A traffic simulation system and method

Abstract

This invention provides a traffic simulation system and method. Through a model parameter selection module, human-driver model parameters are selected based on environmental codes obtained through an encoding module, and simulation is performed through a simulation operation module. The environmental codes include current vehicle environmental information, environmental information of historical vehicles passing through the current location, the selected human-driver model parameters, and simulation evaluation results. This system and method enable vehicle entities in traffic simulation to select appropriate human-driver model parameters based on their environment and surrounding traffic flow to generate their driving trajectories. Furthermore, the experience gained from this selection can be propagated among vehicle entities through environmental pheromones, creating a learning and optimization effect.

Description

Technical Field

[0001] This invention relates to the field of autonomous driving technology, and in particular to a traffic simulation system and method. Background Technology

[0002] Traffic simulation refers to the use of simulation technology to study traffic behavior. It is a technique that tracks and describes the changes in traffic movement over time and space. Traffic simulation has stochastic characteristics, can be microscopic or macroscopic, and involves mathematical models describing the real-time movement of transportation systems over a certain period. The role of traffic simulation is to reproduce or predict the traffic operation of existing or future systems, thereby explaining and analyzing complex traffic phenomena, identifying the root causes of problems, and ultimately optimizing the traffic system under study. Based on the degree of description of the research object by the traffic simulation model, it can be divided into microscopic simulation, mesoscopic simulation, macroscopic simulation, and traffic planning simulation. Microscopic traffic simulation provides the highest level of detail in describing the elements and behaviors of the traffic system. For example, microscopic traffic simulation models describe traffic flow using individual vehicles as the basic unit, realistically reflecting microscopic behaviors such as following, overtaking, and lane changes on the road.

[0003] In the field of autonomous driving simulation, traffic simulation engines, such as SUMO, are typically used to generate and simulate the vehicles and behaviors surrounding the tested vehicle. This aims to simulate the tested vehicle in a relatively realistic traffic flow environment to test the algorithm's performance. The realism and detail of the description of the driving behavior of each vehicle agent (VA) in the simulation system directly determine the validity of the simulation results. Research on enabling VAs in simulation systems to learn and simulate human driving behavior is collectively referred to as human-driver model research. These studies can be broadly categorized into two types based on methodology. One type is based on explicit decision-making models, which analyze the state and trajectory data of human-driven vehicles to construct decision-making models such as utility models and game theory models, and then use these models to construct driving behavior in specific driving scenarios. The other type is based on deep reinforcement learning methods, which use factors such as travel time, speed, and acceleration / deceleration consumption as feedback to the driver, constructing a "black box" human-driver model based on a deep reinforcement learning framework. However, existing research methods require constructing vehicle motion behavior models for specific driving scenarios, such as unrestrained lane changes, restrained lane changes, and left turns without signal protection. It is difficult to automatically use the corresponding human-driver model for all VAs in any scenario within a comprehensive simulation scenario that includes multiple driving scenarios. Summary of the Invention

[0004] To address some or all of the problems in the prior art, the present invention provides a traffic simulation system, comprising:

[0005] The storage module is used to store the environmental pheromone concentration at the current location, wherein the environmental pheromone concentration includes environmental information when historical vehicles passed through the current location, the parameters of the human-driving model used, and the simulation evaluation results;

[0006] The encoding module is used to convert environmental pheromone concentration and current vehicle environmental information into vectors of a specified length, which are then used as input to the model parameter selection module.

[0007] The model parameter selection module is used to select human-driver model parameters based on the environmental pheromone concentration and the current vehicle's environmental information; and

[0008] The simulation operation module is used to perform simulation operations based on the selected human-driving model parameters.

[0009] Furthermore, the environmental information includes the current vehicle's own information and the information of neighboring vehicles.

[0010] Furthermore, the current vehicle's own information includes: absolute location latitude and longitude, traffic object identification code of the map corresponding to its location, current speed, and current acceleration.

[0011] Furthermore, the information of adjacent vehicles includes information on the vehicles in front and behind the current vehicle in its lane, as well as information on the vehicles in front and behind the current vehicle in adjacent lanes. The information of each vehicle includes: distance from the current vehicle, current speed, and acceleration.

[0012] Furthermore, the environmental information also includes instantaneous traffic flow, current time, and weather information.

[0013] Furthermore, the traffic simulation system also includes a model parameter storage module, which stores human-driving model parameters, including at least one classical dynamics model and its parameters, and at least one neural network model and its parameters.

[0014] Furthermore, the simulation evaluation results include speed, and / or the smoothness of speed, and / or distance to adjacent vehicles, and / or consistency between the simulated driving trajectory and the driving plan and / or measurement data.

[0015] Based on the traffic simulation system described above, another aspect of the present invention provides a traffic simulation method, comprising:

[0016] Based on the environmental pheromones stored in the storage module and the vehicle's current environmental information, select the human-driver model parameters; and

[0017] The simulation was performed using the parameters of the human-driving model.

[0018] Furthermore, using the ant colony algorithm, the parameters of the human-driving model are selected based on the concentration of environmental pheromones, including:

[0019] If the environmental pheromone concentration is empty, then a human driver model parameter is randomly selected from the model parameter storage module; otherwise...

[0020] Based on the environmental pheromone concentration, a selection probability matrix is ​​constructed, and the human-driving model parameters of the current vehicle are selected based on the selection probability matrix.

[0021] Furthermore, the concentrations of the environmental pheromones are sorted to construct a selection probability matrix.

[0022] Furthermore, the traffic simulation method also includes:

[0023] The simulation results are evaluated, and the vehicle's human driving model parameters and evaluation results are stored as environmental information concentration in the storage module.

[0024] This invention provides a traffic simulation system and method that abstracts driver-vehicle model types and parameters into a high-dimensional space. Utilizing an ant colony algorithm-based heuristic evolutionary algorithm, vehicle agents (VAs) encode information such as their surrounding environment and vehicle states, and reference feedback from other VAs passing through the same location to learn and evolve their driver-vehicle model parameters. After a VA passes through a scene, it leaves feedback to other vehicles in the environment, evaluating its speed, driving smoothness, and safe distance from other vehicles. Through this multi-driver-vehicle model fusion mechanism, VAs in traffic simulation can select appropriate driver-vehicle model parameters based on their environment and surrounding traffic flow to generate their driving trajectories. This experience gained through selection can be propagated among VAs via environmental pheromones, resulting in a learning and optimization effect. Attached Figure Description

[0025] To further illustrate the above and other advantages and features of the various embodiments of the present invention, a more specific description of the various embodiments of the present invention will be presented with reference to the accompanying drawings. It is to be understood that these drawings depict only typical embodiments of the invention and are therefore not intended to limit its scope. In the drawings, identical or corresponding parts will be indicated by identical or similar reference numerals for clarity.

[0026] Figure 1 This diagram illustrates the structure of a traffic simulation system according to an embodiment of the present invention.

[0027] Figure 2 A schematic diagram of a model parameter storage module according to an embodiment of the present invention is shown; and

[0028] Figure 3The diagram shows a flowchart of a traffic simulation method according to an embodiment of the present invention. Detailed Implementation

[0029] In the following description, the invention is described with reference to various embodiments. However, those skilled in the art will recognize that the embodiments may be practiced without one or more specific details or in conjunction with other alternatives and / or additional methods or components. In other instances, well-known structures or operations are not shown or described in detail so as not to obscure the inventive points of the invention. Similarly, for illustrative purposes, specific numbers and configurations are set forth to provide a comprehensive understanding of embodiments of the invention. However, the invention is not limited to these specific details.

[0030] In this specification, references to "an embodiment" or "this embodiment" mean that a particular feature, structure, or characteristic described in connection with that embodiment is included in at least one embodiment of the invention. The phrase "in one embodiment" appearing throughout this specification does not necessarily refer to the same embodiment in all instances.

[0031] It should be noted that the embodiments of the present invention describe the method steps in a specific order; however, this is only for illustrating the specific embodiment and not for limiting the order of the steps. On the contrary, in different embodiments of the present invention, the order of the steps can be adjusted according to actual needs.

[0032] Existing traffic simulation solutions typically divide the test problem into several independent scenarios, each containing a specific, single driving and test task, such as unrestricted lane changing, restricted lane changing, following other vehicles, queuing, unprotected left turns, and right turns. For each scenario, a specific driver-driver model is constructed to drive the VA (Vehicle Ability) in the simulation system. However, in real-world driving environments, driving scenarios and tasks are diverse. Even if a driver's decision-making pattern can be categorized into a certain model based on the environment, such as for unrestricted lane changing scenarios, heterogeneity still exists between different drivers' specific behaviors. For example, whether a driver leans towards aggressive or conservative driving ultimately manifests as different travel speeds and the probability of executing actions. Correspondingly, the parameters in the driver-driver model have different probability distribution characteristics. Therefore, in existing solutions, the parameter combinations of the driver-driver model usually need to be pre-generated according to a certain distribution, such as a normal distribution, and then allocated during the actual simulation process. Thus, existing testing methods that focus on specific single scenarios cannot be applied to comprehensive, large-scale simulation environments.

[0033] To make traffic simulation schemes closely resemble large-scale, complex real-world traffic environments, this invention proposes a traffic simulation system and method. This system can be applied to large-scale, complex traffic simulation environments and can autonomously learn and optimize, representing a fusion technology of multiple human-driver models. Specifically, the system and method abstract human-driver model types and their parameters into a high-dimensional space. Utilizing an ant colony algorithm's heuristic evolutionary algorithm, the VA (Vehicle Assignment) learns and evolves its parameters based on encoding information such as its surrounding environment and vehicle states, and by referencing feedback from other VAs leaving their positions in the environment. After a VA passes through a scene, it leaves feedback to other vehicles, providing feedback based on its passing speed, driving smoothness, and safe distance from other vehicles. The system and method construct the VA model assignment problem in traffic simulation as an optimization problem, allowing it to be solved using a swarm intelligence algorithm framework rather than a rule-based approach. On the other hand, the system constructs a model library and parameter space to integrate various traditional vehicle dynamics models with neural network-based human-driver models, enabling simulation of complex scenarios. Furthermore, the selection of VA models ensures heterogeneity. In addition, within the model optimization framework of the method, VA is based on pheromone concentration to select models and parameters. This ensures that more suitable models and parameters have a higher probability of being selected by subsequent vehicles, while also guaranteeing a certain probability of selecting new models, thus avoiding getting trapped in local optima.

[0034] The technical solution of the present invention will be further described below with reference to the accompanying drawings of the embodiments.

[0035] Figure 1 A schematic diagram of a traffic simulation system according to an embodiment of the present invention is shown. As shown in the figure, a traffic simulation system includes a storage module 101, an encoding module 102, a model parameter selection module 103, and a simulation running module 104.

[0036] The storage module 101 is used to store the environmental pheromone concentration at the current location, wherein the environmental pheromone concentration includes environmental information when historical vehicles passed through the current location, the human-driving model parameters used, and simulation evaluation results. In one embodiment of the present invention, the environmental information includes at least the surrounding vehicle relationships when the historical vehicle passed through the current location and the human-driving model parameters used by the vehicle. The surrounding vehicle relationships include, for example, the following parameters: distance between vehicles in the same lane (PVg), speed of the vehicle in front (PVv), acceleration of the vehicle in front (PVa), distance between vehicles behind (RVg), speed of the vehicle behind (RVv), acceleration of the vehicle behind (RVa), distance between vehicles in the left lane (LLVg), speed of the vehicle in front (LLVv), acceleration of the vehicle in front (LLVa), distance between vehicles behind (LFVg), speed of the vehicle behind (LFVv), acceleration of the vehicle behind (LFVa), distance between vehicles in the right lane (RLVg), speed of the vehicle in front (RLVv), acceleration of the vehicle in front (RLVa), distance between vehicles behind (RFVg), speed of the vehicle behind (RFVv), and acceleration of the vehicle behind (LFVa). The simulation evaluation results may include, for example, passing speed, and / or the smoothness of passing speed, and / or distance to adjacent vehicles, and / or consistency between the simulated driving trajectory and the driving plan and / or measurement data.

[0037] The encoding module 102 is used to convert the environmental pheromone concentration and the current vehicle's environmental information stored in the storage module 101 into environmental codes, which are then used as input to the model parameter selection module 103. The current vehicle's environmental information includes at least the vehicle's own information and information about neighboring vehicles. The vehicle's own information may include, for example, the vehicle's absolute position latitude and longitude (po), the traffic object identification code (id) on the map corresponding to the vehicle's location, the current vehicle speed (SVv), and the current acceleration (SVa). The information about neighboring vehicles is similar to the surrounding vehicle relationships in the environmental pheromone concentration data, and may include information such as the distance, current speed, and acceleration of the vehicles in front and behind the current vehicle in its lane, as well as the distances to the vehicles in front and behind in adjacent lanes.

[0038] In addition, in order to better select the human-driving model parameters, in one embodiment of the present invention, the environmental information may also include information such as the instantaneous traffic flow pcu / h, the current time, and the weather.

[0039] Since the environmental pheromone concentration and the current vehicle's environmental information may vary as mentioned above, and some information may be empty, in one embodiment of the present invention, the environmental encoding actually converts the environmental pheromone concentration and the current vehicle's environmental information into a vector of a specified length.

[0040] In one embodiment of the present invention, the absolute location latitude and longitude (po) of the vehicle is encoded using Google's Plus Code encoding standard. Plus Code is a latitude and longitude encoding method that can represent any location on Earth. After encoding latitude and longitude, Plus Code typically consists of ten characters, excluding the plus sign. Plus Code removes easily confused letters and some unpleasant characters, using only 20 characters containing partial numbers: 2, 3, 4, 5, 6, 7, 8, 9, C, F, G, H, J, M, P, Q, R, V, W, and X. The latitude and longitude, after encoding, are reduced from two fields (longitude and latitude) to a single field, reducing the number of fields. Furthermore, the float type is no longer used; a fixed-length string can be directly used. For example, the plus code for Xujiahui is 8Q335CVQ+. Based on the length range in the plus code hierarchy, the corresponding length range is 5.5km, and a 6-bit code is appropriate, i.e., 8Q335C. Then, eight map sheets around this map sheet are selected: 8Q3349, 8Q334C, 8Q334F, 8Q3359, 8Q335F, 8Q3369, 8Q336C, and 8Q336F, ultimately forming a nine-square grid map sheet. Furthermore, the plus code defines an additional rule that can extend the code to 11 or 12 bits. An 11-bit code roughly represents a range of 3 meters, which should be able to describe the front or back door of a building, or the size of a car, providing finer positioning granularity. In this invention, the plus code uses an 11-bit extended length.

[0041] In another embodiment of the present invention, the traffic object identification code id of the map corresponding to the location of the vehicle only serves as the identification code of the traffic object in the electronic map. However, the identification code usually contains its category information and unique identifier information. Therefore, the category information can use conventional letters, such as INTENT to represent the intersection entrance and INTEXT to represent the intersection exit. The unique identifier information is usually a string of real numbers, which is only used to distinguish and does not participate in numerical calculation.

[0042] The current vehicle's own information, as well as the information of adjacent vehicles, mainly the main vehicle (VA), and the driving status and relative positional relationships of the current vehicle and the interacting vehicles (VAs) surrounding it, can be encoded as a fixed-length vector. When there is no vehicle at the corresponding position, the length can be padded with 0 values.

[0043] The current time can be encoded using relative time.

[0044] Instantaneous traffic flow (pcu / h) can be represented by equivalent flow rate, for example, to indicate traffic conditions.

[0045] Environmental pheromone concentrations are mostly numerical information, so they are similar to the current vehicle's own information and the information of neighboring vehicles. As long as the encoding results in a fixed-length vector, it is sufficient.

[0046] It should be understood that in other embodiments of the present invention, other encoding methods may also be used, as long as the above information can be converted into a format that the model parameter selection module 103 can accept.

[0047] The model parameter selection module 103 is used to select human-driver model parameters based on the environment code. In one embodiment of the present invention, an ant colony algorithm is used to select the human-driver model parameters. The basic idea of ​​applying the ant colony algorithm to solve optimization problems is as follows: the walking paths of ants represent the feasible solutions to the problem to be optimized, and all paths of the entire ant colony constitute the solution space of the problem to be optimized. Ants with shorter paths release more pheromones, and as time progresses, the pheromone concentration accumulated on the shorter paths gradually increases, and the number of ants choosing that path also increases. Finally, the entire ant colony will converge on the optimal path under the effect of positive feedback, which corresponds to the optimal solution to the problem to be optimized. Based on this, in one embodiment of the present invention, the ant colony algorithm is used to find the most suitable combination of human-driver model parameters for the current vehicle under its current conditions.

[0048] In one embodiment of the present invention, to facilitate the model parameter selection module 103 in selecting a suitable human-driver model and model parameters for the current vehicle, the system also includes a model parameter storage module 105. The model parameter storage module 105 employs a model library and hyperparameter encoding to construct a solution space. Figure 2 A schematic diagram of a model parameter storage module according to an embodiment of the present invention is shown. Figure 2 As shown, the solution space is first partitioned into upper-level partitions according to classical dynamics models, neural network models, etc., and then the lower-level partitions consist of the parameter spaces of each specific model. The classical dynamics models include, for example, Krauss car following, IDM car following, LC2013 lane change, LC2015 lane change models, etc., and the neural network models include, for example, the DRL model. Taking the IDM car following model as an example, the specific structure of the parameter space of the model in the solution space is described in detail:

[0049] The Intelligent Driving Model (IDM) is a classic car-following model proposed by Treiber et al., commonly used in microscopic simulation systems. The IDM model is defined as follows:

[0050]

[0051]

[0052] Where v0 represents the desired speed, s0 represents the minimum vehicle distance, T is the desired headway, a is the acceleration, b is the braking deceleration, and δ is the exponential parameter. The parameter space can then be considered as a 6-dimensional space composed of these 6 parameters, with typical parameter value ranges shown in Table 1.

[0053] parameter Range of values <![CDATA[v0]]> 30-120m / s <![CDATA[s0]]> 1-5m T 1.1-2.5s a <![CDATA[0.5-1.5m / s 2 ]]> b <![CDATA[0.5-2m / s 2 ]]> δ 3-5

[0054] Table 1

[0055] In common simulation systems, the default values ​​assigned to IDM model parameters are often quite conservative, which may result in a very low probability of lane-changing behavior occurring in the simulation, deviating from reality. To avoid this, in one embodiment of the present invention, a reasonable variation space is set for each parameter, allowing the learning algorithm framework to automatically adapt to appropriate values, thereby making the simulation results more consistent with reality. Similarly, in embodiments of the present invention, the parameters of other classical dynamics models and neural network models can also be set with variation spaces in this way.

[0056] The simulation operation module 104 is used to perform simulation operation according to the selected human driving model parameters.

[0057] Based on the traffic simulation system described above, by selecting human-driver model parameters according to the environmental pheromone concentration and the current vehicle's environmental information stored in the storage module, and using the human-driver model parameters to run the simulation, traffic simulation of a multi-driver model can be realized. Figure 3 This diagram illustrates a traffic simulation method according to an embodiment of the present invention. Figure 3 As shown, a traffic simulation method includes:

[0058] First, in step 301, information encoding. The encoding module converts the environmental pheromone concentration and the current vehicle's environmental information into a vector of a specified length to obtain the environmental code;

[0059] Next, in step 302, it is determined whether the environmental pheromone concentration is empty. If it is empty, proceed to step 303; otherwise, proceed to step 304.

[0060] In step 303, the human-driving model parameters are randomly selected. An empty environmental pheromone concentration indicates that no other vehicle has passed through this location before the current vehicle. Therefore, a set of human-driving models and their parameters can be randomly selected for the current vehicle from the model parameter storage module. In one embodiment of the invention, a set of human-driving models and parameters can also be formulated based on historical experience and the current vehicle's environmental information, making the initial solution not completely random, or allowing for a certain probability of searching a new solution space.

[0061] In step 304, a selection probability matrix is ​​constructed. If the environmental pheromone concentration is not empty, it means that other vehicles have passed through this location before the current vehicle. Therefore, the environmental pheromone concentration of the preceding vehicle is first obtained and sorted to construct a selection probability model. In this step, the environmental pheromone concentration of the preceding vehicle mainly refers to the pheromone concentration of the human-driving model and parameter combination it uses.

[0062] Next, in step 305, the driver-driver model parameters are selected. Based on the selection probability matrix, the driver-driver model parameters for the current vehicle are selected, so that models with higher pheromone content receive a relatively higher selection probability; and

[0063] Finally, in step 306, simulation and evaluation are performed. Based on the selected driver model and parameters, a simulation is run, and the vehicle's driving state is evaluated, generating environmental pheromones that are fed back into the environment. The purpose of the evaluation is to ensure that suitable driver models and parameters are more likely to be selected by subsequent vehicles. Therefore, in one embodiment of the invention, the vehicle's passing speed, driving smoothness, consistency with the driving objective, and safe distance from surrounding vehicles are several key factors considered in the evaluation.

[0064] The traffic simulation system and method provided by this invention can use a multi-driver model fusion mechanism to allow the VA in the traffic simulation to select a suitable driver model and parameters based on the environment and surrounding traffic flow to generate its driving trajectory. Moreover, the experience formed after this selection can be spread among the VAs through pheromones in the environment, forming a learning and optimization effect.

[0065] Although various embodiments of the invention have been described above, it should be understood that they are presented by way of example only and not as limitations. It will be apparent to those skilled in the art that various combinations, modifications, and alterations can be made without departing from the spirit and scope of the invention. Therefore, the breadth and scope of the invention disclosed herein should not be limited by the exemplary embodiments disclosed above, but should be defined solely by the appended claims and their equivalents.

Claims

1. A traffic simulation system, characterized in that, include: A storage module is configured to store the environmental pheromone concentration at the current location, wherein the environmental pheromone concentration includes environmental information when historical vehicles passed through the current location, the parameters of the human-driving model used, and simulation evaluation results, and the environmental information includes the current vehicle's own information and the information of neighboring vehicles. The encoding module is configured to convert environmental pheromone concentration and current vehicle environmental information into environmental codes, which are then used as input to the model parameter selection module. The model parameter selection module is configured to select a human-driving model and parameters based on the environment code. The human-driving model and parameters are stored in the model parameter storage module, including at least one classical dynamics model and its parameters, and at least one neural network model and its parameters. For each scenario, a specific human-driving model is constructed. Each scenario contains a specific and single driving and testing task, which includes: unconstrained lane change, or constrained lane change, or following, or queuing, or unprotected left turn, or right turn. as well as The simulation operation module is configured to perform simulation operations based on the selected human-driving model and parameters.

2. The traffic simulation system as described in claim 1, characterized in that, The vehicle's own information includes: absolute location latitude and longitude, traffic object identification code on the map corresponding to its location, current speed, and current acceleration.

3. The traffic simulation system as described in claim 1, characterized in that, The information of adjacent vehicles includes information on the vehicles in front and behind the current vehicle in its lane, as well as information on the vehicles in front and behind the current vehicle in adjacent lanes. The information of each vehicle includes: distance from the current vehicle, current speed, and acceleration.

4. The traffic simulation system as described in claim 1, characterized in that, The environmental information also includes instantaneous traffic flow, current time, and weather information.

5. The traffic simulation system as described in claim 1, characterized in that, The simulation evaluation results include speed, and / or the smoothness of speed, and / or distance to adjacent vehicles, and / or consistency between the simulated driving trajectory and the driving plan and / or measurement data.

6. A traffic simulation method, characterized in that, Including the following steps: Provides the environmental pheromone concentration at the current location, wherein the environmental pheromone concentration includes environmental information when historical vehicles passed through the current location, the parameters of the human-driving model used, and simulation evaluation results, and the environmental information includes the current vehicle's own information and the information of neighboring vehicles; Convert environmental pheromone concentrations and current vehicle environmental information into environmental codes; Based on the environment code, a human-driving model and its parameters are selected. The human-driving model and its parameters are stored in the model parameter storage module, including at least one classical dynamics model and its parameters, and at least one neural network model and its parameters. For each scenario, a specific human-driving model is constructed. Each scenario contains a specific and single driving and testing task. The driving and testing tasks include: unrestrained lane change, or restrained lane change, or following another vehicle, or queuing, or unprotected left turn, or right turn. as well as The simulation was performed using the aforementioned human-driven model and its parameters.

7. The traffic simulation method as described in claim 6, characterized in that, The ant colony algorithm is used to select human-driver model parameters based on the environmental pheromone concentration, including: If the environmental pheromone concentration is empty, a driver model and its parameters are randomly selected from the model parameter storage module; otherwise... Based on the environmental pheromone concentration, a selection probability matrix is ​​constructed, and the driver model and its parameters for the current vehicle are selected based on the selection probability matrix.

8. The traffic simulation method as described in claim 7, characterized in that, The environmental pheromone concentrations are sorted to construct a selection probability matrix.

9. The traffic simulation method as described in claim 6, characterized in that, It also includes the following steps: The simulation results are evaluated, and the vehicle's human driving model parameters and evaluation results are stored as environmental information concentration in the storage module.

Images