure 2] FIG. 2 is a table showing the overall configuration of individual behaviors in the simulation according to the present embodiment. [Figure 3] FIG. 3 is a diagram showing a crossing judgment model by integrating individual judgments for a plurality of vehicles according to the present embodiment. [Figure 4] FIG. 4 is a diagram for explaining a judgment model incorporating visual behavior according to the present embodiment. [Figure 5] FIG. 5 is a diagram for explaining a judgment model incorporating visual behavior according to the present embodiment. [Figure 6] FIG. 6 is a diagram for explaining the observation of pedestrian behavior under the influence of a smartphone operation task according to the present embodiment. [Figure 7] FIG. 7 is a diagram for explaining the observation of pedestrian behavior under the influence of a smartphone operation task according to the present embodiment. [Figure 8] FIG. 8 is a diagram for explaining the observation of pedestrian behavior under the influence of a smartphone operation task according to the present embodiment. [Figure 9] FIG. 9 is a diagram showing an example of cooperation between the information processing apparatus according to the present embodiment and a simulator. [Figure 10] FIG. 10 is a diagram for explaining a crossing judgment model generated from the相关性 of behaviors and attributes according to another embodiment. [Figure 11]Figure 11 is a diagram illustrating the pedestrian crossing decision model according to this embodiment. [Figure 12] Figure 12 illustrates a simulation using a pedestrian model to which attribute parameters according to this embodiment have been applied. [Figure 13] Figure 13 shows the behavioral characteristics of cyclists. [Figure 14] Figure 14 shows the behavioral characteristics of cyclists. [Figure 15] Figure 15 is a diagram illustrating the bicycle simulator according to this embodiment. [Figure 16] Figure 16 is a diagram illustrating the model configuration and the scene to be modeled according to this embodiment. [Figure 17] Figure 17 is a diagram illustrating the model configuration and the scene to be modeled according to this embodiment. [Figure 18] Figure 18 illustrates the explanatory variables of the cyclist judgment model according to this embodiment. [Figure 19] Figure 19 shows the relationship between the deceleration behavior of a cyclist and distance according to this embodiment. [Figure 20] Figure 20 is a block diagram showing the functions of the information processing device according to this embodiment. [Figure 21] Figure 21 is an example of a flowchart of the information processing program according to this embodiment. [Modes for carrying out the invention]

[0020] An example of an information processing device 10 according to one embodiment of the present disclosure will be described with reference to Figures 1 to 21. The information processing device 10 is an information processing device that generates a judgment model of the behavior of traffic participants when traversing a traffic environment and executes a simulation. Traffic participants refer to all individuals, including pedestrians, cyclists, automobiles, and new mobility devices, who travel on roads and other public transportation. The traffic environment refers to the entire environment in which these traffic participants interact. The information processing device 10 (see Figure 20) is a so-called computer, also known as a server, personal computer (hereinafter referred to as PC), notebook PC, tablet PC, or smartphone.

[0021] First, we will explain the overview of the Other-Reactive Individual Behavior Model 1, referring to Figure 1. Figure 1 is a diagram illustrating the Other-Reactive Individual Behavior Model. Humans have various attributes, characteristics, and psychology, as well as various means of movement. Attributes include, for example, male, female, child, and elderly. Characteristics include, for example, aggressive and conservative. Psychology includes, for example, calm and anxious.

[0022] Human behavior is influenced by the actions of others. In other words, people perceive the actions of others, judge what action they should take, and then actually act. This sequence of cognition → judgment → action allows for the creation of predictive models that forecast human behavior. Individual differences in behavior contribute to judgment (judgments take into account individual cognitive and motor abilities), so by representing individual differences with parameters and changing the combinations with others, various behaviors can be represented. In short, even in the same scenario, different people will behave differently.

[0023] Referring to Figure 2, the overall configuration of the individual behavior model 1 in the simulation will be explained. Figure 2 is a diagram showing the overall configuration of the individual behavior model 1 in the simulation according to this embodiment. Simulation means to carry out something in a simulated manner, and in this embodiment, it includes the meaning of computer simulation, which means to simulate or test the target phenomenon on a computer, and to analyze and measure the phenomenon that occurs in a virtual space constructed on the computer. Specifically, the simulation is performed under constantly changing environmental conditions, and the calculated or measured results are accumulated each time.

[0024] As mentioned above, humans possess a variety of attributes and psychological traits. Here, we categorize these attributes. For example, attributes can be categorized into groups, such as elderly groups and children's groups. Attributes can also be categorized by region, such as urban residents and sparsely populated residents. Furthermore, attributes can be categorized by differences in lifestyle, such as the wealthy and the poor. These human attributes are collected through questionnaires, including surveys and tests. The attribute data collected through questionnaires, along with data on gait and cognitive function, are calculated and analyzed to express them as parameters. From these parameter representations, a cognitive model 2 incorporating visual behavior is generated. Cognitive model 2 refers to a pattern that derives the result of an object performing cognition in a virtual space based on avatar measurement data, artificial intelligence, and mathematically modeled data. Specifically, it is a pattern that derives the cognition of whether to recognize only vehicles coming from the front or also vehicles coming from the left. Additionally, an action intention estimation model 3 is generated by calculating and analyzing attribute data collected through questionnaires, along with data on gait and cognitive function. The behavioral intention estimation model 3 first estimates situational awareness and risk assessment from attitude, and then estimates risk assessment from situational awareness. The behavioral intention is estimated from risk assessment, psychological utility, and economic utility. Surrounding circumstances are also considered. Surrounding circumstances include, for example, the situation to the left and right of a pedestrian, and the situation of multiple vehicles and other pedestrians. Based on behavioral intention estimation model 3, cognitive model 2, and surrounding circumstances, decision model 4 is generated. Decision model 4 refers to a pattern that derives the result of an object making an action decision using artificial intelligence in a virtual space. Decision model 4 includes crossing decision model 5 and route decision model 6. Decision model 4 is a pattern that derives the result of making an action decision when crossing a crosswalk or going to a destination, for example, a pattern that derives the result of deciding whether to cross immediately when the light turns green or to check left and right before crossing, and a pattern that derives the result of deciding whether to go straight to the destination or avoid obstacles. These models are mathematical models.

[0025] Next, with reference to Figure 3, a cross-sectional decision model based on the integration of individual decisions for multiple vehicles according to this embodiment will be described. Figure 3 is a diagram illustrating the cross-sectional decision model based on the integration of individual decisions for multiple vehicles according to this embodiment. Cross-sectional decision models, which integrate individual decisions, can be divided into two types: integrated decision-and-motion models used for pedestrians and cyclists, and integrated decision-and-control models used for automobiles. The left side of Figure 3 shows the unified decision model (conventional model). This model considers simultaneously whether a pedestrian standing in front of a crosswalk should cross or wait for vehicles coming from both the front and the left. On the other hand, the right side of Figure 3 shows the integrated decision model. In this model, the decision of whether to cross or wait for vehicles coming from the front and the decision of whether to cross or wait for vehicles coming from the left are made separately, and these individual decisions are integrated. These individual decisions are based on visual behavior, which is then modeled. This makes it possible to represent crossing decisions using a small-scale model.

[0026] Next, we will explain a pedestrian crossing decision model that incorporates visual behavior, referring to Figures 4 and 5. Figures 4 and 5 are diagrams illustrating a pedestrian crossing decision model that incorporates visual behavior.

[0027] The upper part of Figure 4 shows visual behavior → individual judgment → final judgment. t represents time, and t-1 represents the time immediately preceding t. O is the input (in this case, gaze, what is currently being looked at). t And what you're looking at in the previous time is O t-1 The symbols represent the following: X represents external factors (such as the speed and position of other traffic participants), S represents a vehicle going straight, R represents a vehicle turning right, and D represents a judgment. Based on the judgment of external factors X and the judgment of the line of sight O, individual judgments are made regarding the vehicle going straight and the vehicle turning right. These individual judgments are then integrated to make a final judgment. Final judgment D t Based on operation M t The process was carried out, and the final decision was D. t-1 Based on operation M t-1 It will take place.

[0028] The lower part of Figure 4 is a graph showing the likelihood of deciding whether to cross or wait over time. The line represents the model's estimated decision, and the background represents the observed decision. As shown in the graph, the left side represents the state before the decision, i.e., before deciding whether to wait or go, where the probability of "Undecided" is 1. The middle part represents the waiting state, where the probability of "WAIT" increases to 0.6. The right side represents the crossing state, where the probability of "GO" increases from 0.5 to 1.

[0029] Figure 5 is a graph showing the relationship between visual behavior and vehicles. A pedestrian is standing in front of a crosswalk, with vehicles approaching from the left and right. The second graph from the top shows the yaw angle of the HMD (head-mounted display), that is, how much the subject's head is tilted. + indicates a tilt to the right, and - indicates a tilt to the left. At 2 seconds, it is +, and at 3 seconds, it is -, indicating that the subject is checking left and right. After that, the yaw angle of the HMD is -, indicating that the subject is looking to the left, that is, fixating on the vehicle approaching from the left. Next, at 6 seconds, it is +, and at 7 seconds, it is -, indicating that the subject is checking left and right. Next, from 7 seconds to 11 seconds, it is +, indicating that the subject is fixating on the vehicle approaching from the right. After that, it becomes - again and then +, indicating that the subject is checking left and right a third time. After that, it becomes 0, so it is thought that the subject crossed the crosswalk looking straight ahead. Thus, a certain pattern can be observed in visual behavior, and it can be seen that it changes in accordance with the relationship with vehicles.

[0030] Next, with reference to Figures 6 to 8, we will explain the results of observing pedestrian behavior under the influence of a smartphone operation task. Figures 6 to 8 illustrate the results of observing pedestrian behavior under the influence of a smartphone operation task. Since cognitive resources are insufficient when walking while operating a smartphone, we used a VR environment to observe and model the characteristics of pedestrians walking while operating a smartphone.

[0031] Figure 6 is a graph showing a comparison of the crossing time of pedestrians using smartphones while walking compared to pedestrians not using smartphones. Crossing time refers to the time from when a pedestrian passes the crosswalk until a vehicle passes it. The left side shows the average crossing time of pedestrians not using smartphones, which is 6.85 seconds, while the right side shows the average crossing time of pedestrians using smartphones, which is 6.38 seconds. Nearly 90% of pedestrians using smartphones while walking experienced a decrease in crossing time.

[0032] Figure 7 is a graph showing the results of comparing the lateral variability of pedestrians crossing a crosswalk while using a smartphone with the lateral variability of the crossing time of pedestrians walking without using a smartphone. The left side shows the lateral variability of pedestrians walking without using a smartphone, and the right side shows the lateral variability of pedestrians walking while using a smartphone. Since the range from -2m to 2m, which represents the x-position of the pedestrian, is larger on the right side, it can be understood that the lateral variability is greater on the right side, which represents pedestrians walking while using a smartphone.

[0033] Figure 8 is a table showing the simulation results of how long it took for subjects wearing VR to notice an approaching vehicle, under different conditions: with and without a smartphone, and with different approaches from the right, left, and both sides. When a vehicle was approaching from the right, the time taken to notice an approaching vehicle was 0.671 with a smartphone compared to 0.642 without a smartphone. When a vehicle was approaching from the left, the time taken was 0.712 with a smartphone compared to 0.660 without a smartphone. When vehicles were approaching from both sides, the time taken was 0.655 with a smartphone compared to 0.632 without a smartphone. This confirms that judgment becomes ambiguous.

[0034] Next, we will explain the combination of the pedestrian crossing judgment model and the route judgment model, referring to Figure 9. Figure 9 is a diagram illustrating the combination of the pedestrian crossing judgment model and the route judgment model.

[0035] Figure 9 shows an example of integration between an information processing device and a simulator. The left side of Figure 9 shows the traffic flow simulation software before a simulation is performed with the model replaced by a pedestrian. The right side shows the software after a simulation is performed with the model replaced by a pedestrian. Replacing the software model with a pedestrian means changing the walking speed from approximately 1 m / s to approximately 1.5 m / s. Before the model replacement, the target pedestrian ▲ follows the pedestrian 〇 in front of them, so they do not step onto the roadway, and such behavior cannot be simulated with a typical model. However, by replacing the model and changing the speed, pedestrian ▲ overtakes pedestrian 〇, causing pedestrian ▲ to step onto the roadway, resulting in a near-miss with a vehicle ■. In other words, by running a simulation with the software model replaced by a pedestrian, it is possible to extract whether or not a new near-miss occurred. Even with the same simulation, different evaluation results can be obtained by changing human behavior.

[0036] Referring to Figure 10, the cross-sectional decision model generated from the relationship between behavior and attributes according to this embodiment will be described. Figure 10 is a diagram illustrating the cross-sectional decision model generated from the relationship between behavior and attributes according to this embodiment.

[0037] Inverse reinforcement learning is performed on the behavioral data 7 to feature the behaviors. Unlike reinforcement learning, which searches for a set optimal solution, inverse reinforcement learning is a technique that learns the optimization index by treating the decision history of an expert as the optimal solution. Through inverse reinforcement learning, behavioral tendencies are clustered, and the behavioral tendencies are classified by estimating intention using a reward function. Here, the reward refers to an external signal that the learner (agent) uses to judge whether its own behavior is good or bad.

[0038] Furthermore, cross-sectional intentions are estimated by linking individual characteristics 8 with attributes. Individual characteristics include attributes and physical abilities, and information about attributes is obtained through questionnaires. Attribute information includes, for example, age and assertiveness. Physical abilities include, for example, effective field of view (UFOV) and walking ability. These individual characteristics are linked to attributes using personality assessments, assertiveness assessments, and risk-taking tendencies to estimate cross-sectional intentions such as whether to go first or yield.

[0039] Furthermore, a human behavior model, including the model and parameters, is constructed from the behavioral data 7. The relationship between behavioral tendency clustering and cross-sectional intentions is analyzed, and the range of parameters is selected based on the results of this analysis.

[0040] Next, with reference to Figure 11, the pedestrian crossing decision model of the information processing device according to this embodiment will be explained. Figure 11 is a diagram illustrating the pedestrian crossing decision model of the information processing device according to this embodiment. The upper left of Figure 11 is an unsignalized intersection where vehicles are stopped. In the lower left and lower center, vehicles are not slowing down. In the situations on the left and in the center, the pedestrian model itself does not change, but independently changes its decision and actions based on the changes in the movement of surrounding vehicles. In the upper left case, the vehicle is stopped, so the pedestrian crosses. In the lower left case, the pedestrian judges that they can be yielded to the vehicle because it has slowed down, and begins to cross. In the center case, the model decides whether it is not possible to cross and waits, or turns back after starting to cross, and acts accordingly. In contrast, on the right side, the behavior of the surrounding cars does not change and they travel at a constant speed. In this case, even though the model is the same, the parameters are changed: in the upper right, the setting is aggressive, meaning the model judges that it is possible to cross before the vehicle passes and begins to cross; in the lower right, the setting is conservative, meaning the model judges that it is not possible to cross before the vehicle passes and waits for the vehicle to pass before crossing. As shown above, it is possible to implement in a single model how changes in the surrounding environment and the state of pedestrians can lead to changes in behavior.

[0041] Next, with reference to Figure 12, a simulation using a pedestrian model to which the attribute parameters of the information processing device according to this embodiment have been applied will be described. Figure 12 is a diagram illustrating a simulation using a pedestrian model to which the attribute parameters of the information processing device according to this embodiment have been applied. The pedestrian in the software that simulates human walking behavior is replaced with 10 pedestrian models, the type is set using the GUI, and the behavior of the set type is shown. The 10 pedestrian models include an aggressive pedestrian model, a conservative pedestrian model, an intermediate pedestrian model, and the aforementioned software pedestrian. The parameters of the pedestrian model are set and the simulation is executed. The lower right shows the parameter settings of the pedestrian model. A parameter film is selected and the start point, passing point, end point, walking speed, attributes, and crossing area are entered. Information regarding vehicles is entered as needed, and the model acts while making judgments according to its attributes in this walking section.

[0042] Next, the behavioral characteristics of cyclist 9 are shown with reference to Figures 13 and 14. Figures 13 and 14 are diagrams illustrating the behavioral characteristics of cyclist 9. As shown in Figure 13, cyclist 9, or bicycle rider 9, is an intermediate entity between pedestrians and automobiles in terms of both speed and degree of freedom of movement, and there are individual differences in the recognition of rules. As shown in Figure 14, bicycle rider 9 frequently interacts with both pedestrians 11 and automobiles 12. For example, this includes riding against traffic or failing to stop, as shown at the bottom of Figure 14. The diversity of movement and the frequency of interactions make modeling highly important.

[0043] Next, the bicycle simulator 16 according to this embodiment will be described with reference to Figure 15. Figure 15 is a diagram illustrating the bicycle simulator 16 according to this embodiment. This bicycle simulator expresses realism in a virtual space for the construction of a cyclist model. The upper left shows the platform 17 of the bicycle simulator 16. The maximum tilt angle is set to 6° to the left and right to enable leaning operations. The lower left shows the steering block 18. The maximum rotation angle is set to 34° to the left and right to reproduce the handlebar restoring force and handlebar load. The upper right shows pedal load control by the cycling trainer. The pedal load during uphill riding is reproduced by transmitting gradient information from the software to the trainer. The lower right shows the implementation of a power meter built into the pedal axle. This makes it possible to record pedal movements related to the cyclist's forward / stop decision. It becomes possible to determine whether the cyclist is coasting or pedaling.

[0044] Next, we will explain the model configuration of the information processing device and the scenes to be modeled, referring to Figures 16 and 17. Figures 16 and 17 are diagrams illustrating the model configuration of the information processing device and the scenes to be modeled.

[0045] Figure 16 shows a model configuration and a scenario for modeling a left-turn collision. In other words, cyclist 9, the rider of bicycle 19, makes the decision to go straight with the intention of going straight. The position is determined by the equations of motion, which include speed, steering angle, and lean angle. The environment 22 includes surrounding traffic participants. Surrounding traffic participants are, for example, cars. Surrounding traffic participants notify cyclist 9 of the surrounding information.

[0046] Figure 17 shows a bicycle attempting to go straight on a road and a car attempting to turn left on the road. The driver of the car uses the driving simulator 21, and the cyclist 9 of the bicycle uses the bicycle simulator 16. The cyclist 9 using the bicycle simulator 16 decelerates by braking and adjusts the speed by pedaling. In this case, data is measured using a multi-player type simulator, and the acceleration and deceleration judgment of the cyclist 9 in the left-turn involvement scene of the bicycle 19 and the car 12 is modeled.

[0047] Next, referring to FIG. 18, the explanatory variables of the cyclist determination model of the information processing apparatus according to the present embodiment will be described. FIG. 18 is a diagram for explaining the explanatory variables of the cyclist determination model of the information processing apparatus according to the present embodiment. An explanatory variable is a variable that explains a certain phenomenon or value (a variable that causes something). In this case, d y : Distance from the vehicle (horizontal), d x : Distance from the vehicle (vertical), l b : Distance to the intersection, l c : Distance to the intersection, v b : Bicycle speed, v c : Vehicle speed, w: Winker, yaw: Vehicle yaw angle become explanatory variables. These are modeled as explanatory variables and compared with actual behavior.

[0048] FIG. 19 is a diagram showing the relationship between the deceleration behavior of the cyclist 9 of the information processing apparatus according to the present embodiment and the distance. As shown on the left side of FIG. 19, when the intersection is far away, the bicycle 19 does not decelerate. It decelerates after coming near the intersection. On the other hand, as shown on the right side of FIG. 19, when close to the intersection, it decelerates about 5 m behind the car 12. The situation such as the distance rather than the speed of the car affects the deceleration judgment.

[0049] Figure 20 is a block diagram showing the functions of the information processing device 10 according to this embodiment. The information processing device 10 includes an acquisition unit 30, a first generation unit 31, a second generation unit 32, and a third generation unit 33. These are functional units realized by the CPU executing an information processing program deployed on RAM. The information processing device 10 is further equipped with storage units such as ROM, RAM, SSD, HDD, and various interfaces (not shown).

[0050] The acquisition unit 30 acquires characteristic information that characterizes traffic participants, which is the premise of the judgment model 4. The judgment model 4 is a judgment model that integrates individual judgments and motion for multiple vehicles, and a judgment model that integrates individual judgments and control. Traffic participants include pedestrians 11, cyclists 9, and drivers of automobiles 12. The judgment model 4 includes a route judgment model 6 that shows the path that traffic participants take when crossing the traffic environment. If there are no obstacles on the path when traffic participants cross the traffic environment, they will go straight, and if a collision with an obstacle is expected on the path, they will take evasive action, which is determined by comparing going straight and taking evasive action.

[0051] The acquisition unit 30 acquires cognitive information, judgment information, and behavioral information relating to a person's reaction to a pedestrian 11, judgment information relating to their decision, and behavioral information relating to their action; cognitive information, judgment information, and behavioral information relating to a person's reaction to a bicycle 19, judgment information relating to their decision, and behavioral information relating to their action; and cognitive information, judgment information, and behavioral information relating to a person's reaction to a car 12. The behavioral model, judgment model 4, and cognitive model 2 change depending on the combination of surrounding circumstances. The surrounding circumstances include the pedestrian 11, the bicycle 19, and the car 12.

[0052] The acquisition unit 30 acquires, as attribute information, information on at least one of the following: the person's age, the number of years and frequency of driving a vehicle, the person's personality, and the results of a questionnaire or test conducted with the person in advance regarding the person's physical abilities.

[0053] The first generation unit 31, in simulating the behavior of traffic participants when crossing a traffic environment, generates a behavioral model of traffic participants based on characteristic information, representing their response to understanding the surrounding environment as an external factor when crossing a traffic environment.

[0054] The second generation unit 32, in simulating the behavior of traffic participants when traversing a traffic environment, generates a cognitive model of the traffic participant based on characteristic information, representing their response to understanding the surrounding environment as an internal factor when traversing a traffic environment.

[0055] The third generation unit 33 generates a decision model 4 for the behavioral decisions of traffic participants when crossing a traffic environment, based on the behavioral model 1 and the cognitive model 2. The decision model 4 is generated by integrating the traffic participant's visual behavior, specifically the decision of whether to cross or wait at the crosswalk in relation to the first vehicle, and the decision of whether to cross or wait at the crosswalk in relation to the second vehicle, which is different from the first vehicle. The decision model 4 is generated by clustering the behavioral tendencies of traffic participants through inverse reinforcement learning using the traffic participant's characteristic information. The third generation unit 33 generates a decision model for the behavioral decisions of traffic participants when crossing a traffic environment, based on the behavioral model 1 and the cognitive model 2 when the traffic participant is a cyclist 9.

[0056] (Regarding information processing methods and information processing programs) Next, with reference to Figure 21, an information processing program according to one embodiment of the present invention will be described along with an information processing method. Figure 21 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 (not shown) of the information processing device 10 based on the information processing program. The information processing program includes an acquisition step S10, a first generation step S20, a second generation step S30, and a third generation step S40, among others. The information processing program implements acquisition, first generation, second generation, and third generation functions, etc., in the processing unit (not shown) of the information processing device 10. These functions are executed in the order shown in the flowchart of Figure 21, but the order can be changed as appropriate. Since each function overlaps with the description of the various functional units of the information processing device 10 described above, a detailed explanation is omitted.

[0057] The acquisition function acquires characteristic information that characterizes traffic participants, which is the basis for the decision model (S10: acquisition step).

[0058] The first generation function, in simulating the behavior of traffic participants when crossing a traffic environment, generates a behavioral model of traffic participants as a response to understanding the surrounding environment as an external factor when crossing a traffic environment, based on the characteristic information (S20: First generation step).

[0059] The second generation function, in simulating the behavior of traffic participants when traversing a traffic environment, generates a cognitive model of the traffic participants as a response to their perception of the surrounding environment, which is an internal factor when traversing a traffic environment, based on the aforementioned characteristic information (S30: Second generation step).

[0060] The third generation function generates the judgment model for the behavioral judgments made by traffic participants when traversing a traffic environment, based on the behavioral model and the cognitive model (S40: Third generation step).

[0061] Furthermore, 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 gist of the present invention as described in the claims. Also, although the word "data" is used in the above-described embodiment, the word "data" can be replaced with "information," and the word "information" can be replaced with "data."

[0062] In this embodiment, we have described a model of traffic participants in an outdoor traffic environment, but the model is not limited to this and can also be applied to models of traffic participants in indoor traffic environments. For example, congestion can occur at airport security checkpoints, and it is possible to create a behavioral model of passengers waiting to pass through security in such a situation, based on the behavior of other passengers. [Explanation of Symbols]

[0063] 1. The Other-Reactive Individual Behavior Model 2 Cognitive Models 3. Models for Estimating Behavioral Intentions 4. Decision Model 5. Cross-sectional decision model 6. Path Decision Model 7. Behavioral Data 8. Personal Characteristics 9 cyclists 10 Information Processing Devices 11 Pedestrians 12 Automobiles 16 Bicycle Simulator 17 Platforms 18 Steering Blocks 19. Automobile 20 Mobile phones 22 Environment 30 Acquisition Department 31 1st generation part 32 Second generation part 33 Third generation part

Claims

1. An information processing device that generates a judgment model of the behavior of traffic participants when traversing a traffic environment and performs simulations, An acquisition unit that acquires characteristic information that characterizes traffic participants, which is the premise of the aforementioned judgment model, In simulating the behavior of traffic participants when traversing a traffic environment, a first generation unit generates a behavioral model of traffic participants as a response to understanding the surrounding environment as an external factor when traversing a traffic environment, based on the aforementioned characteristic information. In simulating the behavior of traffic participants when traversing a traffic environment, a second generation unit generates a cognitive model of the traffic participant as a response to the perception of the surrounding environment as an internal factor when traversing a traffic environment, based on the aforementioned characteristic information, The system includes a third generation unit that generates a judgment model for the behavioral judgments of traffic participants when traversing a traffic environment, based on the behavioral model and the cognitive model. Information processing device.

2. The aforementioned decision model is a decision model that integrates individual decisions and motion for multiple vehicles, and a decision model that integrates individual decisions and control, and is generated by integrating the decision of whether to cross the pedestrian crossing or wait for a first vehicle, and the decision of whether to cross the pedestrian crossing or wait for a second vehicle, which is different from the first vehicle, based on the visual behavior of the traffic participant. The information processing apparatus according to claim 1.

3. The aforementioned decision model includes a path decision model that indicates the route taken by a traffic participant when crossing the traffic environment. If there are no obstacles on the route the traffic participant takes when crossing the traffic environment, they will proceed straight; if a collision with an obstacle is expected on the route, they will take evasive action, which is determined by a comparison of proceeding straight and taking evasive action. The information processing apparatus according to claim 2.

4. The aforementioned decision model is generated by clustering the behavioral tendencies of traffic participants through inverse reinforcement learning using the characteristic information of traffic participants. The information processing apparatus according to claim 1.

5. The information processing device according to claim 1, wherein the traffic participants include pedestrians, cyclists, and drivers of automobiles.

6. The information processing apparatus according to claim 1, wherein the third generation unit generates the judgment model for the behavioral judgment of a traffic participant when crossing a traffic environment, based on the behavioral model and cognitive model when the traffic participant is a cyclist.

7. The aforementioned surrounding conditions include pedestrians, cyclists, and automobiles. The acquisition unit acquires cognitive information, judgment information, and behavioral information relating to a person's reaction to a pedestrian, cognitive information, judgment information, and behavioral information relating to a person's reaction to a bicycle, cognitive information, judgment information, and behavioral information relating to a person's reaction to a car, Depending on the combination of the aforementioned surrounding circumstances, the behavioral model, the judgment model, and the cognitive model change. The information processing apparatus according to claim 1.

8. The acquisition unit acquires, as attribute information, information relating to at least one of the following: the person's age, the person's years of experience driving a vehicle and the frequency of driving, the person's personality, and the results of a questionnaire or test conducted with the person in advance regarding the person's physical abilities. The information processing apparatus according to claim 1.

9. An information processing method that generates a judgment model of the behavior of traffic participants when traversing a traffic environment and performs simulations, wherein a computer The acquisition step involves obtaining characteristic information that characterizes traffic participants, which is the premise of the aforementioned judgment model, In simulating the behavior of traffic participants when traversing a traffic environment, a first generation step is performed to generate a behavioral model of traffic participants as a response to the surrounding environment as an external factor when traversing a traffic environment, based on the aforementioned characteristic information, In simulating the behavior of traffic participants when traversing a traffic environment, a second generation step is performed to generate a cognitive model of the traffic participant as a response to the perception of the surrounding environment as an internal factor when traversing a traffic environment, based on the aforementioned characteristic information. A third generation step of generating a judgment model for the behavioral judgments of traffic participants when traversing a traffic environment, based on the behavioral model and the cognitive model, An information processing method that performs the following.

10. An information processing program that generates a judgment model of the behavior of traffic participants when traversing a traffic environment and executes simulations, which is performed on a computer. The acquisition function acquires characteristic information that characterizes traffic participants, which is the premise of the aforementioned judgment model, In simulating the behavior of traffic participants when traversing a traffic environment, a first generation function generates a behavioral model of traffic participants as a response to the surrounding environment as an external factor when traversing a traffic environment, based on the aforementioned characteristic information, In simulating the behavior of traffic participants when traversing a traffic environment, a second generation function generates a cognitive model of the traffic participant as a response to the perception of the surrounding environment as an internal factor when traversing a traffic environment, based on the aforementioned characteristic information, A third generation function that generates the judgment model for the behavioral judgments of traffic participants when traversing a traffic environment, based on the behavioral model and the cognitive model, An information processing program that executes [something].

Images