Intelligent automobile man-machine parallel in-loop test method
Through the intelligent automobile human-machine parallel in-ring test method, the driving simulator and vehicle dynamics model are used to solve the problems of fixed driving routes and low simulation in the existing test methods, and efficient and multi-scene automobile tests and data are achieved Acquisition improves the fidelity and stability of the test.
Patent Information
- Application Number
- CN202411240107.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-05
- Publication Date
- 2024-12-31
AI Technical Summary
The existing automobile testing methods have problems such as fixed driving routes, engineers need to visit the site, time-consuming and low simulation, making it difficult to achieve complex scenario simulation and efficient parts trial production.
Using intelligent automobile human-machine parallel in-loop testing method, the car driving simulator combines simulation control system, real-time simulation system and human-machine interaction system to provide a virtual simulation environment, simulate vehicle dynamic model and driver operation, and achieve high fidelity Driving scene simulation and data acquisition.
It realizes trial product testing on computers, reduces the cost and time of physical production, provides a high-fidelity virtual driving test site environment, supports multi-scene and environment simulation, and improves the simulation accuracy and stability of vehicle dynamic models. .
Smart Images
Figure CN119225202A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to an automobile testing method, and in particular to an intelligent automobile human-machine parallel in-the-loop testing method. Background Art
[0002] In today's automotive industry, the application of driving simulators has become one of the essential equipment for the development of automotive-related technologies. Manufacturers in the automotive industry need to produce prototypes for the technical development of new products and conduct driving tests on the company's dedicated test tracks and roads, such as Figure 1 As shown. However, because the test route cannot be changed during testing through the test line, and engineers are required to be on site every time, it brings many inconveniences in terms of time. In addition, in actual highway testing, it is difficult to realize the scenarios envisioned by technicians, and data collection requires a lot of time. During repeated testing, the trial production of parts also consumes huge funds. Existing research attempts to use pure virtual simulation methods to provide a platform foundation for driver-related research. However, pure virtual simulation methods have low realism and drivers cannot respond to scene evolution in real time. Summary of the invention
[0003] In order to solve the above technical problems, the present invention provides a human-machine parallel in-the-loop test method for intelligent automobiles, which adopts an automobile driving simulator for testing. The automobile driving simulator includes a cockpit and a human-machine co-driving simulation system. The human-machine co-driving simulation system mainly consists of three parts: a simulation control system, a real-time simulation system, and a human-machine interaction system. Among them, the simulation control system consists of a main control computer and a servo control system; the real-time simulation system mainly includes vehicle dynamics model upgrade, motion posture calculation, vision, and sound integration algorithm; the human-machine interaction system mainly includes driver operation collection and feeling simulation, instrument interaction and display. The simulation control system provides a virtual simulation environment for the real-time simulation system and the human-machine interaction system, and projects the image from the driver's perspective onto the display; the real-time simulation system calculates the vehicle dynamics model, estimates the motion posture, etc. in real time, thereby ensuring that the simulation process can meet the real-time requirements of the entire system; the human-machine interaction system establishes a connection between the virtual simulation environment and the driver, uses the instrument to display the current state of the vehicle, and collects driver behavior data through data acquisition equipment.
[0004] The simulation control system mainly realizes simulator-related servo control, including steering wheel torque simulation control and electric cylinder servo control; wherein, the motion platform where the cockpit is located already has an electric cylinder controller, therefore, the electric cylinder servo control scheme is to communicate through the main control computer and the electric cylinder controller via the data bus; the steering wheel torque simulation control includes a sensor signal processing and calibration module, a driver torque characterization module, and a steering wheel torque control module, wherein the sensor signal processing and calibration module is used to collect steering wheel torque and angle signals, the driver torque characterization module is used to calculate the required expected torque, and the steering wheel torque control module performs torque simulation control and steering wheel return control.
[0005] The real-time simulation system mainly realizes the calling of vehicle dynamics model and the calculation of motion posture of automobile driving simulator, and solves the current motion posture of virtual automobile in three-dimensional scene to obtain the expected simulator posture, and then transmits the posture data to electric cylinder controller, and realizes the synchronization of vision and hearing at the same time.
[0006] The real-time simulation system adopts a self-developed three-dimensional nonlinear real-time vehicle dynamics model, which mainly includes: aerodynamic model, body dynamics model, steering system model, engine model, transmission system model, braking system model, suspension system model, wheel dynamics model, tire dynamics model and road surface model, each of which corresponds to a corresponding graphical operation interface; the driver outputs a steering command to the steering system model, the steering system model transmits the steering angle to the wheel dynamics model, the wheel dynamics model transmits its output mechanical signals to the steering system model and the suspension system model respectively, and the steering system model feeds back the return torque to the driver; the suspension system model transmits its output mechanical signals to the body dynamics model, and transmits its simulated motion state to the wheel dynamics model; the body dynamics model transmits the speed signal to the aerodynamic model and receives the output of the aerodynamic model The air resistance is simulated and the motion state is transmitted to the suspension system model, the wheel dynamics model and the driver respectively; the body dynamics model transmits the displacement signal to the road surface model, and the road surface model outputs the road surface state, in which the friction coefficient is transmitted to the tire dynamics model, and the road shape signal is transmitted to the wheel dynamics model. The tire dynamics model receives the motion load output by the wheel dynamics model, and outputs the mechanical signal to the wheel dynamics model; the driver outputs the throttle control command to the engine model, and the engine model interacts with the transmission system model for torque and speed signals; the driver outputs the gear clutch control command to the transmission system model, and the transmission system model outputs the torque signal to the wheel dynamics model, and receives the motion state signal output by the wheel dynamics model; the driver outputs the braking command to the braking system model, and the braking system model outputs the torque signal to the wheel dynamics model, and receives the motion state signal output by the wheel dynamics model.
[0007] The human-computer interaction system includes various control systems of the cockpit, such as the road feeling simulation system, throttle system, brake system, instrument cluster, combination switch and car seat, etc., which are used to imitate the overall structure of the real car, so that the driver can operate the steering wheel, gearshift mechanism, brake pedal and accelerator pedal in the simulator according to the driving scene provided, so as to achieve the touch and visual sense of the real car driving process in a limited space. Among them, the mechanical structure of the road feeling simulation system is modified by the power steering system of the real car, mainly including: steering wheel, steering column, road feeling simulation motor, torque sensor, angle sensor, angle limit mechanism, road feeling controller; the throttle system adopts the electronic throttle and gearshift mechanism of the real car; the brake system adopts the master cylinder system for modification to simulate the real brake pedal force; the throttle, brake and parking information are obtained by the throttle pedal displacement sensor, brake pedal displacement / force sensor and displacement sensor respectively. The instrument cluster, combination switch and car seat and other components are provided by the real car, and data collection and integration are carried out later.
[0008] The main working process of the human-machine co-driving simulation system includes: the main control computer of the simulation control system collects the control quantities of each control system in the cockpit of the human-machine interaction system through the information collection sensor, transmits it to the information processing and control circuit, and directly controls the on and off of the indicator light through the control circuit after processing and calculating the information. At the same time, the collected quantity is sent to the real-time simulation system according to the communication protocol. The real-time simulation system calls the vehicle dynamics model and the motion posture of the car driving simulator at the previous moment, calculates the current motion posture of the virtual car in the three-dimensional scene, and then transmits the posture data to the simulation control system to control the movement of the motor in the cockpit motion platform to drive the movement of the cockpit. The main control computer simultaneously transmits data such as the speed and engine speed of the car driving simulator and the road information in the vision to the information processing and control circuit. After conversion, the corresponding control quantity is output to control each instrument, and each calculated simulation value is displayed on the instrument panel. At the same time, the motor in the steering wheel torque control module is controlled to rotate, driving the steering wheel to simulate the real steering feeling; the vision processing module in the vision system is called to generate the vision at the current moment, and the vision picture is projected in real time onto the large screen placed in front of the cockpit through the projector to realize the simulation of the driving scene; the sound synthesis module is called to synthesize the movement sounds such as the engine and road friction in the current driving vision, and play them through the speakers.
[0009] The test method is as follows:
[0010] The area of the virtual experimental site, road properties and weather conditions are set. In order to obtain more comprehensive relationship curves between the vehicle's driving state and the steering wheel feedback torque, during the experiment, the driver drives the vehicle at will in the virtual site, but the vehicle's driving path must include a double lane change condition to obtain the relationship between the steering wheel torque and the vehicle's driving state during lane changing and turning.
[0011] Several drivers were selected to participate in the test, and each driver completed the double lane change at a different speed. The speed when passing the double lane change is basically the maximum speed of the test.
[0012] First, a training data set is constructed with the manipulation data of a single driver. In order to obtain more extensive information on changes in vehicle status and steering wheel status, the training data set composed of each driver's driving data is composed of data obtained from the vehicle speed when passing through double lane changes; other test data that are not involved in the network training process are used as prediction data sets to verify the prediction performance of the established neural network steering wheel feedback torque model for unknown data; in order to remove the noise generated during the data acquisition process, the signals in the above test data are filtered using a Butterworth filter.
[0013] The training data set composed of a certain driver is composed of several test conditions, recording the trajectory curves under several test conditions, and drawing the change curves of the input and output variables over time after several test conditions are connected together; the total duration of each data set exceeds 120s, and the data set includes the steering wheel angle range, vehicle speed and lateral acceleration range.
[0014] Beneficial effects of the present invention:
[0015] The present invention uses a driving simulator simulation test. Before the physical production of the trial product is carried out, a mathematical model can be used to perform simulation on a computer interface. It can be directly embedded in the hardware equipment of the driving simulator, and most of the test experiments can be completed on the computer. The present invention has the relevant experimental conditions for the human-computer interaction and "human-computer co-driving" driving right switching control mechanism experiments, meets the experimental conditions for studying the driver's reaction characteristics in an emergency, and introduces the driver's subjective evaluation and the complex traffic environment formed by the interactive dynamics into the development of automobile products. It can replace the actual vehicle field test to realize the testing, verification and calibration of the whole vehicle and the electronic control system. The whole vehicle is integrated in the simulator, which is convenient for changing different models. The present invention can provide a high-fidelity automobile virtual driving test field environment, high-fidelity 3D scene modeling and rendering of multiple scenes (towns, highways, mountainous areas, test fields) and multiple environments (day, night, rain, snow and fog). The vehicle dynamics model of the present invention has high simulation accuracy; it can simulate for a long time, is stable and reliable, and has good robustness; the core code is written in ANSI-C and is easy to transplant; it has high computational efficiency and strong real-time performance; it can run in real time and non-real time under Windows and Simulink environments; the model supports real-time operation on hardware platforms such as dSPACE and NI, and the simulation step length is less than 1ms. Each model in the vehicle dynamics model is relatively independent and replaceable, and each model provides a secondary development interface for user customization; each model can read model parameters from the commercial software CarSim, and supports embedded joint simulation with CarSim, and users can also import customized vehicle appearance files. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] Figure 1 A schematic diagram of the application of a vehicle driving simulator in the development of a control system is provided as a background technology of the present invention;
[0017] Figure 2 This is a schematic diagram of the human-machine co-driving simulation system architecture of the present invention;
[0018] Figure 3 It is a schematic diagram of the vehicle dynamics model of the present invention;
[0019] Figure 4 It is a schematic diagram of the human-computer interaction system of the present invention;
[0020] Figure 5 This is a physical diagram of the human-computer interaction system of the present invention;
[0021] Figure 6 A schematic diagram of a double lane shifting path in an embodiment of the present invention;
[0022] Figure 7 1 is a trajectory diagram of driver A in the embodiment of the present invention when the test conditions are A_V40_1, A_V60_2 and A_V80_1;
[0023] Figure 8 1 is an input-output variable diagram of driver A in the embodiment of the present invention when the test conditions are A_V40_1, A_V60_2 and A_V80_1;
[0024] Fig. 9 1 is a trajectory diagram of the driver C in the embodiment of the present invention when the test conditions are C_V40_1, C_V60_1 and C_V80_2;
[0025] Fig.10 1 is an input-output variable diagram of the driver C in the embodiment of the present invention when the test conditions are C_V40_1, C_V60_1 and C_V80_2;
[0026] Fig.11 1 is a trajectory diagram of the driver F in the embodiment of the present invention when the test conditions are F_V40_1, F_V60_1 and F_V80_1;
[0027] Fig.12 1 is an input-output variable diagram of the driver F in the embodiment of the present invention when the test conditions are F_V40_1, F_V60_1 and F_V80_1.
[0028] 1. Base platform 2. Seat 3. Front guard 4. Three-screen display 5. Center console 6. Combination switch 7. Steering wheel 8. Automatic gear group 9. Foot pedal 10. Power control box 11. Casters 12. Safety belt. DETAILED DESCRIPTION
[0029] This embodiment provides a human-machine parallel in-the-loop test method for an intelligent vehicle. The method uses a vehicle driving simulator for testing. The vehicle driving simulator includes a cockpit and a human-machine co-driving simulation system. The human-machine co-driving simulation system mainly consists of three parts: a simulation control system, a real-time simulation system, and a human-machine interaction system. Figure 2 As shown; the simulation control system is composed of a main control computer and a servo control system; the real-time simulation system mainly includes vehicle dynamics model upgrades, motion posture calculations, vision, and sound integration algorithms; the human-computer interaction system mainly includes driver operation acquisition and sensory simulation, instrument interaction and display.
[0030] The simulation control system mainly realizes simulator-related servo control, including steering wheel torque simulation control and electric cylinder servo control; wherein, the motion platform where the cockpit is located already has an electric cylinder controller, therefore, the electric cylinder servo control scheme is to communicate through the main control computer and the electric cylinder controller via the data bus; the steering wheel torque simulation control includes a sensor signal processing and calibration module, a driver torque characterization module, and a steering wheel torque control module, wherein the sensor signal processing and calibration module is used to collect steering wheel torque and angle signals, the driver torque characterization module is used to calculate the required expected torque, and the steering wheel torque control module performs torque simulation control and steering wheel return control.
[0031] The real-time simulation system mainly realizes the calling of vehicle dynamics model and the calculation of motion posture of automobile driving simulator, and solves the current motion posture of virtual automobile in three-dimensional scene to obtain the expected simulator posture, and then transmits the posture data to electric cylinder controller, and realizes the synchronization of vision and hearing at the same time.
[0032] The present invention adopts a self-developed three-dimensional nonlinear real-time vehicle dynamics model, such as Figure 3The vehicle dynamics model mainly includes: aerodynamic model, body dynamics model, steering system model, engine model, transmission system model, brake system model, suspension system model, wheel dynamics model, tire dynamics model and road surface model, each of which corresponds to a corresponding graphical operation interface; the driver outputs a steering command to the steering system model, the steering system model transmits the steering angle to the wheel dynamics model, the wheel dynamics model transmits its output mechanical signals to the steering system model and the suspension system model respectively, and the steering system model feeds back the return torque to the driver; the suspension system model transmits its output mechanical signals to the body dynamics model, and transmits its simulated motion state to the wheel dynamics model; the body dynamics model transmits the speed signal to the aerodynamic model, and receives the air resistance output by the aerodynamic model to simulate the motion state. The vehicle dynamics model transmits the displacement signal to the road model, the road model outputs the road state, the friction coefficient is transmitted to the tire dynamics model, the road shape signal is transmitted to the wheel dynamics model, the tire dynamics model receives the motion load output by the wheel dynamics model, and outputs the mechanical signal to the wheel dynamics model; the driver outputs the throttle control command to the engine model, and the engine model interacts with the transmission system model for torque and speed signals; the driver outputs the gear clutch control command to the transmission system model, and the transmission system model outputs the torque signal to the wheel dynamics model and receives the motion state signal output by the wheel dynamics model; the driver outputs the braking command to the braking system model, and the braking system model outputs the torque signal to the wheel dynamics model and receives the motion state signal output by the wheel dynamics model. The vehicle dynamics model has high simulation accuracy; it can be simulated for a long time, stable and reliable, and has good robustness; the core code is written in ANSI-C and is easy to transplant; it has high computational efficiency and strong real-time performance; it can run in real time and non-real time under Windows and Simulink environments; the model supports real-time operation on hardware platforms such as dSPACE and NI, and the simulation step is less than 1ms. Each model in the vehicle dynamics model is relatively independent and replaceable, and each model provides a secondary development interface for user customization. Each model can read model parameters from the commercial software CarSim and supports embedded co-simulation with CarSim. Users can also import customized vehicle appearance files.
[0033] The human-computer interaction system includes various control systems of the cockpit, such as the road feel simulation system, throttle system, brake system, instrument cluster, combination switch and seat, etc., which are used to simulate the overall structure of the real car, so that the driver can operate the steering wheel, gear shift mechanism, brake pedal and accelerator pedal in the simulator according to the driving scene provided, so as to achieve the touch and visual sense of the real car driving process in a limited space. Among them, the mechanical structure of the road feel simulation system is modified by the power steering system of the real car, mainly including: steering wheel, steering column, road feel simulation motor, torque sensor, angle sensor, angle limit mechanism, road feel controller; the throttle system adopts the electronic throttle and gear shift mechanism of the real car; the brake system is modified by the master cylinder system to simulate the real brake pedal force; the throttle, brake and parking information are obtained by the throttle pedal displacement sensor, brake pedal displacement / force sensor and displacement sensor respectively. Components such as instrument cluster, combination switch and seat are provided by the real car, and data collection and integration are carried out later. The overall design scheme of the human-computer interaction system is as follows: Figure 4 As shown, it includes a base platform 1, a seat 2, a front guard 3, a three-screen display 4, a center console 5, a combination switch 6, a steering wheel 7, an automatic gear group 8, and a pedal 9. A power control box 10 is provided on the base platform 1, and a caster 11 is provided at the lower part. The seat 2, the front guard 3, the three-screen display 4, the center console 5, the combination switch 6, the steering wheel 7, the automatic gear group 8, and the pedal 9 are respectively arranged on the upper part of the base platform 1, the three-screen display 4 is arranged above the center console 5, the front guard 3, the three-screen display 4, the center console 5, the combination switch 6, the steering wheel 7, the automatic gear group 8, and the pedal 9 are arranged in front of the seat 2, and a seat belt 12 is provided on the seat 2; the actual object is as shown Figure 5 shown.
[0034] The main working process of the human-machine co-driving simulation system includes: the main control computer of the simulation control system collects the control quantities of each control system in the cockpit of the human-machine interaction system through the information collection sensor, transmits it to the information processing and control circuit, and directly controls the on and off of the indicator light through the control circuit after processing and calculating the information. At the same time, the collected quantity is sent to the real-time simulation system according to the communication protocol. The real-time simulation system calls the vehicle dynamics model and the motion posture of the car driving simulator at the previous moment, calculates the current motion posture of the virtual car in the three-dimensional scene, and then transmits the posture data to the simulation control system to control the movement of the motor in the cockpit motion platform to drive the movement of the cockpit. The main control computer simultaneously transmits data such as the speed and engine speed of the car driving simulator and the road information in the vision to the information processing and control circuit. After conversion, the corresponding control quantity is output to control each instrument, and each calculated simulation value is displayed on the instrument panel. At the same time, the motor in the steering wheel torque control module is controlled to rotate, driving the steering wheel to simulate the real steering feeling; the vision processing module in the vision system is called to generate the vision at the current moment, and the vision picture is projected in real time onto the large screen placed in front of the cockpit through the projector to realize the simulation of the driving scene; the sound synthesis module is called to synthesize the movement sounds such as the engine and road friction in the current driving vision, and play them through the speakers.
[0035] The test method is as follows:
[0036] The virtual test site is a 400m×400m flat dry asphalt road with no wind. In order to obtain more comprehensive relationship curves between the vehicle driving state and the steering wheel feedback torque, during the experiment, the driver drives the vehicle at will in the virtual field, but the vehicle driving path must include a double lane change condition to obtain the relationship between the steering wheel torque and the vehicle driving state during the lane change and turning process. The double lane change condition refers to the test standard in ISO3888-1:1999, and the path design in the test site is as follows: Figure 6 shown.
[0037] A total of 7 drivers participated in the test, as shown in Table 1. Each driver completed the double lane change at a speed of approximately 40km / h, 60km / h and 80km / h. The speed when passing the double lane change was basically the maximum speed of the test.
[0038] Table 1 Basic information of the driver and details of the completed test conditions
[0039]
[0040]
[0041] *In the test condition column, A_V40_1 indicates the first test in which driver A passed a double lane change at a speed of approximately 40 km / h. The first subsection represents the driver code, the second subsection represents the approximate speed when passing a double lane change, and the third subsection represents the number of tests under this condition.
[0042] First, a training data set is formed with the manipulation data of a single driver. If research requires, the training data sets formed by different drivers can also be merged into the same training data set. In order to obtain a wider range of vehicle status and steering wheel status change information, the training data set composed of each driver's driving data consists of three sets of data with vehicle speeds of approximately 40km / h, 60km / h and 80km / h when passing through double lane changes. Other test data that did not participate in the network training process are used as prediction data sets to verify the prediction performance of the established neural network steering wheel feedback torque model for unknown data. In order to remove the noise generated during the data acquisition process, the signals in the above test data are filtered using a Butterworth filter.
[0043] The following is a brief introduction to the training data set consisting of the driving data of driver A, driver C, and driver F. The training data set consisting of driver A consists of test conditions A_V40_1, A_V60_2, and A_V80_1. Figure 7 are the trajectory curves under three test conditions. Figure 8 The curves of input and output variables changing with time after the three test conditions are connected together; the training data set composed of driver C consists of test conditions C_V40_1, C_V60_1 and C_V80_2. Fig. 9 are the trajectory curves under three test conditions. Fig.10 The curves of input and output variables changing with time after the three test conditions are connected together; the training data set composed of driver F consists of test conditions F_V40_1, F_V60_2 and F_V80_1, Fig.11 are the trajectory curves under three test conditions. Fig.12 The curves of input and output variables changing with time after the three test conditions are connected together. Figure 8 , 10 As shown in Figure 12, the total duration of each data set exceeds 120s. The steering wheel angle range, vehicle speed and lateral acceleration range that constitute the data set are relatively wide. In particular, from the lateral acceleration curve, it can be seen that some test conditions have entered the nonlinear region. It can be considered that this data set fully expresses the nonlinear relationship between the steering wheel feedback torque and the motion state of the steering wheel and the vehicle.
[0044] The human-machine parallel system in-the-loop testing method constructed by the present invention can analyze and study the behavior and characteristics of real drivers operating cars. The human-machine co-driving simulation system has good performance and can provide a training data set for modeling a driver characteristic recognition neural network model.
Claims
1. A human-machine parallel in-the-loop testing method for an intelligent vehicle, characterized in that: The car driving simulator is used for testing. The car driving simulator includes a cockpit and a human-machine co-driving simulation system. The human-machine co-driving simulation system mainly consists of three parts: simulation control system, real-time simulation system, and human-machine interaction system. Among them, the simulation control system consists of a main control computer and a servo control system, and mainly realizes steering wheel force feedback control; the real-time simulation system mainly includes vehicle dynamics model upgrades, motion posture calculations, vision, and sound integration algorithms; the human-machine interaction system mainly includes driver operation collection and feeling simulation, instrument interaction and display; the simulation control system provides a virtual simulation environment for the real-time simulation system and the human-machine interaction system, and projects the image from the driver's perspective onto the display; the real-time simulation system calculates the vehicle dynamics model in real time and estimates the motion posture; the human-machine interaction system establishes a connection between the virtual simulation environment and the driver, uses the instrument to display the current status of the vehicle, and collects driver behavior data through data acquisition equipment; The simulation control system mainly realizes simulator-related servo control, including steering wheel torque simulation control and electric cylinder servo control; wherein, the motion platform where the cockpit is located is equipped with an electric cylinder controller, and the electric cylinder servo control scheme communicates with the electric cylinder controller through the main control computer via the data bus; the steering wheel torque simulation control includes a sensor signal processing and calibration module, a driver torque characterization module, and a steering wheel torque control module, wherein the sensor signal processing and calibration module is used to collect steering wheel torque and angle signals, the driver torque characterization module is used to calculate the required expected torque, and the steering wheel torque control module performs torque simulation control and steering wheel return control; The real-time simulation system mainly realizes the calling of the vehicle dynamics model and the calculation of the motion posture of the car driving simulator, and solves the current motion posture of the virtual car in the three-dimensional scene to obtain the desired simulator posture, and then transmits the posture data to the electric cylinder controller, while realizing the synchronization of vision and hearing; The human-computer interaction system includes a cockpit road feeling simulation system, a throttle system, a brake system, a combination instrument, a combination switch and a seat, which is used to simulate the overall structure of a real car, so that the driver can operate the steering wheel, the gearshift mechanism, the brake pedal and the accelerator pedal in the simulator according to the driving scene provided, thereby achieving a realistic reproduction of the touch and visual sense during the driving process of a real car in a limited space; The main working process of the human-machine co-driving simulation system includes: the main control computer of the simulation control system collects the control quantities of each control system in the cockpit of the human-machine interaction system through the information collection sensor, transmits it to the information processing and control circuit, processes and calculates the information, and directly controls the on and off of the indicator light through the control circuit. At the same time, the collected quantity is sent to the real-time simulation system according to the communication protocol. The real-time simulation system calls the vehicle dynamics model and the motion posture of the car driving simulator at the previous moment, calculates the current motion posture of the virtual car in the three-dimensional scene, and then transmits the posture data to the simulation control system to control the motor movement in the cockpit motion platform to drive the cockpit movement; the main control The computer transmits the speed, engine speed and other data of the car driving simulator and the road information in the vision to the information processing and control circuit at the same time. After conversion, the corresponding control quantity is output to control each instrument, and each calculated simulation value is displayed on the instrument panel. At the same time, the motor in the steering wheel torque control module is controlled to rotate, driving the steering wheel to simulate the real steering feeling; the vision processing module in the vision system is called to generate the vision at the current moment, and the vision picture is projected in real time on the large screen placed in front of the cockpit through the projector to realize the simulation of the driving scene; the sound synthesis module is called to synthesize the movement sounds of the engine, road friction, etc. in the current driving vision, and play them through the speakers; The test method is as follows: The area of the virtual field, road properties and weather conditions of the experiment were set. In order to obtain more comprehensive relationship curves between the vehicle's driving state and the steering wheel feedback torque, during the experiment, the driver drove the vehicle at will in the virtual field, but the vehicle's driving path must include a double lane change condition to obtain the relationship between the steering wheel torque and the vehicle's driving state during lane changing and turning. Several drivers were selected to participate in the test. Each driver completed the double lane change at a different speed. The speed when the double lane change was completed was the highest speed of the test. First, the training data set is composed of the manipulation data of a single driver. In order to obtain more extensive information on the changes in vehicle status and steering wheel status, the training data set composed of each driver's driving data is composed of the data obtained from the vehicle speed when passing through double lane changes; other test data that are not involved in the network training process are used as prediction data sets to verify the prediction performance of the established neural network steering wheel feedback torque model for unknown data; The training data set composed of a certain driver is composed of several test conditions, recording the trajectory curves under several test conditions, and drawing the change curves of the input and output variables over time after several test conditions are connected together; the total duration of each data set exceeds 120s, and the data set includes the steering wheel angle range, vehicle speed and lateral acceleration range.
2. The method for parallel human-machine in-the-loop testing of an intelligent vehicle according to claim 1, characterized in that: The real-time simulation system adopts a self-developed three-dimensional nonlinear real-time vehicle dynamics model, which mainly includes: aerodynamic model, body dynamics model, steering system model, engine model, transmission system model, braking system model, suspension system model, wheel dynamics model, tire dynamics model and road surface model, each of which corresponds to a corresponding graphical operation interface; the driver outputs a steering command to the steering system model, the steering system model transmits the steering angle to the wheel dynamics model, the wheel dynamics model transmits its output mechanical signals to the steering system model and the suspension system model respectively, and the steering system model feeds back the return torque to the driver; the suspension system model transmits its output mechanical signals to the body dynamics model, and transmits its simulated motion state to the wheel dynamics model; the body dynamics model transmits the speed signal to the aerodynamic model and receives the output of the aerodynamic model The air resistance is simulated and the motion state is transmitted to the suspension system model, the wheel dynamics model and the driver respectively; the body dynamics model transmits the displacement signal to the road surface model, and the road surface model outputs the road surface state, in which the friction coefficient is transmitted to the tire dynamics model, and the road shape signal is transmitted to the wheel dynamics model. The tire dynamics model receives the motion load output by the wheel dynamics model, and outputs the mechanical signal to the wheel dynamics model; the driver outputs the throttle control command to the engine model, and the engine model interacts with the transmission system model for torque and speed signals; the driver outputs the gear clutch control command to the transmission system model, and the transmission system model outputs the torque signal to the wheel dynamics model, and receives the motion state signal output by the wheel dynamics model; the driver outputs the braking command to the braking system model, and the braking system model outputs the torque signal to the wheel dynamics model, and receives the motion state signal output by the wheel dynamics model.
3. The method for parallel human-machine in-the-loop testing of an intelligent vehicle according to claim 1, characterized in that: The mechanical structure of the road feel simulation system is modified using the power steering system of a real vehicle, and mainly includes: a steering wheel, a steering column, a road feel simulation motor, a torque sensor, a corner sensor, a corner limit mechanism, and a road feel controller; the throttle system uses the electronic throttle and shift mechanism of a real vehicle; the braking system is modified using the master cylinder system to simulate the actual brake pedal force; the throttle, braking, and parking information are obtained through the throttle pedal displacement sensor, the brake pedal displacement / force sensor, and the displacement sensor, respectively; components such as the combination instrument, combination switch, and seat are provided by the real vehicle, and data collection and integration are performed subsequently.