Automatic simulation test system and method for intelligent driving

Abstract

The invention relates to an automatic simulation test system and method for intelligent driving, and relates to the technical field of automatic driving, and the system comprises a data acquisition module which is used for collecting traffic environment data through multiple sensors; the model building module is used for building and initializing a multi-level simulation model based on the collected traffic environment data; the data fusion module is used for sensing the traffic environment in real time through the multi-level simulation model and performing data fusion on various data output by the multi-level simulation model to obtain fused data; the strategy generation module is used for generating a driving strategy of the vehicle according to the fusion data; the simulation test module is used for running a simulation model to carry out simulation test on the driving strategy to obtain a corresponding simulation result; and the simulation optimization module is used for analyzing the simulation result, optimizing the driving strategy to obtain an optimization strategy, and carrying out the next round of simulation test according to the optimization strategy. The system can improve the accuracy, the performance and the application effect of the driving simulation test.

Description

Technical Field

[0001] The present invention relates to the field of autonomous driving technology, and in particular to an automated simulation test system, method, electronic device, and non-transitory computer-readable storage medium for intelligent driving. Background Art

[0002] With the development of intelligent vehicles, autonomous driving technology is becoming increasingly advanced. Autonomous driving simulation testing methods can enable autonomous vehicles to adopt driving strategies similar to those of human drivers, thereby improving passenger comfort.

[0003] However, as intelligent vehicles interact more closely with their driving environments, the driving environments and traffic conditions they must navigate are becoming increasingly complex. Existing autonomous driving simulation testing methods often struggle to generate accurate and effective driving strategies for these complex traffic scenarios, failing to meet practical needs. Summary of the Invention

[0004] In response to the technical problems existing in the prior art, the present invention provides an automated simulation test system, method, electronic device and non-transitory computer-readable storage medium for intelligent driving, which can improve the performance and application effect of intelligent driving automated simulation test.

[0005] The technical solution of the present invention to solve the above technical problems is as follows:

[0006] The present invention provides an automated simulation test system for intelligent driving, the system comprising:

[0007] A data acquisition module, used to collect traffic environment data through multiple sensors;

[0008] A model building module, configured to build and initialize a multi-level simulation model based on the collected traffic environment data; the multi-level simulation model includes a macroscopic traffic flow model, a mesoscopic traffic network model, and a microscopic vehicle behavior model;

[0009] A data fusion module, configured to perceive the traffic environment in real time through the multi-level simulation model and to fuse various data output by the multi-level simulation model to obtain fused data;

[0010] A strategy generation module, configured to generate a vehicle driving strategy based on the fused data;

[0011] A simulation test module, configured to perform a simulation test on the driving strategy by running a simulation model to obtain corresponding simulation results;

[0012] The simulation optimization module is used to analyze the simulation results, optimize the driving strategy to obtain an optimized strategy, and perform the next round of simulation testing according to the optimized strategy.

[0013] Optionally, the model building module is further configured to:

[0014] Obtain the traffic density and maximum traffic density in the traffic environment;

[0015] Obtaining the speed of each vehicle in the traffic environment;

[0016] An adjustment coefficient is obtained, and a change in the traffic density of the traffic environment is determined through the traffic density, the maximum traffic density, the speed, and the adjustment coefficient of the macro traffic flow model.

[0017] Optionally, the model building module is further specifically configured to:

[0018] The equation describing the change of traffic density in the traffic environment through the macroscopic traffic flow model is expressed as:

[0019]

[0020] Among them, ρ(x,t) represents the traffic density at position x at time t, u(x,t) represents the speed, ρ max is the maximum traffic density, and γ is the adjustment coefficient, which indicates the speed at which the density returns to the maximum density.

[0021] Optionally, the model building module is further specifically configured to:

[0022] Obtaining the traffic flow density of the first path output by the macro traffic flow model;

[0023] Obtaining a path speed, a path length, a maximum traffic density, a maximum speed, and a maximum path length of the first path;

[0024] The path weight of the first path is determined according to the traffic density, path speed, path length, maximum traffic density, maximum speed and maximum path length of the first path through the meso-level traffic network model.

[0025] Optionally, the path weight is expressed as:

[0026]

[0027] Among them, α, β, and γ are the first weight, second weight, and third weight respectively, and W ij is the weight from node i to node j, p ij represents the path traffic density, u ij represents the path speed, l ij Path length, f(p ij ,u ij ,l ij ) is the calculation function of path weight, ρmax 、u max 、l max are the maximum traffic density, maximum speed and maximum path length respectively.

[0028] Optionally, the model building module is further specifically configured to:

[0029] Acquire a second vehicle that interacts with the first vehicle;

[0030] Constructing a utility function based on the action selection of the first vehicle at a historical moment, the action selection of the second vehicle at a historical moment, the state of the first vehicle, and a preset moment;

[0031] An action choice of the first vehicle at the preset time is determined according to the utility.

[0032] Optionally, the model building module is further configured to:

[0033] determining a behavior selection of the first vehicle by using the microscopic vehicle behavior model, wherein the behavior selection is obtained based on an internal utility function;

[0034] Wherein, the utility function is expressed as:

[0035]

[0036] Among them, N i is the set of other vehicles that interact with vehicle i, a ij , β ij is the interaction weight, g ij and h ij is the interaction function between vehicles;

[0037] The action selection of the first vehicle at the preset time is expressed as:

[0038] a i (t)=argmax(a i )E[U i (a i ,a -i ,s i ,t)];

[0039] Among them, a i (t) represents the action choice of vehicle i at the preset time t, E represents the expected value, U i represents the utility function of vehicle i, a -i represents the action choices of other vehicles, s i Represents the state of vehicle i.

[0040] Optionally, the fused data includes a state estimate of the vehicle at the preset time, expressed as:

[0041]

[0042] in, represents the estimated state at the preset time t, z(t) represents the sensor observation value at the preset time t, and K(t) represents the Kalman gain at the preset time t;

[0043] The calculation formula of Kalman gain K(t) is as follows:

[0044] K(T)=P(t|t-1)H T (HP(t|t-1)H T +R) -1 ;

[0045] Where P(t|t-1) is the prediction error covariance matrix, H is the observation model matrix, and R is the observation noise covariance matrix.

[0046] Optionally, the strategy generation module is further specifically configured to:

[0047] Determine the speed and position of the first vehicle at the next moment after the preset moment, expressed as:

[0048]

[0049] Among them, t is the current preset time, represents the velocity of particle i at time t+1, ω represents the inertia weight, c represents the learning factor, r represents the random number, represents the individual optimal position of particle i at the preset time t, g t represents the optimal position of the group, represents the position of particle i at time t, represents the position of particle i at time t+1, represents the periodic disturbance term.

[0050] Optionally, the method comprises:

[0051] Collect traffic environment data through multiple sensors;

[0052] Based on the collected traffic environment data, a multi-level simulation model is constructed and initialized; the multi-level simulation model includes a macro traffic flow model, a meso traffic network model, and a micro vehicle behavior model;

[0053] Real-time perception of the traffic environment through the multi-level simulation model, and data fusion of various types of data output by the multi-level simulation model to obtain fused data;

[0054] generating a driving strategy for the vehicle based on the fused data;

[0055] Performing simulation tests on the driving strategy by running a simulation model to obtain corresponding simulation results;

[0056] Analyze the simulation results, optimize the driving strategy to obtain an optimized strategy, and conduct the next round of simulation tests based on the optimized strategy.

[0057] The present invention also provides an automated simulation test method for intelligent driving, the method comprising:

[0058] Collect traffic environment data through multiple sensors;

[0059] Based on the collected traffic environment data, a multi-level simulation model is constructed and initialized; the multi-level simulation model includes a macro traffic flow model, a meso traffic network model, and a micro vehicle behavior model;

[0060] Real-time perception of the traffic environment through the multi-level simulation model, and data fusion of various types of data output by the multi-level simulation model to obtain fused data;

[0061] generating a driving strategy for the vehicle based on the fused data;

[0062] Performing simulation tests on the driving strategy by running a simulation model to obtain corresponding simulation results;

[0063] Analyze the simulation results, optimize the driving strategy to obtain an optimized strategy, and conduct the next round of simulation tests based on the optimized strategy.

[0064] In addition, to achieve the above-mentioned purpose, the present invention also proposes an electronic device, comprising: a memory for storing a computer software program; a processor for reading and executing the computer software program, thereby realizing an automated simulation test method for intelligent driving as described above.

[0065] In addition, to achieve the above-mentioned purpose, the present invention also proposes a non-transitory computer-readable storage medium, in which a computer software program is stored. When the computer software program is executed by a processor, it implements an automated simulation test method for intelligent driving as described above.

[0066] The beneficial effects of the present invention are:

[0067] (1) The present invention constructs a multi-level simulation model, including a macroscopic traffic flow model, a mesoscopic traffic network model, and a microscopic vehicle behavior model, which can comprehensively simulate traffic environments and vehicle behaviors at different levels, making the generated driving strategies closer to actual traffic scenarios. The microscopic vehicle behavior model based on game theory takes into account the interaction between vehicles and can simulate more complex and realistic driving strategies, thereby improving the performance of autonomous vehicles in complex traffic environments.

[0068] (2) The present invention introduces dynamic environmental perception technology, which uses multi-sensor fusion and machine learning algorithms to perceive the traffic environment in real time, ensuring that the autonomous vehicle can accurately obtain and respond to changes in the surrounding environment. The use of Kalman filtering for multi-sensor data fusion can effectively process sensor noise and improve the reliability and accuracy of environmental perception.

[0069] (3) The present invention uses an improved particle swarm optimization algorithm to generate driving strategies and introduces periodic disturbance terms to enhance the optimization effect, making the strategy generation process more efficient and stable. The improved optimization algorithm can quickly converge to the optimal strategy, reduce calculation time, and improve the efficiency of simulation testing.

[0070] In summary, the present invention can significantly improve the performance and application effect of the autonomous driving simulation test method, and provide strong support for the development of intelligent driving technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] Figure 1 A scene diagram of an automated simulation test method for intelligent driving provided by the present invention;

[0072] Figure 2 This is a flow chart of an automated simulation test method for intelligent driving provided by the present invention;

[0073] Figure 3 A schematic structural diagram of an automated simulation test system for intelligent driving provided by the present invention;

[0074] Figure 4 A schematic diagram of the hardware structure of a possible electronic device provided by the present invention;

[0075] Figure 5 A schematic diagram of the hardware structure of a possible computer-readable storage medium provided by the present invention. DETAILED DESCRIPTION

[0076] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making any creative efforts shall fall within the scope of protection of the present invention.

[0077] In the description of the present invention, the terms "first" and "second" are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of the technical features indicated. Therefore, a feature specified as "first" or "second" may explicitly or implicitly include one or more of the specified features. In the description of the present invention, "plurality" means two or more, unless otherwise specifically defined.

[0078] In the description of the present invention, the term "for example" is used to mean "used as an example, illustration or illustration". Any embodiment of the present invention described as "for example" is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is given to enable any person skilled in the art to implement and use the present invention. In the following description, details are listed for the purpose of explanation. It should be understood that a person of ordinary skill in the art can recognize that the present invention can be implemented without using these specific details. In other examples, well-known structures and processes are not elaborated in detail to avoid obscuring the description of the present invention with unnecessary details. Therefore, the present invention is not intended to be limited to the embodiments shown, but is consistent with the widest scope consistent with the principles and features disclosed herein.

[0079] See also Figure 1 , Figure 1 This is a scene diagram of an automated simulation test method for intelligent driving provided by the present invention. Figure 1 As shown, the terminal and server are connected via a network, such as a wired or wireless network. Terminals include, but are not limited to, portable devices such as mobile phones and tablets installed with various network platform applications, as well as fixed devices such as computers, kiosks, and advertising machines. The server provides various business services to users, including service push servers and user recommendation servers.

[0080] It should be noted that Figure 1The scenario diagram of an automated simulation test method for intelligent driving shown is only an example. The terminal, server, and application scenario described in the embodiment of the present invention are intended to more clearly illustrate the technical solution of the embodiment of the present invention, and do not generate any limitation on the technical solution provided by the embodiment of the present invention. Ordinary technicians in this field can know that with the evolution of the system and the emergence of new business scenarios, the technical solution provided by the embodiment of the present invention is also applicable to similar technical problems.

[0081] Among them, the terminal can be used to:

[0082] Collect traffic environment data through multiple sensors;

[0083] Based on the collected traffic environment data, a multi-level simulation model is constructed and initialized; the multi-level simulation model includes a macro traffic flow model, a meso traffic network model, and a micro vehicle behavior model;

[0084] Real-time perception of the traffic environment through the multi-level simulation model, and data fusion of various types of data output by the multi-level simulation model to obtain fused data;

[0085] generating a driving strategy for the vehicle based on the fused data;

[0086] Performing simulation tests on the driving strategy by running a simulation model to obtain corresponding simulation results;

[0087] Analyze the simulation results, optimize the driving strategy to obtain an optimized strategy, and conduct the next round of simulation tests based on the optimized strategy.

[0088] See also Figure 2 , Figure 2 This is a structural diagram of an automated simulation test system for intelligent driving provided by the present invention.

[0089] like Figure 2 As shown, an automated simulation test system for intelligent driving proposed in an embodiment of the present invention includes:

[0090] The data acquisition module 201 is used to collect traffic environment data through multiple sensors.

[0091] In some embodiments, the multiple sensors may include radar, cameras, ultrasonic sensors, inertial measurement units, and GPS.

[0092] In some embodiments, multiple sensors operate simultaneously to collect a variety of information about the traffic environment in real time. Each sensor provides highly accurate data based on its strengths. For example, LiDAR provides precise distance measurement, cameras provide rich visual information, and radar provides reliable detection in adverse weather conditions.

[0093] The collected data often contains noise and redundant information and requires preprocessing, such as denoising and enhancing camera images, and filtering and removing invalid points in radar and LiDAR data.

[0094] In some embodiments, data fusion techniques can be used to integrate data from different sensors. Common data fusion methods include Kalman filtering, extended Kalman filtering (EKF), and particle filtering. These methods can effectively combine the advantages of different sensors to improve the accuracy and reliability of environmental perception.

[0095] In some embodiments, the process of collecting traffic environment data by multiple sensors is as follows:

[0096] Multi-sensor collaborative work:

[0097] Multiple sensors in the autonomous driving system (radar, LiDAR, camera, ultrasonic sensor, IMU, GPS, etc.) work together to collect various information in the traffic environment in real time.

[0098] Data synchronization and preprocessing:

[0099] The collected data are preprocessed through time synchronization and spatial alignment to remove noise and redundant information and ensure data consistency and accuracy.

[0100] Data Fusion:

[0101] Data fusion techniques such as Kalman filtering combine data from different sensors to form a global perception of the surrounding environment. For example, Kalman filtering can combine distance information from radar and LiDAR with visual information from cameras through a weighted average method to form a more accurate environmental model.

[0102] By collecting and fusing traffic environment data using multiple sensors, this invention enables autonomous driving systems to obtain comprehensive and accurate perception of their surroundings. This provides a solid foundation for autonomous driving decision-making and control, improving system safety and reliability. Multi-sensor fusion technology can effectively address various challenges in complex and ever-changing traffic environments, enhancing the intelligence level of autonomous vehicles.

[0103] The model building module 202 is used to build and initialize a multi-level simulation model based on the collected traffic environment data.

[0104] In some embodiments, the multi-level simulation model includes a macroscopic traffic flow model, a mesoscopic traffic network model, and a microscopic vehicle behavior model.

[0105] In some embodiments, the model building module 202 is further configured to:

[0106] Obtain the traffic density and maximum traffic density in the traffic environment;

[0107] Obtaining the speed of each vehicle in the traffic environment;

[0108] An adjustment coefficient is obtained, and a change in the traffic density of the traffic environment is determined through the traffic density, the maximum traffic density, the speed, and the adjustment coefficient of the macro traffic flow model.

[0109] In some embodiments, the model building module 202 is further configured to:

[0110] The equation describing the change of traffic density in the traffic environment through the macroscopic traffic flow model is expressed as:

[0111]

[0112] Among them, ρ(x,t) represents the traffic density at position x at time t, u(x,t) represents the speed, ρ max is the maximum traffic density, and γ is the adjustment coefficient, which indicates the speed at which the density returns to the maximum density.

[0113] In practice, ρ(x,t) represents the traffic density at location x at time t, usually expressed as vehicles per unit length (e.g., vehicles / km). It describes the number of vehicles at a given location and time.

[0114] u(x, t) represents the speed, which is usually expressed in distance per unit time (e.g., kilometers per hour), and describes the speed of the vehicle at a given location and time;

[0115] ρ max is the maximum traffic density, which indicates the maximum vehicle density that the road can physically bear. It is a constant used to limit the upper limit of traffic density.

[0116] γ is the adjustment coefficient, which indicates the speed at which the density returns to the maximum density and reflects the system's response speed to density changes. The larger the value, the faster the system adjusts.

[0117] In the specific implementation, It represents the change of traffic density over time and space.

[0118] It represents the rate of change of the traffic density at position x at time t over time. This term describes the change in vehicle density per unit time.

[0119] It represents the rate of change of the product of traffic density and speed at position x over space. This term describes the density change caused by vehicle movement, that is, the density change caused by the vehicle moving from one position to another.

[0120] γ(ρ max -ρ(x,t)) indicates that the traffic density tends to the maximum traffic density ρ max Specifically:

[0121] ρ max -ρ(x,t) represents the difference between the current traffic density and the maximum traffic density. This difference is positive when the traffic density is lower than the maximum density and negative when the traffic density exceeds the maximum density.

[0122] γ represents the adjustment speed coefficient. It controls the speed at which traffic density returns to its maximum density. Therefore, the significance of the entire equation lies in describing how traffic density varies over time and space, and how to adjust the coefficient γ to gradually bring traffic density toward its maximum density. This model can simulate the dynamic changes in traffic flow, helping to accurately reflect the actual conditions in complex traffic scenarios during simulation tests.

[0123] In some embodiments, the model building module 202 is further configured to:

[0124] Obtaining the traffic flow density of the first path output by the macro traffic flow model;

[0125] Obtaining a path speed, a path length, a maximum traffic density, a maximum speed, and a maximum path length of the first path;

[0126] The path weight of the first path is determined according to the traffic density, path speed, path length, maximum traffic density, maximum speed and maximum path length of the first path through the meso-level traffic network model.

[0127] In some embodiments, the path weight is expressed as:

[0128]

[0129] Among them, α, β, and γ are the first weight, second weight, and third weight respectively, and W ij is the weight from node i to node j, p ij represents the path traffic density, u ij represents the path speed, l ij Path length, f(p ij ,u ij ,l ij ) is the calculation function of path weight, ρ max 、u max 、lmax are the maximum traffic density, maximum speed and maximum path length respectively.

[0130] In the specific implementation, α, β, and γ are the first weight, second weight, and third weight, respectively, which are used to adjust the relative importance of different factors in the path weight calculation. These weights are adjusted according to the needs of the specific application scenario;

[0131] W ij is the path weight from node i to node j. The larger the weight value, the worse the traffic condition from node i to node j.

[0132] p ij It represents the traffic density of the path, usually expressed as the number of vehicles per unit length;

[0133] u ij Indicates the path speed, usually expressed as distance per unit time;

[0134] l ij path length;

[0135] f(p ij ,u ij ,l ij ) is the calculation function of path weight;

[0136] ρ max 、u max 、l max They are maximum traffic density, maximum speed, and maximum path length. Maximum traffic density indicates the maximum vehicle density that a road can physically support. Maximum speed indicates the highest achievable speed on the road. Maximum path length is used to normalize path length.

[0137] Indicates the impact of the path traffic density on the weight. ij The normalized value multiplied by the first weight α reflects the congestion level of the route. The greater the traffic density, the higher the weight.

[0138] Indicates the influence of path speed on weight. ij The normalized value is then inverted and multiplied by the second weight β to reflect the contribution of low speed to the path weight. The lower the speed, the higher the weight.

[0139] Indicates the effect of path length on weight. ij After normalization, it is multiplied by the third weight γ, which reflects the physical distance of the path. The longer the path, the higher the weight.

[0140] It can be seen that the path weight W ijIt is a comprehensive function of traffic density, speed, and path length. By adjusting the weight coefficients α, β, and γ, you can flexibly reflect the importance of different factors in route selection. For example, if you want to prioritize routes with lower traffic density, you can increase α; if you are more concerned about driving speed, you can increase β; if path length is the primary consideration, you can increase γ.

[0141] Through the above methods, the present invention enables the autonomous driving system to accurately evaluate the pros and cons of each path in a simulation environment, thereby selecting the optimal driving strategy and improving the accuracy and effectiveness of simulation testing.

[0142] In some embodiments, the model building module 202 is further configured to:

[0143] Acquire a second vehicle that interacts with the first vehicle;

[0144] Constructing a utility function based on the action selection of the first vehicle at a historical moment, the action selection of the second vehicle at a historical moment, the state of the first vehicle, and a preset moment;

[0145] An action choice of the first vehicle at the preset time is determined according to the utility.

[0146] In some embodiments, the model building module 202 is further configured to:

[0147] determining a behavior selection of the first vehicle by using the microscopic vehicle behavior model, wherein the behavior selection is obtained based on an internal utility function;

[0148] Wherein, the utility function is expressed as:

[0149]

[0150] Among them, N i is the set of other vehicles that interact with vehicle i, a ij , β ij is the interaction weight, g ij and h ij is the interaction function between vehicles;

[0151] The action selection of the first vehicle at the preset time is expressed as:

[0152] a i (t)=argmax(a i )E[U i (a i ,a -i ,s i ,t)];

[0153] Among them, a i(t) represents the action choice of vehicle i at the preset time t, E represents the expected value, U i represents the utility function of vehicle i, a -i represents the action choices of other vehicles, s i Represents the state of vehicle i.

[0154] In the specific implementation, a i (t) represents the action selected by vehicle i at the preset time t, which may include driving behaviors such as acceleration, deceleration, and lane change;

[0155] E stands for expected value, indicating that the goal in the formula is to maximize the utility function U i expected value;

[0156] U i represents the utility function of vehicle i;

[0157] a -i Indicates the action options of other vehicles;

[0158] s i Represents the state of vehicle i, including the vehicle's speed, position, acceleration, etc.

[0159] argmax(a i ) means finding the utility function U by finding the maximum parameter i Maximizing action choice a i .

[0160]

[0161] Among them, N i is the set of other vehicles that interact with vehicle i, a ij , β ij is the interaction weight, g ij and h ij is the interaction function between vehicles.

[0162] In the specific implementation, a i The core idea of (t) is: at a preset time t, each vehicle i will choose an action option a i (t) makes its utility function U i The expected value E[U i (a i ,a -i ,s i ,t)] is maximized. In the specific implementation:

[0163] Select a i (t) is a decision that the first vehicle, for example, vehicle i, needs to make at each preset time t, i.e., choosing an action a i, for example, this action may be acceleration, deceleration, lane change, etc.

[0164] Expected value E[U i (a i ,a -i ,s i ,t)] is maximized, the goal of vehicle i is to choose an action a i , so that the expected value is maximized, the utility function U i Indicates that vehicle i chooses a in a specific action i 、Other second vehicle's action option a -i , own status i And the "satisfaction" or "benefit" at the preset time t, the expected value E represents the average value of the utility function taking into account uncertain factors.

[0165] Utility function U i (a i ,a -i ,s i ,t) is a multivariable function that may include the weighted sum of multiple factors, such as the vehicle’s speed, travel time, safety distance, etc. In the decision-making process, vehicle i can consider the action options a of other vehicles. -i , because the behavior of other vehicles will affect the utility of vehicle i. Vehicle i’s state s i Including the vehicle's current speed, position, acceleration, etc. These state information will affect the vehicle's decision-making process and the calculation of the utility function.

[0166] Through these methods, the simulation model can simulate the optimal behavior of each vehicle in complex traffic environments, taking into account the actions of other vehicles and its own state, thereby generating driving strategies similar to those of human drivers. This not only improves the performance of autonomous vehicles in real-world traffic environments, but also enhances passenger comfort and safety.

[0167] The data fusion module 203 is used to perceive the traffic environment in real time through the multi-level simulation model and to fuse various types of data output by the multi-level simulation model to obtain fused data.

[0168] In some embodiments, the fused data may include a state estimate of the vehicle at the predetermined time, expressed as:

[0169]

[0170] in, represents the estimated state at the preset time t, z(t) represents the sensor observation value at the preset time t, and K(t) represents the Kalman gain at the preset time t;

[0171] The calculation formula of Kalman gain K(t) is as follows:

[0172] K(T)=P(t|t-1)H T (HP(t|t-1)H T +R) -1 ;

[0173] Where P(t|t-1) is the prediction error covariance matrix, H is the observation model matrix, and R is the observation noise covariance matrix.

[0174] In the specific implementation, represents the estimated state at a preset time t, which is the best estimate of the current system state obtained by fusing sensor observations and previous state estimates;

[0175] z(t) represents the sensor observation value at the preset time t. The sensor provides measurement data at the current moment, but this data usually contains noise;

[0176] K(t) represents the Kalman gain at the preset time t. The Kalman gain plays a role in balancing the observed value and the previous state estimate during the estimation process.

[0177] This formula indicates that the current state is estimated by the Kalman gain K(t) for the current observation z(t) and the previous state estimate Specifically, when the Kalman gain is large, it is more dependent on the current observation value; when the Kalman gain is small, it is more dependent on the previous state estimate.

[0178] The calculation formula of Kalman gain K(t) is as follows:

[0179] K(T)=P(t|t-1)H T (HP(t|t-1)H T +R) -1 ;

[0180] Among them, P(t|t-1) is the prediction error covariance matrix, which represents the uncertainty of the predicted state. It is the covariance of the prediction error of the current state without considering the current observation value.

[0181] H is the observation model matrix, which describes how the system state is mapped to the observation value;

[0182] R is the observation noise covariance matrix, which represents the uncertainty of the noise in the observations.

[0183] In some embodiments, the current state and error covariance matrix can be predicted based on the previous state estimate and the system model. The predicted state is then updated using the current observation. The calculation of the Kalman gain K(t) is crucial to the update process, determining the degree of reliance on the current observation and the previous state estimate.

[0184] In some embodiments, the state at the previous time t-1 can be estimated Predict the state at the current time t, and then calculate the prediction error covariance matrix P(t|t-1). The Kalman gain K(t) can be calculated using the prediction error covariance matrix P(t|t-1), the observation model matrix H, and the observation noise covariance matrix R. Based on the current observation value z(t) and the Kalman gain K(t), the state estimate at the current time t is updated.

[0185] Through the above approach, the present invention demonstrates that the Kalman filter's estimate of the current state is a weighted average of the observed value and the previous state estimate. The magnitude of the Kalman gain, K(t), determines the degree to which the system trusts the current observation. Using the Kalman filter, autonomous driving systems can accurately perceive and estimate the vehicle's state in real time in complex and dynamic traffic environments, improving system stability and reliability.

[0186] The strategy generation module 204 is configured to generate a driving strategy for the vehicle based on the fused data.

[0187] In some embodiments, the policy generation module 204 is further configured to:

[0188] Determine the speed and position of the first vehicle at the next moment after the preset moment, expressed as:

[0189]

[0190] Among them, t is the current preset time, represents the velocity of particle i at time t+1, ω represents the inertia weight, c represents the learning factor, r represents the random number, represents the individual optimal position of particle i at the preset time t, g t represents the optimal position of the group, represents the position of particle i at time t, represents the position of particle i at time t+1, represents the periodic disturbance term.

[0191] In the specific implementation, represents the velocity of particle i at time t+1, indicating the direction and magnitude of the particle's movement in the next time step; ω represents the inertia weight, indicating the degree to which the particle's current velocity affects its velocity at the next moment. The larger the inertia weight, the stronger the tendency of the particle to maintain its current direction of motion; c represents the learning factor, which regulates the influence of individual cognition, group cognition, and periodic perturbations respectively; r represents a random number, usually between [0,1], used to introduce randomness and increase the diversity of the algorithm; represents the individual best position of particle i at the preset time t, and represents the best position found by the particle in history; g t represents the optimal position of the group, which means the optimal position found by the entire particle group at time t; Represents the position of particle i at time t, represents the value of the sine function at time t, and is used to introduce periodic changes to prevent the particle swarm from falling into a local optimal solution; represents the periodic disturbance term.

[0192] The simulation test module 205 is used to perform simulation tests on the driving strategy by running a simulation model to obtain corresponding simulation results.

[0193] A simulation model can be a high-fidelity virtual environment that simulates real-world traffic scenarios and driving conditions. It typically includes environmental elements such as road infrastructure, traffic regulations, other vehicles, and pedestrians.

[0194] Driving strategies are the driving decisions and control actions made by an autonomous driving system in a specific environment. These strategies include speed control, lane keeping, lane changing, obstacle avoidance, and other operations.

[0195] In some embodiments, a virtual traffic environment that meets the test requirements can be built on the simulation platform. This environment should simulate the target test scenario as realistically as possible, including road layout, traffic flow, traffic lights, and other dynamic obstacles.

[0196] In some embodiments, the driving strategy to be tested can be loaded into the simulation model. The driving strategy is generated by the autonomous driving algorithm and makes decisions based on sensor input and environmental perception information.

[0197] In some embodiments, a simulation model can be activated to simulate the autonomous vehicle's driving process in a virtual traffic environment. During the simulation, the behavior and response of the autonomous driving system, including vehicle speed, acceleration, direction, position changes, etc., are recorded.

[0198] In some embodiments, simulation results data can be collected after running a simulation test. This data includes vehicle trajectory, travel time, fuel consumption, passenger comfort, interactions with other vehicles and pedestrians, etc. The simulation results are analyzed using various evaluation metrics, such as safety (whether a collision occurred), efficiency (travel time and fuel consumption), and comfort (acceleration changes).

[0199] In some embodiments, the simulation results can be used to analyze the strengths and weaknesses of the driving strategy, identify potential problems and areas for improvement, and continuously adjust and improve the driving strategy through iterative optimization to enhance the performance and reliability of the autonomous driving system.

[0200] This method enables extensive testing and optimization in a virtual environment, eliminating the need for costly and high-risk real-world experiments. Simulation testing not only identifies and corrects driving strategy issues but also verifies system performance in a variety of complex traffic scenarios, improving the safety, efficiency, and passenger comfort of autonomous driving systems. This approach is an essential tool in the development of autonomous driving technology.

[0201] The simulation optimization module 206 is used to analyze the simulation results, optimize the driving strategy to obtain an optimized strategy, and perform the next round of simulation testing according to the optimized strategy.

[0202] In some embodiments, simulation result analysis may include:

[0203] Data collection: After the simulation test is completed, all data generated during the simulation process is collected. This data includes the vehicle's trajectory, speed, acceleration, fuel consumption, collision events, and interactions with other traffic participants.

[0204] Metrics evaluation: Use a series of evaluation metrics to analyze the simulation results. These metrics can include:

[0205] Safety: such as the number of collisions and near-collision incidents.

[0206] Efficiency: such as average speed, driving time, fuel efficiency.

[0207] Comfort: such as changes in acceleration and deceleration, and passenger comfort ratings.

[0208] Compliance: such as whether traffic rules and traffic lights are obeyed.

[0209] In some embodiments, identifying problems and areas for improvement may include:

[0210] Problem identification: Identify problems that arise during simulation testing based on evaluation metrics. For example, if the number of collisions is high, the obstacle avoidance strategy may need to be improved; if fuel consumption is excessive, acceleration and deceleration control may need to be optimized.

[0211] Root cause analysis: Analyze the root causes of these issues. For example, whether the obstacle avoidance strategy failed to respond in a timely manner, whether the speed control was not smooth enough, whether there were errors in environmental perception, etc.

[0212] In some embodiments, policy adjustments may include:

[0213] Parameter adjustment: Based on the analysis results, adjust the parameters in the driving strategy. For example, adjust the obstacle avoidance distance threshold, acceleration and deceleration smoothness parameters, etc.

[0214] Algorithm improvement: When necessary, improve the core algorithm, for example, improve the path planning algorithm, optimize the control algorithm, enhance the environment perception and decision-making algorithm, etc.

[0215] In some embodiments, the optimized strategy generation may be to integrate the adjusted parameters with the improved algorithm to generate a new optimized driving strategy, ensuring that the new strategy can theoretically solve the problems found in the previous simulation test.

[0216] In some embodiments, the optimized driving strategy can be loaded into the simulation model for the next round of simulation testing, preparing for the next round. The simulation model is then started and a new round of simulation testing is run. The performance of the new strategy in the same or different traffic environments is recorded. Based on the results of the new round of simulation testing, the above analysis and optimization process can be repeated, gradually improving the driving strategy until all evaluation indicators meet the expected targets.

[0217] Through this approach, the present invention enables extensive testing and optimization in a virtual environment, continuously improving driving strategies and enhancing the system's adaptability and reliability in complex traffic environments. This approach not only identifies and corrects problems in driving strategies but also verifies the system's performance in various traffic scenarios, providing a solid technical foundation for actual deployment.

[0218] See also Figure 3 , provides a flowchart of an automated simulation test method for intelligent driving of the present invention, comprising the following steps:

[0219] Step 301: Collect traffic environment data through multiple sensors;

[0220] Step 302: construct and initialize a multi-level simulation model based on the collected traffic environment data; the multi-level simulation model includes a macro traffic flow model, a meso traffic network model, and a micro vehicle behavior model;

[0221] Step 303: Real-time perception of the traffic environment through the multi-level simulation model, and data fusion of various data output by the multi-level simulation model to obtain fused data;

[0222] Step 304: Generate a vehicle driving strategy based on the fused data;

[0223] Step 305: Perform a simulation test on the driving strategy by running a simulation model to obtain a corresponding simulation result;

[0224] Step 306: Analyze the simulation results, optimize the driving strategy to obtain an optimized strategy, and perform the next round of simulation testing based on the optimized strategy.

[0225] It should be noted that the relevant content of the above steps has been described in the aforementioned part about the automated simulation test system for intelligent driving and will not be repeated here.

[0226] See also Figure 4 , Figure 4 Schematic diagram of an embodiment of an electronic device provided by an embodiment of the present invention. Figure 4 As shown, an embodiment of the present invention provides an electronic device 400, including a memory 410, a processor 420, and a computer program 411 stored in the memory 410 and executable on the processor 420. When the processor 420 executes the computer program 411, the following steps are implemented:

[0227] Collect traffic environment data through multiple sensors;

[0228] Based on the collected traffic environment data, a multi-level simulation model is constructed and initialized; the multi-level simulation model includes a macro traffic flow model, a meso traffic network model, and a micro vehicle behavior model;

[0229] Real-time perception of the traffic environment through the multi-level simulation model, and data fusion of various types of data output by the multi-level simulation model to obtain fused data;

[0230] generating a driving strategy for the vehicle based on the fused data;

[0231] Performing simulation tests on the driving strategy by running a simulation model to obtain corresponding simulation results;

[0232] Analyze the simulation results, optimize the driving strategy to obtain an optimized strategy, and conduct the next round of simulation tests based on the optimized strategy.

[0233] See also Figure 5 , Figure 5 Schematic diagram of an embodiment of a computer-readable storage medium provided in an embodiment of the present invention. Figure 5As shown, this embodiment provides a computer-readable storage medium 500 on which a computer program 411 is stored. When the computer program 411 is executed by a processor, the following steps are implemented:

[0234] Collect traffic environment data through multiple sensors;

[0235] Based on the collected traffic environment data, a multi-level simulation model is constructed and initialized; the multi-level simulation model includes a macro traffic flow model, a meso traffic network model, and a micro vehicle behavior model;

[0236] Real-time perception of the traffic environment through the multi-level simulation model, and data fusion of various types of data output by the multi-level simulation model to obtain fused data;

[0237] generating a driving strategy for the vehicle based on the fused data;

[0238] Performing simulation tests on the driving strategy by running a simulation model to obtain corresponding simulation results;

[0239] Analyze the simulation results, optimize the driving strategy to obtain an optimized strategy, and conduct the next round of simulation tests based on the optimized strategy.

[0240] It should be noted that, in the above embodiments, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant description of other embodiments.

[0241] It will be understood by those skilled in the art that embodiments of the present invention may be provided as systems, methods, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0242] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded computer, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1A system that specifies the functions of a box or boxes.

[0243] These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture including an instruction system that is implemented in the process. Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0244] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 The steps for the function specified in one or more boxes.

[0245] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0246] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if such changes and modifications fall within the scope of the claims and their equivalents, the present invention is intended to include such changes and modifications.

Claims

1. An automated simulation test system for intelligent driving, characterized in that: The system comprises: A data acquisition module, used to collect traffic environment data through multiple sensors; A model building module for building and initializing a multi-level simulation model based on the collected traffic environment data; the multi-level simulation model includes a macro traffic flow model, a meso traffic network model, and a micro vehicle behavior model; A data fusion module, configured to perceive the traffic environment in real time through the multi-level simulation model and to fuse various data output by the multi-level simulation model to obtain fused data; A strategy generation module, configured to generate a vehicle driving strategy based on the fused data; A simulation test module, configured to perform a simulation test on the driving strategy by running a simulation model to obtain corresponding simulation results; The simulation optimization module is used to analyze the simulation results, optimize the driving strategy to obtain an optimized strategy, and perform the next round of simulation testing according to the optimized strategy.

2. The automated simulation test system for intelligent driving according to claim 1, characterized in that: The model building module is also used to: Obtain the traffic density and maximum traffic density in the traffic environment; Obtaining the speed of each vehicle in the traffic environment; An adjustment coefficient is obtained, and a change in the traffic density of the traffic environment is determined through the traffic density, the maximum traffic density, the speed, and the adjustment coefficient of the macro traffic flow model.

3. The automated simulation test system for intelligent driving according to claim 2, characterized in that: The model building module is further specifically configured to: The equation describing the change of traffic density in the traffic environment through the macroscopic traffic flow model is expressed as: Among them, ρ(x, t) represents the traffic density at position x at time t, u(x, t) represents the speed, ρ max is the maximum traffic density, and Y is the adjustment coefficient, which indicates the speed at which the density returns to the maximum density.

4. The automated simulation test system for intelligent driving according to claim 3, characterized in that: The model building module is further specifically configured to: Obtaining the traffic flow density of the first path output by the macro traffic flow model; Obtaining a path speed, a path length, a maximum traffic density, a maximum speed, and a maximum path length of the first path; The path weight of the first path is determined according to the traffic density, path speed, path length, maximum traffic density, maximum speed and maximum path length of the first path through the meso-level traffic network model.

5. The automated simulation test system for intelligent driving according to claim 4, characterized in that: The path weight is expressed as: Among them, α, β, and γ are the first weight, second weight, and third weight respectively, and W i j is the weight from node i to node j, p ij represents the path traffic density, u ij represents the path speed, l ij Path length, f(p ij ,u ij , l ij ) is the calculation function of path weight, ρ max 、u max 、l max are the maximum traffic density, maximum speed and maximum path length respectively.

6. The automated simulation test system for intelligent driving according to claim 5, characterized in that: The model building module is further specifically configured to: Acquire a second vehicle that interacts with the first vehicle; Constructing a utility function based on the action selection of the first vehicle at a historical moment, the action selection of the second vehicle at a historical moment, the state of the first vehicle, and a preset moment; An action choice of the first vehicle at the preset time is determined according to the utility.

7. The automated simulation test system for intelligent driving according to claim 6, characterized in that: The model building module is also used to: determining a behavior selection of the first vehicle by using the microscopic vehicle behavior model, wherein the behavior selection is obtained based on an internal utility function; Wherein, the utility function is expressed as: Among them, N i is the set of other vehicles that interact with vehicle i, a ij , β ij is the interaction weight, g ij and h ij is the interaction function between vehicles; The action selection of the first vehicle at the preset time is expressed as: a i (t)=argmax(a i ) E[U i (a i ,a -i ,s i ,t)]; Among them, a i (t) represents the action choice of vehicle i at the preset time t, E represents the expected value, U i represents the utility function of vehicle i, a -i represents the action choices of other vehicles, s i Represents the state of vehicle i.

8. The automated simulation test system for intelligent driving according to claim 7, characterized in that: The fused data includes the vehicle's state estimate at the preset time, expressed as: in, represents the estimated state at the preset time t, z(t) represents the sensor observation value at the preset time t, and K(t) represents the Kalman gain at the preset time t; The calculation formula of Kalman gain K(t) is as follows: K(T)=P(t|t-1)H T (HP(t|t-1)H T +R) -1 ; Where P(t|t-1) is the prediction error covariance matrix, H is the observation model matrix, and R is the observation noise covariance matrix.

9. The automated simulation test system for intelligent driving according to any one of claims 1 to 8, characterized in that: The strategy generation module is further specifically configured to: Determine the speed and position of the first vehicle at the next moment after the preset moment, expressed as: Among them, t is the current preset time, represents the velocity of particle i at time t+1, ω represents the inertia weight, c represents the learning factor, r represents the random number, represents the individual optimal position of particle i at the preset time t, gt represents the optimal position of the group, represents the position of particle i at time t, represents the position of particle i at time t+1, represents the periodic disturbance term.

10. An automated simulation test method for intelligent driving, characterized in that: The method comprises: Collect traffic environment data through multiple sensors; Based on the collected traffic environment data, a multi-level simulation model is constructed and initialized; the multi-level simulation model includes a macro traffic flow model, a meso traffic network model, and a micro vehicle behavior model; Real-time perception of the traffic environment through the multi-level simulation model, and data fusion of various types of data output by the multi-level simulation model to obtain fused data; generating a driving strategy for the vehicle based on the fused data; Performing simulation tests on the driving strategy by running a simulation model to obtain corresponding simulation results; Analyze the simulation results, optimize the driving strategy to obtain an optimized strategy, and conduct the next round of simulation tests based on the optimized strategy.

Images