CN114323698B - A real vehicle experimental platform testing method for human-machine co-driving smart cars - Google Patents

Abstract

The invention relates to a testing method for a real vehicle experimental platform for human-machine co-driving smart cars, which mainly includes the following steps: Step 1: On the basis of the E-HS3 real vehicle platform, add a controllable motor and torque/ The angle sensor meets the interface of human-machine co-driving, allowing the platform to have three modes: human driver driving, intelligent driving system driving, and human driver and intelligent driving system shared driving; Step 2: Based on step 1, provide human-machine co-driving The system deploys millimeter-wave radar, cameras, GPS positioning, industrial computers, and underlying control systems to form a closed-loop system containing external information; Step 3: Based on the characteristics of the real vehicle experiment of human-machine co-driving, design a test framework that is consistent with human-machine co-driving. The present invention designs an efficient and reliable human-machine co-driving experimental platform for the experiments and testing problems of human-machine co-driving smart cars, and designs a corresponding test control framework. The platform can test human-machine co-driving in a real vehicle environment. characteristics for effective verification.

Description

A real vehicle experimental platform testing method for human-machine co-driving smart cars

Technical field

The invention relates to a testing method for a human-machine co-driving smart car test platform, and in particular to a driving experiment platform and a testing method in a real vehicle environment that meet the requirements of human-machine co-driving smart car.

Background technique

Autonomous vehicles have developed rapidly in the past few years, but there are still many safety issues and legal policy issues in fully autonomous driving at the L4 and L5 levels. Therefore, the advanced assisted driving system (ADAS) based on shared control has gained According to more and more research, unlike previous assisted driving, the intelligent controller in human-machine co-driving can continuously assist the driver in safe driving in the control field, thereby improving driving safety and reducing the driver's driving burden. During the research process of human-machine co-driving research technology, its testing and evaluation technology has been widely studied in the industry, especially the construction of human-machine co-driving experimental platform and the formulation of evaluation standards.

Regarding the construction of experimental platforms for human-machine co-driving, hardware-in-the-loop experimental platforms have been widely developed in the industry. Jiang Haobin of Jiangsu University and others developed a set of PCs, human-machine co-driving steering ECUs, driving simulators, and front-end steering systems. A human-machine co-driving hardware-in-the-loop experimental platform with torque/angle sensor, rear torque/angle sensor, CAN card, and data collector. This platform has two modes of human driving and machine driving, which can effectively reduce development costs, but this The platform cannot consider the vehicle movement characteristics in the actual road environment (Chinese patent: CN, CN107727417A, "A hardware-in-the-loop simulation test platform for human-machine co-driving steering system").

Zhu Bing of Jilin University and others have disclosed a driving test platform for human-machine co-driving of smart cars. The platform mainly introduces the mechanical structure principles of the experimental platform, but still lacks testing in landing scenarios (Chinese patent: CN, CN109493681A, "A Driving test platform for human-machine co-driving of smart cars”). It can be seen that the experimental platform based on human-machine co-driving still needs further research considering the actual road environment and actual vehicle conditions.

Regarding the testing and evaluation standards for human-machine co-driving experiments, the industry currently conducts different evaluations based on different objects. Shi Shuming of Jilin University and others built a fault generation and control switching detection module to form a set of human-machine co-driving reliability standards. Evaluation method, but it still lacks specific elaboration on indicators such as driver burden and human-machine co-driving collaborative performance (Chinese patent: CN, CN107871418A, "An experimental platform for evaluating the reliability of human-machine co-driving"). Therefore, the testing and evaluation methods of human-machine co-driving experiments still need to be further improved, and more attention should be paid to the evaluation of lane keeping performance, driver operating load, and human-machine co-driving collaborative performance.

Contents of the invention

In response to the above problems, the purpose of the present invention is to provide a real vehicle experimental platform testing method for human-machine co-driving smart cars. The platform should introduce a human-computer co-driving interface and deploy sensors containing external information and industrial computers for information processing. , On the other hand, a corresponding experimental test framework is designed based on the characteristics of human-machine co-driving.

In order to achieve the above objectives, the present invention proposes a real vehicle experimental platform testing method for human-machine co-driving smart cars. The technical solution adopted includes the following steps: Step 1: Based on the E-HS3 real vehicle platform, first Shield the EPS power-assisted motor that comes with the original car to offset the impact of the original car's power-assisted system on the human-machine co-driving system. Add a controllable motor and torque/angle sensor to the steering system to construct an interface that satisfies human-machine co-driving. The platform has three modes: human driver driving, intelligent driving system driving, and human driver and intelligent driving system shared driving; Step 2: Based on step 1, deploy millimeter-wave radar, cameras, GPS positioning, Industrial computer and underlying control system, thus forming a closed-loop test system containing external information; Step 3: Based on the characteristics of the real vehicle experiment of human-machine co-driving, design a test framework that is consistent with human-machine co-driving. And evaluate the experimental results with subjective and objective experimental evaluation plans.

Further, (1) based on the E-HS3 real vehicle platform, design a steering system that meets the human-machine co-driving interface, which includes the following steps:

① First of all, since the torque and angle signals of the E-HS3 steering system of the original experimental vehicle cannot be obtained, the original torque and angle signals of the vehicle must be shielded to avoid the impact of this signal on the analysis of the modified human-machine co-driving system, that is, The power assist T _eps of the original car EPS is approximately equal to 0.

T _eps ≈0 (1)

② Then install a torque/angle reading sensor and a torque/angle motor on the steering column. The sensor can read the torque and angle of the vehicle's steering. The motor can execute the control signal and output the torque/angle on the known industrial computer. Under the premise of the control signal (that is, the controller output), plus the driver's control input, the resultant torque/angle acting on the additional motor can finally be obtained. This solution can meet the standards of tactile human-machine co-driving. When the controller output is shielded, it is the human-only driving mode; when the driver does not operate the steering wheel and the controller continues to output the control value, it is the controller driving mode; when the driver continues to operate the steering wheel and the controller continues to output the control value , which is the human-machine co-driving mode.

Human driver driving alone:

T _sensor =T _d (2)

Intelligent driving system driving:

T _sensor =T _c (3)

Shared driving between human drivers and intelligent driving systems:

T _sensor =T _d +T _c (4)

Among them, T _sensor is the torque read by the sensor installed on the steering column, T _d is the torque exerted by the driver on the steering wheel, and T _c is the torque exerted by the intelligent driving system on the steering column.

Further, (2) establish a closed-loop human-machine co-driving test system containing external information, which mainly includes the following steps:

The industrial computer is connected to the sensing system through 2-channel CAN (one channel is responsible for sending the vehicle's yaw angular velocity and speed to the camera, and the other channel is responsible for obtaining the information processed by the camera and millimeter-wave radar). The camera and millimeter wave radar are connected through 1 CAN. The industrial computer is connected to the high-precision positioning system through a USB channel (used to obtain the vehicle's position, direction angle and other information), and the GPS module and the 4G module are connected through a RS232 channel (used to solve the network problem of the positioning system). The underlying controller is connected to the industrial computer through 1-channel CAN, and is required to be able to send commands to the controller and receive steering wheel angle and torque information from the controller. Since the industrial computer only has 2 CAN channels, the CAN that sends the vehicle's yaw angular velocity and vehicle speed to the camera is connected in parallel with the CAN of the underlying controller. It is required that there should be no conflict between CAN IDs. The vehicle hardware is directly powered by the vehicle battery, providing 12V DC power.

Further, ① Deployment of perception system: This system is a set of fused perception devices including millimeter wave radar + camera. It is used to obtain the position of obstacles in front on the premise of inputting the vehicle's speed V _ego and yaw angular velocity γ _ego information. (x _obs ,y _obs ), direction angle ψ _obs , speed V _obs and its size information size _obs (l,w,h). The position information of the obstacle output by the sensor is described by a trajectory:

Traj _obs ={(x _obs ,y _obs ):f(x _obs ,y _obs )＝0} (5)

If it is a lane line, it is in the form of a binary linear function:

Traj _obs ={(x _obs ,y _obs ):Ax _obs +By _obs +C＝0} (6)

Among them, A, B, and C are all coefficients of the corresponding function.

Further, ② Deploy a high-precision positioning system: This system is a high-precision positioning system based on differential (RTK) GPS and inertial measurement unit (IMU), which is used to obtain the position (x _ego , y _ego ) and direction of the own vehicle. Positioning information such as angle ψ _ego and angular velocity γ _ego . This part of the information can be directly parsed and then input into the industrial computer for online processing. It should be noted that since differential technology requires network support, an additional set of 4G modules needs to be deployed to meet the network requirements of the system. Since the position information, direction angle, speed and other information of the vehicle and the obstacle are known, the relative position d _rel , relative direction angle ψ _rel , relative yaw angular velocity γ _rel and other information of the vehicle and the obstacle can be obtained.

ψ _rel ＝ψ _ego -ψ _obs (8)

γ _rel ＝γ _ego -γ _obs (9)

Further, ③Deploy the upper-layer processing unit: This system uses a vehicle-grade industrial computer as the carrier for information processing and real-time calculation of control commands. It mainly accepts vehicle CAN bus information, high-precision positioning system position, attitude and other information, and fusion perception system road and obstacle information. At the same time, it can output steering wheel torque/angle and acceleration control commands for the underlying actuator to process and control. The calculation of the command is mainly based on the control signal calculated in real time based on information processing, that is, the torque T _c ,

δ _c ＝C ₁ d _rel +C ₂ ψ _rel +C ₃ γ _rel (10)

T _c =Kδ _c (11)

Among them, C ₁ is the coefficient about the relative position of the own vehicle and the obstacle, C ₂ is the coefficient about the relative angle between the own vehicle and the obstacle, and C ₃ is the coefficient about the relative yaw angular velocity of the own vehicle and the obstacle. There is a conversion coefficient K between the output control torque T _c and the control angle δ _c .

Further, ④ deploy the underlying control system: the steering underlying controller that accepts steering wheel angle or torque commands, and the longitudinal control system that accepts acceleration commands. At the same time, the underlying control system can also provide chassis CAN bus status information such as vehicle speed, steering wheel angle and torque to the upper-level processing unit.

Further, (3) based on the characteristics of the real vehicle experiment of human-machine co-driving, design a test control method framework that is consistent with human-machine co-driving, and evaluate the experimental results with a subjective and objective experimental evaluation plan, which mainly includes the following steps:

The test framework mainly generates a reference trajectory based on the position information of the high-precision positioning system and external environment information, and then obtains the deviation amount based on its own position information and attitude information, and finally obtains a torque regulator based on fuzzy PID and active disturbance rejection, thereby achieving control The purpose of the vehicle's lateral movement includes: ① Obtaining the required steering angle using the fuzzy PID method, ② Obtaining the torque T _c of the tracking angle δ _c using the auto-disturbance rejection method.

Further, ① Obtain the required steering angle using the fuzzy PID method:

In order to make the controller fast and stable, corresponding fuzzy rules are designed to update the parameters k _p , k _I , k _d in the PID algorithm in real time, and obtain the real-time updated parameters respectively. And then get/> Then the final output control angle is δ _c ,

Among them, δ _d is the target angle based on the reference trajectory, which acts as a feedforward response in the subsequent active disturbance rejection controller. Its main purpose is to obtain a fast and accurate angle response.

② Obtain the moment T _c of the tracking angle δ _c using the method of auto-disturbance rejection:

Since there is a coefficient K between the control force and the control angle, the present invention calculates the control torque T _c by designing an automatic disturbance rejection algorithm to track the control angle δ _c as much as possible.

in, x _i , i = 1, 2 are system state quantities, r, h are adjustable parameters, T is the sampling step, z _i , i = 1, 2, 3 are expansion state quantities, β _i , i = 1, 2 ,3 are adjustable parameters, α, δ, b are adjustable parameters, e _i , i = 1, 2 are the deviations between the input state and the expanded state quantity, u is the adjustment torque acting on the steering column.

There is a coefficient K between the control force and the control angle. The automatic disturbance rejection algorithm is designed to calculate the control torque T _c to track the control angle δ _c , see formula (16).

Furthermore, for the evaluation of the experimental results and the subjective evaluation of the human-machine co-driving system, the four indicators of driving safety, driving accuracy, driving comfort, and overall driving experience during the human-machine co-driving process were evaluated. Driving safety includes whether driving safety can be improved, traffic accidents can be reduced, and whether the driver's wrong behavior can be compensated; driving accuracy includes the driver's subjective understanding of the accuracy and smoothness in the driving process; driving comfort includes A subjective evaluation of the driver's physical and mental load; the overall evaluation includes the driver's trust in the human-machine co-driving system, and the evaluation items cover gender (1), driving experience (2), age (3), and driving safety. (4-5), driving accuracy (6-7), driving comfort (8-10), and overall evaluation (11-13). ;The score includes 1-5 points (1-disagree, 2-strongly agree, 3-neutral, 4-slightly agree, 5-strongly agree);

(4) For the objective evaluation of the human-machine co-driving system, lane keeping performance, driver operating load, and human-machine co-driving collaborative performance are considered respectively.

Lane keeping performance mainly considers the degree of lane departure, trajectory tracking accuracy, and average passing time; driver operation compliance mainly considers changes in driver torque, lateral acceleration changes, and lateral acceleration change rates; human-machine collaborative control performance mainly considers the human-machine co-driving process steering correction force and driver steering correction frequency.

①Lane keeping performance indicators:

Lane departure degree: including relative position deviation d _rel and relative angle deviation ψ _rel ;

Trajectory tracking accuracy: including the deviation between the actual trajectory and the expected trajectory.

Traj _error ＝{(x _ego -x _obs ,y _ego -y _obs ):A _error Δx+B _error Δy+C _error ＝0} (17)

Among them, A _error , B _error , and C _error are correlation coefficients.

Average passing time: Enable vehicles to pass quickly while ensuring the degree of lane departure and accuracy of trajectory tracking error.

Among them, l is the test road distance, v(d _rel ,ψ _rel ,Traj _pre ) is the vehicle longitudinal speed affected by the lane departure degree and trajectory tracking error, and t is the vehicle passing time. d _pre is the distance safety threshold, ψ _pre is the direction angle safety threshold, and Traj _pre is the trajectory safety threshold.

②Driver operating load index:

Driver's moment: Obtain the driver's moment when the driver drives alone and the driver's moment T _d,d and T _d,cop in the human-machine co-driving mode respectively.

Lateral acceleration and change rate: Obtain the driver's torque when the driver is driving alone and the lateral acceleration and change rate a _y and Δa _y in the human-machine co-driving mode.

③Performance indicators of human-machine co-driving collaboration:

Steering correction force: the deviation between the controller torque and the steering back-aligning torque T _cor =T _c -T _al .

Among them, T _al =-F _yf (t _m +t _p ) is the steering back-to-alignment moment, F _yf is the tire lateral force, t _m is the distance between the extension line of the kingpin caster and the intersection between the ground and the tire surface, t _p is the tire support distance, F _yf can be approximately estimated, and t _m and t _p can take common parameters.

Driver steering correction rate: the number of times the driver corrects the steering wheel within a certain period of time. The present invention adopts the above technical solution and has the following advantages compared with the existing human-machine co-driving experimental platform:

1. The steering system of the present invention is based on the modified E-HS3 of the actual vehicle, which can meet the needs of human-machine co-driving and form a tactile human-machine co-driving mode under the real vehicle platform, and has three different switchable driving modes: Human Driver driving, intelligent driving system driving, and human driver and intelligent driving system shared driving. Compared with other hardware-in-the-loop platforms, this platform can effectively meet the needs of real-vehicle experiments of human-machine co-driving systems. 2. The present invention constructs a closed-loop human-machine co-driving real vehicle test system that contains external information. Through online processing of external road and obstacle information, vehicle position and posture and other information, it can more accurately verify the real-life human-machine co-driving. Movement characteristics in the environment.

3. Based on the characteristics of human-machine co-driving, the present invention designs a test control framework that is consistent with human-machine co-driving, and evaluates the experimental results with a subjective and objective experimental evaluation plan, so that the real vehicle experimental platform for human-machine co-driving can Effective execution and evaluation.

Description of the drawings

Figure 1 is a schematic diagram of the steering system that meets the human-machine co-driving interface.

Figure 2 shows the overall hardware architecture of the human-machine co-driving experimental platform.

Figure 3 is the test control structure of the human-machine co-driving experimental platform.

Figure 4 is the subjective evaluation analysis of the human-machine co-driving experiment.

Figure 5 shows the longitudinal velocity tracking characteristics.

Figure 6 shows the trajectory tracking changes when the driver drives alone and when humans and machines drive together.

Figure 7 shows the changes in driver torque when the driver drives alone and when the driver and the machine drive together.

Figure 8 shows the changes in vehicle lateral deviation when the driver drives alone and when the driver and the machine drive together.

Detailed ways

The technical solution of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are some, not all, of the embodiments of the present invention. The examples listed below are only for further understanding and implementation of the technical solutions of the present invention, and do not constitute further limitations on the claims of the present invention. Therefore. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of the present invention.

The specific content and implementation of the present invention will be further described below.

(1) Modify the steering system to comply with the human-machine co-driving interface

Taking into account the richness of the driving modes of this human-machine co-driving real vehicle experimental platform, which has three modes: human driver driving, intelligent driving system driving, and human driver and intelligent driving system shared driving, the human-machine co-driving platform needs The following conditions:

① When a human driver is driving, he or she can completely block the control signal of the intelligent system.

②When the intelligent system is driving alone, it can completely block human control signals.

③When the human driver and the intelligent driving system share driving, the human driver and the intelligent system can work together on the steering system and can perceive each other.

In order to meet the above conditions, the present invention has improved the steering system of the Hongqi E-HS3 car. Its structure is shown in Figure 1. The dotted line box is the original vehicle’s power-assist motor (shield), and the solid line box is the additional Motor and torque/angle sensor, which can read the torque and angle of vehicle steering, and the motor can execute control signals. Under the premise that the industrial computer output torque/angle control signal (i.e. controller output) is known, plus the driver's control input, the final installed torque/angle sensor can obtain the resultant torque/angle acting on the steering column. , this solution can meet the standard of tactile human-machine co-driving. Since the original vehicle EPS open source signal cannot be obtained, the present invention can only shield it. The final result is that the steering assist of the vehicle is reduced. The motor outputs a small inertia moment to act on the steering column, and the result can be approximately ignored, thus This avoids the impact of the output of the original car motor on the human-machine co-driving experiment. Since the power assist T _eps of the original car motor is small enough, it can be ignored, that is, T _eps ≈ 0.

Table 1 Vehicle lateral test data after modifying the steering system

When a human driver drives, the torque obtained by the sensor is the torque applied to the steering wheel by the human driver, as shown in equation (2).

When the intelligent system is driving alone, the torque obtained by the sensor is the torque applied by the intelligent system to the steering column, as shown in equation (3).

When the human driver and the intelligent driving system share driving, the torque obtained by the sensor is the sum of the torques of the human driver and the intelligent system, as shown in equation (4).

Therefore, it can be shown that this structure can provide three driving modes for human-machine co-driving real vehicle experiments: human driver driving, intelligent driving system driving, and human driver and intelligent driving system shared driving.

In order to further verify the effectiveness of the structure, a lateral steering test was conducted on the modified steering system, and the steering wheel was turned left and right at different angles. The results are shown in Table 1. From the results in Table 1, it can be seen that after modification The zero-point drift of the vehicle's steering system is almost within 0.4deg, the steady-state error of the steering angle is guaranteed to be within 0.5deg, and the maximum overshoot of the steering wheel is 6.9deg, all within a reasonable range, that is, the steering system meets the steering performance.

(2) Establish a closed-loop human-machine co-driving test system that includes external information

The main content of this part is to deploy the perception system, high-precision positioning system, upper-layer processing unit and bottom-layer control system for the vehicle. Its hardware architecture is shown in Figure 2, and the hardware information is shown in Table 2, which mainly includes the following contents:

① Deployment perception system: This system is a fusion perception device including millimeter wave radar + camera. On the premise of inputting the vehicle's speed V _ego and yaw angular velocity γ _ego information, it is used to obtain the position of the obstacle ahead (x _obs , y _obs ), direction angle ψ _obs , speed v _obs and its size information size _obs (l,w,h). The position information of the obstacle output by the sensor is described by a trajectory, such as equation (5). If it is a lane line, it is in the form of a binary linear function, such as equation (6).

②Deploy a high-precision positioning system: This system is a high-precision positioning system based on differential (RTK) GPS and inertial measurement unit (IMU), which is used to obtain the position of the vehicle (x _ego , y _ego ), direction angle ψ _ego , angular velocity γ _ego and other positioning information. This part of the information can be directly parsed and then input into the industrial computer for online processing. It should be noted that since differential technology requires network support, an additional set of 4G modules needs to be deployed to meet the network requirements of the system. Since the position information, direction angle, speed and other information of the vehicle and the obstacle are known, the relative position d _rel , relative direction angle ψ _rel , relative yaw angular velocity γ _rel and other information of the vehicle and the obstacle can be obtained, as shown in Equation (7) )-(9).

③Deploy the upper-layer processing unit: This system uses a vehicle-grade industrial computer as the carrier for information processing and real-time calculation of control commands. It mainly accepts vehicle CAN bus information, high-precision positioning system position, attitude and other information, and fusion perception system road and obstacle information. At the same time, it can output steering wheel torque/angle and acceleration control commands for the underlying actuator to process and control. The calculation of the command is mainly based on the control signal calculated in real time through information processing, which includes the rotation angle δ _c and the torque T _c , such as equations (11) and (12).

④Deploy the underlying control system: the steering underlying controller that accepts steering wheel angle or torque commands, and the longitudinal control system that accepts acceleration commands. At the same time, the underlying control system can also provide chassis CAN bus status information such as vehicle speed, steering wheel angle and torque to the upper-level processing unit.

In terms of hardware line connection, the industrial computer is connected to the sensing system through 2-channel CAN (one channel is responsible for sending the vehicle's yaw angular velocity and speed to the camera, and the other channel is responsible for obtaining the information processed by the camera and millimeter-wave radar). The camera and millimeter wave radar are connected through 1 CAN. The industrial computer is connected to the high-precision positioning system through a USB channel (used to obtain the vehicle's position, direction angle and other information), and the GPS module and the 4G module are connected through a RS232 channel (used to solve the network problem of the positioning system). The underlying controller is connected to the industrial computer through 1-channel CAN, and is required to be able to send commands to the controller and receive steering wheel angle and torque information from the controller. Since the industrial computer only has 2 CAN channels, the CAN that sends the vehicle's yaw angular velocity and vehicle speed to the camera is connected in parallel with the CAN of the underlying controller. It is required that there should be no conflict between CAN IDs. The vehicle hardware is directly powered by the vehicle battery, providing 12V DC power.

Table 2 Main hardware information

(3) Design a test control method framework that is consistent with the human-machine co-driving system

The test structure of the present invention is shown in Figure 3. The longitudinal control is controlled through the PD algorithm, and the vehicle's accelerator pedal is continuously adjusted to keep the vehicle's longitudinal speed at a stable constant. The lateral control is mainly based on the position information of the high-precision positioning system. , the external environment information generates a reference trajectory, and then obtains the lateral deviation based on its own position information and attitude information. Finally, the steering angle or torque of the steering motor is adjusted by the fuzzy PID-based turning angle mode or the fuzzy PID-based and active-disturbance rejection torque mode. Achieve the purpose of controlling the lateral movement of the vehicle.

Vertical control:

① Use PD method to control the longitudinal speed of the vehicle:

V _error =V _ego -V _d (19)

Among them, V _ego is the actual longitudinal speed of the vehicle, V _d is the target longitudinal speed of the vehicle, V _error is the longitudinal speed deviation of the vehicle, and u _V is the adjustment amount of the accelerator pedal.

Lateral controls:

② Use the fuzzy PID method to obtain the required steering angle, and the results are shown in equations (12)-(15).

③ Obtain the moment T _c of the tracking angle δ _c using the method of auto-disturbance rejection. The result is as shown in Equation (16).

For the subjective evaluation of the human-machine co-driving system, the specific evaluation items are shown in Table 3. The evaluation items cover gender (1), driving experience (2), age (3), driving safety (4-5), and driving accuracy ( 6-7), driving comfort (8-10), and overall evaluation (11-13). ;The score includes 1-5 points (1-disagree, 2-strongly agree, 3-neutral, 4-slightly agree, 5-strongly agree).

Table 3 Subjective evaluation items

Finally, the subjective evaluation results of different drivers are obtained, and the results can be averaged mean square error/> Root mean square error/> processing to further analyze different drivers’ subjective evaluations of the human-machine co-driving system.

(4) For the objective evaluation of the human-machine co-driving system, as shown in Table 4, the average value, mean square error, and root mean square error processing can also be considered when analyzing the data.

①Lane keeping performance indicators:

Lane departure degree: including relative position deviation d _rel and relative angle deviation ψ _rel ; trajectory tracking accuracy: including the deviation between the actual trajectory and the expected trajectory, such as equation (17); average transit time: ensuring the lane deviation degree and trajectory tracking error accuracy Under the conditions, the vehicle can pass quickly, as shown in Equation (18).

②Driver operating load index:

Driver's moment: Obtain the driver's moment when the driver drives alone and the driver's moment T _d,d and T _d,cop in the human-machine co-driving mode respectively. Lateral acceleration and change rate: Obtain the driver's torque when the driver is driving alone and the lateral acceleration and change rate a _y and Δa _y in the human-machine co-driving mode.

③Performance indicators of human-machine co-driving collaboration:

Steering correction force: the deviation between the controller torque and the steering back-aligning torque T _cor =T _c -T _al . Among them, T _al =-F _yf (t _m +t _p ) is the steering back-to-alignment moment, F _yf is the tire lateral force, t _m is the distance between the extension line of the kingpin caster and the intersection between the ground and the tire surface, t _p is the tire support distance, F _yf can be approximately estimated, and t _m and t _p can take common parameters. Driver steering correction rate: the number of times the driver corrects the steering wheel within a certain period of time.

Table 4 Objective evaluation items

On the basis of the three parts, step one ensures the hardware interface of human-machine co-driving, step two ensures the closed-loop test including external information, and step three ensures the testing method and evaluation plan of human-machine co-driving. In order to further verify the human-machine co-driving The present invention conducted effective experimental verification based on the test structure diagram 3 of the comprehensive test effect of the real-vehicle co-driving experimental platform.

① The experimental road is an unmanned roundabout entering and exiting the road condition, which approximately simulates the double-shifting condition.

②The reference trajectory is calculated based on positioning information (GPS), attitude and heading information (IMU), and environmental information (millimeter wave radar + camera).

③In order to verify the effectiveness of the experimental platform, PD adaptive cruise control is used for longitudinal control, the vehicle speed is controlled at 15km/h, and fuzzy PID + automatic disturbance rejection control algorithm is used for lateral control. The driver can also control the steering wheel in real time based on environmental information to control the vehicle. .

④Select drivers of different ages to conduct real-vehicle human-machine co-driving tests.

The statistics of the evaluation results of drivers of different ages on subjective evaluation items are shown in Table 5, and the processed data analysis is shown in Figure 4.

By analyzing the average values, it can be concluded that in terms of safety, drivers have a relatively conservative attitude towards the shared driving system's ability to improve driving safety, but they are more convinced that the shared driving system can compensate for driving errors and fatigue driving. In terms of accuracy, drivers agree that the shared driving system can make it easier to keep the vehicle in the middle of the lane and make the vehicle drive more smoothly; in terms of comfort, drivers believe that the shared driving system can make the driver's driving easier. It is physically taxing and can adapt to the co-driving system, but it is not recognized that the co-driving system will reduce the driver's psychological burden on driving. This may be due to the inconsistency between humans and machines, so there is a need for communication between the driver and the intelligent controller. A running-in process is carried out to improve the rationality of the human-machine co-driving system; finally, it can be seen that the driver can trust the co-driving system as a whole and thinks that it can be considered to be equipped with a co-driving system.

By analyzing the variance and average, it can be concluded that there are large differences between different drivers on the issue of whether the shared driving system can improve driving safety. This may be due to the fact that a small number of drivers may have a psychological potential to conflict with the shared driving system. Therefore, it is felt that the controller will compete with the driver, thus affecting safety; however, from the overall evaluation, different drivers have value for the shared driving system, and think that they can consider equipping the shared driving system. The differences are minimal, which shows that in the above experiments, Basically, drivers of different genders, driving experience, and age groups generally have a positive and optimistic attitude towards the shared driving system.

Table 4 Statistics of subjective evaluation items for human-machine co-driving

The objective evaluation of the experimental results is shown in Figures 5-8. Figure 5 shows that the human-machine co-driving real vehicle experimental platform can not only achieve lateral control, but also study longitudinal control issues, and can then conduct a series of horizontal and vertical couplings. Test experiments. Figure 6 shows that the human-machine co-driving real-vehicle experimental platform can be used to analyze the changing characteristics of trajectory tracking when the driver drives alone and when human-machine co-driving. It can be seen that the trajectory tracking accuracy under human-machine co-driving is higher than that when the driver drives alone. ; Figure 7 shows that the platform can be used to analyze the driver's torque operating load when the driver is driving alone or when the driver and the machine are driving together, thereby analyzing the impact on the driver. It can be seen that there may be differences between the driver and the intelligent system during the process of human-machine joint driving. The process of confrontation; Figure 8 shows that the platform can be used to analyze the different lateral deviations between driver driving alone and human-machine co-driving, that is, it can analyze the trajectory tracking deviation of different driving modes. It can be seen that when the driver drives alone, relatively large deviations may suddenly occur. Large deviation, while the change of lateral deviation during human-machine co-driving is relatively stable.

This invention only lists some feasible experiments and analyses. In fact, the platform can meet the trajectory tracking test and driver load test in three modes: driver driving alone, intelligent system driving alone, and human-machine co-driving. , anti-interference test under human-machine co-driving, characteristic analysis of changes in different variables (lateral deviation, lateral acceleration, yaw angular velocity, etc.), model and algorithm verification and other experiments. In addition, it can also conduct horizontal and vertical analysis of vehicle dynamics in different modes. Study the coupling situation.

Although embodiments of the present invention have been disclosed above, they are not limited to the uses set forth in the specification and description. It can be applied to various fields suitable for the present invention. Additional modifications can be readily implemented by those skilled in the art. Therefore, the invention is not limited to the specific details and illustrations shown and described herein without departing from the general concept as defined by the claims and their equivalent scope.

Claims (7)

1. A real vehicle experiment platform test method for a man-machine co-driving intelligent vehicle comprises the following design steps:

step one: on the basis of an E-HS3 real vehicle platform, an EPS power-assisted motor of an original vehicle is shielded, a controllable motor and a moment/angle sensor are additionally arranged in a steering system, so that an interface meeting the requirements of man-machine co-driving is constructed, and the platform has three modes of human driver driving, intelligent driving system driving and shared driving of a human driver and the intelligent driving system;

step two: on the basis of the first step, a millimeter wave radar, a camera, a GPS positioning system, an industrial personal computer and a bottom layer control system are deployed for the co-driving system of the man-machine, so that a closed-loop co-driving test system of the man-machine containing external information is formed;

step three: aiming at the characteristics of man-machine co-driving real vehicle experiments, a test frame conforming to man-machine co-driving is designed; wherein:

in the second step, the closed-loop man-machine co-driving test system specifically comprises: (1) deploying a sensing system, (2) deploying a high-precision positioning system, (3) deploying an upper-layer processing unit, and (4) deploying a bottom-layer control system;

the deployment awareness system: the system is a set of fusion sensing device comprising millimeter wave radar and a camera, and is used for inputting the speed V of a vehicle _ego And yaw rate gamma _ego On the premise of the information, for acquiring a position (x _obs ,y _obs ) Direction angle psi _obs Velocity V _obs Size information size of the display device _obs (l, w, h) wherein the position information of the obstacle output by the sensor is described by a trace:

Traj _obs ＝{(x _obs ,y _obs ):f(x _obs ,y _obs )＝0} (5)

if the lane line is the form of a binary linear function:

Traj _obs ＝{(x _obs ,y _obs ):Ax _obs +By _obs +C＝0} (6)

wherein A, B and C are the coefficients of the corresponding functions;

the deployment high-precision positioning system comprises: the system is a set of high-precision positioning system based on differential (RTK) GPS and Inertial Measurement Unit (IMU) and is used for acquiring the position (x) _ego ,y _ego ) Direction angle psi _ego Angular velocity gamma _ego Waiting for positioning information; the part of information can be directly analyzed and then input into an industrial personal computer for online processing; because the differential technology needs network support, a set of 4G modules needs to be deployed additionally to meet the network requirements of the system; the relative position distance d between the vehicle and the obstacle is obtained by knowing the position information, the direction angle and the angular velocity information of the vehicle and the obstacle _rel Relative direction angle ψ _rel Relative yaw rate gamma _rel Information;

ψ _rel ＝ψ _ego -ψ _obs (8)

γ _rel ＝γ _ego -γ _obs (9)；

the deployment upper layer processing unit: the system takes a vehicle-standard industrial personal computer as a carrier and is used for carrying out information processing and real-time calculation of control commands; the information processing comprises information input of a sensing system, high-precision positioning system input, vehicle-mounted CAN information input and control signal input; the calculation of the control command is based mainly on control signals calculated in real time by information processing, i.e. the torque T _c ，

δ _c ＝C ₁ d _rel +C ₂ ψ _rel +C ₃ γ _rel (10)

T _c ＝Kδ _c (11)

wherein ,C₁ is a coefficient related to the relative position of the bicycle and the obstacle, C ₂ Is a coefficient related to the relative angle between the vehicle and the obstacle, C ₃ The coefficient of the yaw rate of the vehicle relative to the obstacle is used for outputting a control torque T _c And controlling the rotation angle delta _c There is a conversion coefficient K between them.

2. The test method of the real vehicle experiment platform for the man-machine co-driving intelligent vehicle according to claim 1, wherein in the first step, on the basis of the E-HS3 real vehicle platform, an EPS (electric power steering) assisting motor of an original vehicle is firstly required to be shielded, so that the influence of an original vehicle assisting system on the man-machine co-driving system is counteracted, namely, the assisting force of the original vehicle EPS is approximately equal to 0, then a controllable motor and a moment/angle sensor are added in a steering system, so that an interface meeting the man-machine co-driving requirement is constructed,

T _eps ≈0 (1)

finally, three modes of human driver driving, intelligent driving system driving and human driver and intelligent driving system sharing driving are formed:

(1) the human driver drives alone:

T _sensor ＝T _d (2)

(2) and (3) driving by an intelligent driving system:

T _sensor ＝T _c (3)

(3) human driver and intelligent driving system share driving:

T _sensor ＝T _d +T _c (4)

wherein ,T_eps Is EPS assistance of the original vehicle, T _sensor Is the torque read by a sensor arranged on the steering column, T _d Is the magnitude of the moment exerted on the steering wheel by the driver, T _c Is the magnitude of the torque applied to the steering column by the intelligent driving system.

3. The method for testing the real vehicle experimental platform of the man-machine co-driving intelligent vehicle according to claim 1, wherein the establishing a closed-loop man-machine co-driving testing system containing external information in the second step comprises the following steps:

the industrial personal computer is connected with the sensing system through 2 paths of CAN, one path is responsible for sending the yaw rate and the vehicle speed of the vehicle to the camera, the other path is responsible for acquiring the information processed by the camera and the millimeter wave radar, the camera is connected with the millimeter wave radar through 1 path of CAN, the industrial personal computer is connected with the high-precision positioning system through 1 path of USB and is used for acquiring the information of the position, the direction angle and the like of the vehicle, and the GPS module is connected with the 4G module through one path of RS232 and is used for solving the network problem of the positioning system; the bottom layer controller is connected with the industrial personal computer through a 1-way CAN, and is required to not only send a command to the controller, but also receive information of steering wheel rotation angle and torque from the controller; because the industrial personal computer only has 2 paths of CAN, the CAN for transmitting the yaw rate and the vehicle speed of the vehicle to the camera is connected in parallel with the CAN of the bottom controller, and no conflict exists between CAN IDs; the vehicle-mounted hardware is directly powered by a vehicle-mounted battery to provide 12V direct current;

the closed-loop man-machine co-driving test system specifically comprises: (1) the method comprises the steps of (1) deploying a sensing system, (2) deploying a high-precision positioning system, (3) deploying an upper-layer processing unit, and (4) deploying a bottom-layer control system.

4. The method for testing a real vehicle experimental platform for a man-machine co-driving intelligent vehicle according to claim 3, wherein the deployment bottom layer control system: a steering floor controller for receiving steering wheel angle or torque commands, and a longitudinal control system for receiving acceleration commands; meanwhile, the bottom layer control system CAN also provide chassis CAN bus state information such as vehicle speed, steering wheel angle, moment and the like for the upper layer processing unit.

5. The method for testing the real vehicle experimental platform for the man-machine co-driving intelligent vehicle according to claim 1, wherein in the third step, aiming at the characteristics of the man-machine co-driving real vehicle experiment, a test control framework which accords with the man-machine co-driving is designed, and the experimental result is evaluated by a subjective and objective experimental evaluation scheme, and the method comprises the following steps:

the test frame mainly generates a reference track according to the position information and the external environment information of the high-precision positioning system, obtains deviation according to the position information and the posture information of the test frame, and finally obtains a moment regulator based on fuzzy PID and active disturbance rejection, thereby achieving the purpose of controlling the transverse movement of the vehicle;

comprising the following steps: (1) obtaining the angle required by steering by a fuzzy PID method, (2) obtaining the tracking angle delta by an active disturbance rejection method _c Moment T of (2) _c 。

6. The method for testing the real vehicle experimental platform of the man-machine co-driving intelligent vehicle according to claim 5, wherein (1) the required steering angle is obtained by a fuzzy PID method:

in order to make the controller quickly and stably, a corresponding fuzzy rule is designed so as to update the parameter k in the PID algorithm in real time _p ,k _I ,k _d Respectively obtaining parameters updated in real timeAnd get->The control angle of the final output is delta _c ，

wherein ,δ_d The target rotation angle based on the reference track is used as a feedforward response to act on a subsequent active disturbance rejection controller, and the main purpose of the target rotation angle is to obtain a rapid and accurate rotation angle response;

the tracking angle delta is obtained by the active disturbance rejection method _c Moment T of (2) _c ：

wherein , x _i i=1, 2 is the system state quantity, r, h is the adjustable parameter, T is the sampling step size, z _i I=1, 2,3 is the expansion state quantity, β _i I=1, 2,3 is an adjustable parameter, α, δ, b is an adjustable parameter, e _i I=1, 2 is the deviation of the input state from the expanded state quantity, u is the regulating moment acting on the steering column;

a coefficient K exists between the control force rejection and the control rotation angle, and the control moment T is calculated by designing an active disturbance rejection algorithm _c To track and control the angle delta _c See equation (16).

7. The method for testing the real vehicle experimental platform for the man-machine co-driving intelligent vehicle according to claim 6, wherein the evaluation of the experimental result is performed by respectively considering four indexes of driving safety, driving accuracy, driving comfort and overall driving experience in the man-machine co-driving process for subjective evaluation of the experimental system for the man-machine co-driving intelligent vehicle; the driving safety comprises whether the driving safety can be improved, whether traffic accidents can be reduced and whether the error behaviors of a driver can be compensated; driving accuracy includes accuracy and smoothness during driving that the driver subjectively understands; driving comfort includes subjective assessment of physical and psychological load of the driver; the overall evaluation comprises the trust degree of a driver on a man-machine co-driving system; the evaluation items respectively comprise gender (1), driving age (2), age (3), driving safety (4-5), driving accuracy (6-7), driving comfort (8-10) and overall evaluation (11-13); scoring included 1-5 points (1-disagree, 2-disagree, 3-neutral, 4-slightly agreeable, 5-very agreeable);

the objective evaluation mainly evaluates the lane keeping performance, the operation load of the driver and the co-driving cooperative performance of the man-machine,

(1) lane keeping performance index:

degree of lane departure: including the relative positional deviation d _rel Relative angle deviation ψ _rel ；

Track tracking accuracy: the deviation between the actual track and the expected track is included;

Traj _error ＝{(x _ego -x _obs ,y _ego -y _obs ):A _error △x+B _error △y+C _error ＝0} (17)

wherein ,A_error ，B _error ，C _error Is a correlation coefficient;

average transit time: under the condition of ensuring the lane departure degree and the track tracking error precision, the vehicle can pass through quickly,

where l is the test road distance, v (d _rel ,ψ _rel ,Traj _pre ) Is the longitudinal speed of the vehicle affected by the degree of lane departure and the tracking error of the track, t is the vehicle transit time, d _pre Is a distance safety threshold, ψ _pre Is the direction angle safety threshold, traj _pre Is a trajectory safety threshold;

(2) driver operation load index:

driver torque: respectively acquiring the driver moment when the driver singly drives and the driver moment T in the man-machine co-driving mode _d,d ,T _d,cop ；

Lateral acceleration and rate of change: respectively acquiring the moment of the driver and the lateral acceleration and the change rate a under the man-machine co-driving mode when the driver singly drives _y ,△a _y ；

(3) And the man-machine co-driving cooperative performance index:

steering correction force: deviation T of controller moment and steering correcting moment _cor ＝T _c -T _al, wherein ,T_al ＝-F _yf (t _m +t _p ) Is steering aligning moment, F _yf Is the lateral force of the tyre, t _m Is the distance t between the ground and the tread of the tire, which is the backward inclined extension line of the main pin _p Is the tire support distance F _yf Can be approximately estimated to obtain t _m ,t _p Usual parameters may be taken;

driver steering correction rate: the number of times the driver corrects the steering wheel in a certain time.