CN102795225A - Method for detecting disturbance state of driver by utilizing driver-side longitudinal control model - Google Patents

Abstract

Translated fromChinese

本发明提供一种利用驾驶员模型检测驾驶员干扰状态的方法，该方法包括：由外部传感器获取车辆的驾驶环境信息，根据车辆的位置和车道边界的关系判断驾驶意图；当驾驶意图是保持车道时，驾驶员模型根据驾驶环境信息和车辆状态信息计算方向盘转角改变量，根据当前车速和由车辆与前车的间隔时间确定的期望纵向加速度计算并输出油门开度；当驾驶意图是换道时，驾驶员模型根据驾驶环境信息和状态信息计算方向盘转角改变量，根据当前车速和由侧向加速度计算的期望纵向加速度计算并输出油门开度；将驾驶员模型计算得到的方向盘转角和油门开度与内部传感器获得的真实驾驶员的控制数据进行比较，判断驾驶员是否受到干扰；重复上述步骤，直至停车。

The present invention provides a method for detecting driver interference state by using a driver model, the method comprising: obtaining the driving environment information of the vehicle by an external sensor, and judging the driving intention according to the relationship between the position of the vehicle and the boundary of the lane; when the driving intention is to keep the lane , the driver model calculates the steering wheel angle change amount according to the driving environment information and vehicle state information, calculates and outputs the accelerator opening according to the current vehicle speed and the expected longitudinal acceleration determined by the interval time between the vehicle and the vehicle in front; when the driving intention is to change lanes , the driver model calculates the steering wheel angle change amount according to the driving environment information and state information, calculates and outputs the throttle opening according to the current vehicle speed and the expected longitudinal acceleration calculated from the lateral acceleration; the steering wheel angle and throttle opening calculated by the driver model Compare with the real driver's control data obtained by internal sensors to determine whether the driver is disturbed; repeat the above steps until the car stops.

Description

Utilize the vertical controlling models of driver side to detect the method for chaufeur disturbance state

Technical field

The present invention relates to a kind of method of utilizing the vertical Comprehensive Control model of driver side to detect the chaufeur disturbance state.This method is aspect the interference of detection chaufeur; Utilize the chaufeur control data (steering wheel angle and accelerator open degree) under the current driving environment of pilot model real-time simulation; The control data of pilot model and the control data of true chaufeur are compared, and whether be interfered according to the comparative result identification of driver.The hardware cost of this method is low, and recognition accuracy is high, is specially adapted to develop the DAS (Driver Assistant System) of vehicle.

Background technology

Chaufeur (is for example disturbed; Absent-minded, eat, drink water/beverage, make a phone call and passenger chat and use onboard system etc.) be a major reason that causes traffic accident; Therefore be necessary to develop the disturbance state that method of inspection detects chaufeur in real time; So that in time remind chaufeur, reduce or avoid the generation of accident in conjunction with traffic.

The detection of chaufeur interference at present mainly concentrates on the physiology that utilizes chaufeur and characterizes, and for example judges through the motion of continuous tracking eyes whether chaufeur receives vision interference etc.Though characterizing, the physiology of chaufeur can embody the situation that chaufeur is disturbed; But also has very strong duplicity; For example when chaufeur receives the interference (for example thinking problem when driving) of cognitive aspect; The facial expression of chaufeur possibly not have anything to change, and is difficult to detect for the existing method of such interference.On the contrary, chaufeur then can reflect driver status truly to the control behavior of vehicle, and the control data when research shows chaufeur control data under normal circumstances and disturbed has very big difference.

Summary of the invention

The purpose of this invention is to provide a kind of new mode that chaufeur disturbs that detects; This method is utilized pilot model; Chaufeur control behavior under the current driving environment of real-time simulation; Through the control data of relatively pilot model real-time simulation and the control data of true chaufeur, can judge exactly whether chaufeur is interfered.

According to the object of the invention; A kind of pilot model comprises: path planning module; Reception is from the driving environment information of vehicle-mounted external sensor, the driving intention of decision-making chaufeur (change or keep track), plans corresponding driving trace, and constantly revises according to the car body position in real time; Take aim at module in advance, according to utilize the characteristic of taking aim in advance of chaufeur to obtain the expected trajectory data from the said data of path planning module; Prediction module obtains the prediction locus data according to the status information from vehicle dynamic model; First comparing module receives expected trajectory data and prediction locus data, and obtains the lateral deviation data through the comparison of these two track datas; The side direction control module outputs to second comparing module and vertical control module respectively according to said lateral deviation direction data calculation dish corner change amount and lateral acceleration and with it; Vertical control module is according to said lateral acceleration calculation expectation longitudinal acceleration, then according to expecting longitudinal acceleration calculation of throttle opening value and exporting vehicle dynamic model to; Second comparing module according to the current steering wheel angle of said steering wheel angle change amount and vehicle dynamic model, is calculated final steering wheel angle and is exported vehicle dynamic model to.

According to the object of the invention; A kind of method of utilizing above-mentioned pilot model to detect the chaufeur disturbance state is proposed; Said method comprises: step a; Obtain the driving environment information of vehicle through vehicle-mounted external sensor,, judge that chaufeur changes or keep the driving intention in track through the relation of vehicle current location and lane boundary of living in relatively; Step b; If judging driving intention is to keep the track; Then pilot model utilizes PD control calculated direction dish corner change amount according to the said driving environment information and the status information of vehicle, and calculates and output accelerator open degree value according to current vehicle speed with by definite expectation longitudinal acceleration of the expected elapsed time (safety time) of vehicle and front truck; Step c; If judge that driving intention is to change; Then pilot model utilizes PD control to obtain steering wheel angle change amount according to the said driving environment information and the status information of vehicle; And, calculate and output accelerator open degree value according to the present speed of vehicle and the expectation longitudinal acceleration of confirming by lateral acceleration; Steps d compares the steering wheel angle of step b or c and accelerator open degree data with the corresponding control data that is obtained true chaufeur by term vehicle internal sensors, judge whether chaufeur is interfered; Step e, repeating step a-d is until parking.

Said method also comprises: before step a, vehicle-mounted external sensor and term vehicle internal sensors are linked to each other with computing machine, debug vehicle-mounted external sensor and term vehicle internal sensors, the initialization pilot model.

Judge the time window of the criterion employing 1s whether chaufeur is interfered; 0.25s renewal amount carry out data handing; Wherein, Judge that criterion that whether chaufeur is interfered utilizes the characteristic of the accumulated deficiency of data in 1s of steering wheel angle, accelerator open degree value and true chaufeur as classification, wherein, judge that the discriminant function whether chaufeur is interfered is that kernel function is the SVMs (SVM) of Gaussian function.

The characteristic of division of aforementioned calculation is inputed to supporting vector machine model, be interfered if the result of model, then can judge chaufeur greater than 0, otherwise, be not interfered.

Pilot model of the present invention is based upon on the existing queuing network cognition system, and when following the trail of expected trajectory, it is the driving performance and the physiology limitation of the true chaufeur of emulation exactly, can embody the driving behavior of experienced driver.The environment sensing sensor in real time of outside vehicle is obtained current driving environment information, and these information real-time are defeated by pilot model.Pilot model calculates the chaufeur control data in real time according to the information of environment sensing sensor.Obtain the real-time control data of true chaufeur through term vehicle internal sensors, just can judge accurately with the control data of true chaufeur whether chaufeur is interfered through the driving data that pilot model relatively calculates.

The invention has the advantages that: proposed a kind ofly to detect the method whether chaufeur is interfered through the Simulation Control data of real-time relatively pilot model and the control data of true chaufeur; This testing process itself can not produce chaufeur and disturb; And hardware realizes that easily cost is low; Drive the driving condition that data can reflect chaufeur truly; Whether be interfered through the emulated data of pilot model and the difference identification chaufeur of the driving data of true chaufeur; Can reduce the False Rate of Interference Detection widely; Improve the accuracy rate that detects, advance the degree of intelligence of DAS (Driver Assistant System).

Description of drawings

Fig. 1 utilizes pilot model to carry out the schematic diagram of chaufeur Interference Detection.

Fig. 2 is the pilot model constructional drawing among Fig. 1.

Fig. 3 utilizes pilot model to carry out the diagram of circuit of chaufeur interference detection method.

The specific embodiment

Describe method of inspection below with reference to accompanying drawings in detail according to chaufeur disturbance state of the present invention.In the present invention, in order to simplify description, suppose that pilot model is in identical expected trajectory with true chaufeur, the invention is not restricted to this certainly.

Fig. 1 utilizes pilot model to carry out the schematic diagram of chaufeur Interference Detection.As shown in Figure 1, it is following to utilize pilot model to carry out the principle of chaufeur Interference Detection: (1) through vehicle-mounted external sensor obtain in real time driving environment information (location information in road type (straight line or curve), the current track of living in of vehicle, the speed of a motor vehicle, with vehicle (front truck and the back car) speed of the distance of the front truck in same track and back car, adjacent lane and with the distance of this car); (2) pilot model receives the data from vehicle-mounted external sensor; Driving behavior under the current driving environment of driving behavior real-time simulation of simulation experienced driver; And output chaufeur control data a to kinetic model closed-loop corrected with the driving behavior that realizes emulation, from the driving control data of pilot model output as the benchmark under the normal driving; (3) when carrying out step (2), obtain the control data of true chaufeur in real time to vehicle through onboard sensor; (4) comparison module receives from the pilot model control data of step (2) with from the control data of the true chaufeur of step (3), just can accurately judge with the control data of true chaufeur whether chaufeur is interfered through pilot model relatively.

Be based upon on the existing queuing network cognition system at the pilot model shown in Fig. 1, when following the trail of expected trajectory, it is the driving performance and the physiology limitation of the true chaufeur of emulation exactly, can embody the driving behavior of experienced driver.Describe pilot model in detail with reference to Fig. 2 below.

As shown in Figure 2, pilot model comprises path planning module, takes aim at module, prediction module,comparing module 1, comparing module 2, side direction control module, vertical control module etc. in advance.

On the one hand; Driving environment information input path planning module from vehicle-mounted external sensor; The path planning module decision-making goes out the driving intention (change or keep track) of chaufeur and plans corresponding driving trace; And constantly revise according to the car body position in real time, take aim at module in advance according to utilizing the characteristic of taking aim in advance of chaufeur to obtain the expected trajectory data from the said data of path planning module.

On the other hand, from the car status information (S of vehicle dynamic model_n(x, y, a_x, a_y, v_x, v_y, yaw), wherein x is a car body side coordinate, y is the car body along slope coordinate, a_xBe longitudinal acceleration, a_yBe lateral acceleration, v_xBe longitudinal velocity, v_yBe side velocity, yaw is the car body yaw angle) the input prediction module, prediction module is according to said status information prediction of output track data.Expected trajectory data and prediction locus data are all importedcomparing module 1, obtain lateral deviation data E (will be described below) throughcomparing module 1 these two track datas are compared.The side direction control module is according to outputing to comparing module 2 and vertical control module respectively from the lateral deviation data E calculated direction dish corner change amount Δ δ (will be described below) ofcomparing module 1 and with it.Vertically control module is calculated final accelerator open degree value α (will be described below) and is exported vehicle dynamic model to, and comparing module 2 is according to calculating output final steering wheel angle δ (will be described below) and export it to vehicle dynamic model from the steering wheel angle change amount Δ δ of side direction control module and the current steering wheel angle (will be described below) of vehicle dynamic model.Realize driving behavior closed-loop corrected of pilot model thus.

Because the implementation (for example, software mode, hardware mode etc.) of the control of the comparison/comparison-control module of the sensor-comparison module/comparing module that when describing Fig. 1 and Fig. 2, relates to belongs to prior art, no longer is repeated in this description at this.

Describing in detail with reference to Fig. 3 below utilizes pilot model to carry out the method for chaufeur Interference Detection.

It is following to utilize pilot model to carry out the process of chaufeur Interference Detection:

Instep 301 and 302; Debug vehicle-mounted external sensor and term vehicle internal sensors; Make its normal operation; Each module of initialization pilot model also guarantees that each module clock is consistent, guarantees that simultaneously the time that true chaufeur steering vehicle advances is consistent with the enabling time of each sensor and pilot model.Each sensor and computing machine (for example, car-mounted computer) are linked to each other, to communicate with pilot model.

Instep 303, the chaufeur steering vehicle advances, and vehicle-mounted external sensor obtains current driving environment information in real time, and term vehicle internal sensors is obtained vehicle state information in real time.Certainly, each sensor possibly interrupted by certain situation (for example, sensor fault, vehicle parking etc.).

Instep 304 and 305; Pilot model obtains the driving environment information of vehicle according to vehicle-mounted external sensor; Through the relation of vehicle current location (that is, the position in track of living in) and lane boundary of living in relatively, the driving intention of judgement chaufeur (change or keep track).

The driving intention that obtains according to step 305 (change or keep track); Pilot model is according to the current driving environment information of obtaining from vehicle-mounted external sensor; Carry out path planning according to the drive safety criterion, and constantly revise in real time according to the location information of vehicle.The path planning that obtains according to driving intention is promptly as the expected trajectory of pilot model, and pilot model is followed the trail of this expected trajectory, obtains corresponding driving behavior benchmark.Like this, the expected trajectory that obtains through path planning can guarantee that pilot model is in identical or roughly the same expected trajectory with true chaufeur.

According to the driving intention thatstep 305 obtains, pilot model is followed the trail of the expected trajectory that instep 305, obtains according to corresponding driving intention (change or keeps track) selection Different control mode.

If the driving intention that instep 305, obtains is to keep the track, then in step 306,307,308 and 309, will be through two cars (Ben Che and front truck thereof) the spacing d of vehicle-mounted external sensor acquisition_n, the front truck speed v_HnReach the car status information S that obtains through term vehicle internal sensors_n(x_n, y_n, a_Xn, a_Yn, v_Xn, v_Yn, yaw_n) send pilot model to.Pilot model take aim at the path planning that module is obtained throughstep 305 in advance, obtain the current time (T that takes aim in advance_p=expected trajectory point P in 1.5s)_n(x_n, y_n), prediction module obtains car status information S through term vehicle internal sensors_n(x_n, y_n, a_Xn, a_Vn, v_Xn, v_Vn, yaw_n) and predict and take aim at time T in advance_pInterior vehicle the position coordinate that will arriveJust can obtain the lateral position error E of expected trajectory and prediction locus thus_nIn order accurately to follow the trail of expected trajectory, will adjust steering wheel angle and reduce the lateral position deviation.In pilot model, utilize PD control to obtain steering wheel angle change amount.Aspect vertical,, prevent that rear-end collision from taking place, and guarantee two following distances (perhaps greater than safety time) in safe range for the safety that guarantees to drive.For this reason, calculate the pitch time t of two cars_Cn, expectation longitudinal acceleration a_XnThen with the pitch time t of two cars_CnWith safety time t_f(t_f=4s) the proportional relation of difference, the relation according to vehicle dynamics obtains expecting longitudinal acceleration and the cooresponding accelerator open degree of present speed (quicken on the occasion of representative, negative value is represented brake) thus, formula is following:

$E_n=y_n^q-y_n---\left(1\right)$

$a_{\mathrm{yn}}=\frac{2\·(E_n-v_{\mathrm{yn}}\·T_p)}{T_p^2}\left(2\right)$

$a_{\mathrm{yn}}^{\′}=\frac{a_{\mathrm{yn}}-a_{y(n-1)}}{T_p}---\left(3\right)$

Δδ_n＝k_p·a_yn+k_d·a′_yn （4）

δ_n=δ_n-1+Δδ_n （5）

$t_{\mathrm{cn}}=[d_n-(v_{\mathrm{xn}}\·T_p+0.5\·T_p^2-v_{\mathrm{hn}}{T}_p)]/(v_{\mathrm{xn}}+a_{\mathrm{xn}}\·T_p)---\left(6\right)$

$\begin{array}{c}{a^d}_{\mathrm{xn}}=K\·(t_{\mathrm{cn}}-t_f)---\left(7\right)\\ {\mathrm{\α}}_n=f(v_{\mathrm{xn}},{a^d}_{\mathrm{xn}})---\left(8\right)\end{array}$

By formula (1), the lateral position deviation E in n step_nDeduct the side direction coordinate of expected trajectory point through the side direction coordinate of prediction locus point.By formula (2), obtain n step side velocity v according to term vehicle internal sensors_Vn, calculate the lateral acceleration a that arrives desired location_YnBy formula (3), can obtain the derivative a ' of n step lateral acceleration divided by the time of taking aim in advance through the difference of n step lateral acceleration and n-1 step lateral acceleration_YnBy formula (4),, obtain n step steering wheel angle change amount Δ δ through PD control_nBy formula (5), the steering wheel angle in n-1 step adds steering wheel angle change amount Δ δ_nJust can obtain final steering wheel angle δ_nAspect the vertical controlled variable of calculating (being throttle or brake aperture), calculate the pitch time t that n goes on foot two cars by formula (6)_Cn, obtain expecting longitudinal acceleration by formula (7)The question blank that is provided by formula (8) at last obtains final accelerator open degree value α_n, f (v wherein_Xn, a^d_Xn) be kinetic function expression formula about vehicle motor, the accelerator open degree that different longitudinal accelerations is corresponding different with longitudinal velocity.

Belong to prior art owing to how to realize PD control, no longer describe at this.

Find out that from top description pilot model is known expected trajectory according to vehicle-mounted external sensor,, export to vehicle dynamic model then through the controlled order of the driving performance of emulation experienced driver under the normal driving situation.

If the driving intention that instep 305, obtains is to change, then in step 310,311,312 and 313, the driving environment information of the adjacent lane that will obtain through vehicle-mounted external sensor and the car status information S that obtains through term vehicle internal sensors_n(x_n, y_n, a_Xn, a_Vn, v_Xn, v_Vn, yaw_n) send pilot model to.The method of calculating of taking aim at module, prediction module and steering wheel angle in advance in the pilot model and step 306,307,308 and 309 identical, difference is aspect longitudinal acceleration.Physiology limitation for the true chaufeur of emulation; When changing, the longitudinal acceleration of expectation is according to lateral acceleration decision, obtains expecting that according to the relation of vehicle dynamics the cooresponding accelerator open degree of longitudinal acceleration (quickens on the occasion of representative; Negative value is represented brake), formula is following:

a^d_xn=K·a_yn+a_d （9）

α_n=f(v_xn,a^d_xn) （10）

Calculate the needed steering wheel angle of tracking expected trajectory by top formula (1)-(5).By the longitudinal acceleration of formula (9) calculation expectation, wherein a_YnBe the lateral acceleration that formula (2) calculates, K, a_dIt is constant.The question blank that is provided by formula (10) calculates present speed and the accelerator open degree value of expecting under the longitudinal acceleration.

Can obtain the driving behavior data M (δ of pilot model emulation under normal driving situation (for example, keep track or change) by step 306-309 or step 310-313_Mn, α_Mn), obtain true chaufeur controlled in real-time data D (δ by term vehicle internal sensors_Dn, α_Dn), judge through the comparison module (see figure 1) whether chaufeur receives the interference of other tasks then, shown in the following formula of its decision criteria:

${\mathrm{RMSE}}_{\mathrm{\δ}}=\sqrt{\frac{\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{({\mathrm{\δ}}_M\left(i\right)-{\mathrm{\δ}}_D\left(i\right))}^2}{n}}---\left(11\right)$

${\mathrm{RMSE}}_{\mathrm{\α}}=\sqrt{\frac{\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{({\mathrm{\α}}_M\left(i\right)-{\mathrm{\α}}_D\left(i\right))}^2}{n}}---\left(12\right)$

$F_d=\mathrm{SVM}({\mathrm{RMSE}}_{\mathrm{\δ}},{\mathrm{RMSE}}_{\mathrm{\α}})$

In this decision criteria, adopt the time window of 1s, the renewal amount of 0.25s.Steering wheel angle, accelerator open degree value that calculates pilot model by formula (11), (12) respectively and the characteristic of the accumulated deficiency of data in 1s of true chaufeur as classification; Utilize the gaussian kernel function certificate, formula (11), (12) calculated feature values are input in the SVMs function (SVM), according to result of calculation; If output valve can judge promptly that greater than 0 influence (step 314) that whether chaufeur be interfered (promptly; The result of model the characteristic of division of aforementioned calculation inputed to supporting vector machine model, if, can judge then that chaufeur is interfered greater than 0; Otherwise, be not interfered).

Vehicle-mounted external sensor and term vehicle internal sensors are obtained driving environment information and vehicle interior status information in real time, constantly send information to pilot model, promptly repeat above-mentioned step 304-315, until parking always.

Claims (7)

Translated fromChinese

1.一种利用驾驶员模型检测驾驶员干扰状态的方法，所述方法包括：1. A method utilizing a driver model to detect driver disturbance state, said method comprising:步骤a，通过车载外部传感器获取车辆的驾驶环境信息，通过比较车辆当前位置和所处车道边界的关系，判断驾驶员换道或保持车道的驾驶意图；Step a, obtain the driving environment information of the vehicle through the vehicle external sensor, and judge the driving intention of the driver to change lanes or keep the lane by comparing the relationship between the current position of the vehicle and the boundary of the lane;步骤b，如果驾驶意图是保持车道，则驾驶员模型根据车辆的所述驾驶环境信息和状态信息利用PD控制计算方向盘转角改变量，并根据当前车速和由车辆与前车的间隔时间确定的期望纵向加速度计算并输出油门开度；Step b, if the driving intention is to keep the lane, the driver model uses PD control to calculate the amount of change in the steering wheel angle according to the driving environment information and state information of the vehicle, and according to the current vehicle speed and the desired time determined by the interval between the vehicle and the vehicle in front Calculate the longitudinal acceleration and output the throttle opening;步骤c，如果驾驶意图是换道，则驾驶员模型根据车辆的所述驾驶环境信息和状态信息利用PD控制计算方向盘转角改变量，并根据车辆的当前速度和由侧向加速度计算得到的期望纵向加速度计算并输出油门开度；Step c, if the driving intention is to change lanes, the driver model uses PD control to calculate the steering wheel angle change amount according to the driving environment information and state information of the vehicle, and calculates the desired longitudinal direction according to the current speed of the vehicle and the lateral acceleration Acceleration calculation and output throttle opening;步骤d，将步骤b或c的方向盘转角和油门开度数据与由车辆内部传感器获得真实驾驶员的控制数据进行比较，判断驾驶员是否受到干扰；Step d, comparing the steering wheel angle and accelerator opening data in step b or c with the real driver's control data obtained from the vehicle's internal sensors, to determine whether the driver is disturbed;步骤e，重复步骤a-d，直至停车，Step e, repeat steps a-d until parking,其中，驾驶员模型包括：Among them, the driver model includes:路径规划模块，接收来自车载外部传感器的驾驶环境信息，决策驾驶员换道或保持车道的驾驶意图，规划相应的行驶轨迹，并实时根据车体位置不断修正；The path planning module receives the driving environment information from the vehicle's external sensors, decides the driver's driving intention to change lanes or keep the lane, plans the corresponding driving trajectory, and continuously corrects it in real time according to the position of the vehicle body;预瞄模块，根据来自路径规划模块的所述数据利用驾驶员的预瞄特性得到预期轨迹数据；The preview module utilizes the preview feature of the driver to obtain expected trajectory data according to the data from the path planning module;预测模块，根据车辆状态信息计算预测轨迹数据；Prediction module, calculates forecast track data according to vehicle state information;第一比对模块，接收预期轨迹数据和预测轨迹数据，并通过这两个轨迹数据的比对获得侧向偏差数据；The first comparison module receives expected trajectory data and predicted trajectory data, and obtains lateral deviation data by comparing the two trajectory data;侧向控制模块，根据所述侧向偏差数据计算方向盘转角改变量和侧向加速度并将其分别输出到第二比对模块和纵向控制模块；The lateral control module calculates the steering wheel angle change and lateral acceleration according to the lateral deviation data and outputs them to the second comparison module and the longitudinal control module respectively;纵向控制模块，根据期望纵向加速度和当前汽车的速度计算油门开度值并输出至车辆动力学模型；The longitudinal control module calculates the accelerator opening value according to the desired longitudinal acceleration and the current speed of the vehicle and outputs it to the vehicle dynamics model;第二比对模块，根据所述方向盘转角改变量和车辆动力学模型的当前方向盘转角，计算最终方向盘转角并输出至车辆动力学模型，由此实现驾驶员模型驾驶行为的闭环校正。The second comparison module calculates the final steering wheel angle according to the change amount of the steering wheel angle and the current steering wheel angle of the vehicle dynamics model and outputs it to the vehicle dynamics model, thereby realizing closed-loop correction of the driving behavior of the driver model.2.根据权利要求1所述的方法，所述方法还包括：在步骤a之前，将车载外部传感器以及车辆内部传感器与计算机相连，调试车载外部传感器及车辆内部传感器，初始化驾驶员模型。 2. The method according to claim 1, further comprising: before step a, connecting the vehicle external sensors and the vehicle internal sensors with the computer, debugging the vehicle external sensors and the vehicle internal sensors, and initializing the driver model. the3.根据权利要求2所述的方法，其中，计算机是车载计算机。3. The method of claim 2, wherein the computer is an onboard computer.4.根据权利要求1所述的方法，其中，根据步骤a，驾驶员模型按照驾驶安全性准则进行路径规划，以获得预期轨迹。4. The method according to claim 1, wherein, according to step a, the driver model performs path planning according to driving safety criteria to obtain an expected trajectory.5.根据权利要求1所述的方法，其中，所述驾驶环境信息包括两车间距、前车速度，所述状态信息通过车辆内部传感器获得，5. The method according to claim 1, wherein the driving environment information includes the distance between two vehicles and the speed of the vehicle in front, and the state information is obtained by a vehicle internal sensor,其中，驾驶员模型将当前预瞄时间内的预期轨迹点与车辆所要到达的位置坐标比较，以得到预期轨迹和预测轨迹的侧向位置误差，以计算方向盘转角改变量。Among them, the driver model compares the expected trajectory point within the current preview time with the position coordinates to be reached by the vehicle to obtain the lateral position error between the expected trajectory and the predicted trajectory, and calculate the steering wheel angle change.6.根据权利要求1所述的方法，其中，判断驾驶员是否受到干扰的准则采用1s的时窗，0.25s的更新量，6. The method according to claim 1, wherein the criterion for judging whether the driver is disturbed adopts a time window of 1s, an update amount of 0.25s,其中，判断驾驶员是否受到干扰的准则是利用模型得到的方向盘转角、油门开度值与真实驾驶员的相应控制数据在1s内的累积差值作为分类的特征，Among them, the criterion for judging whether the driver is disturbed is to use the cumulative difference between the steering wheel angle and accelerator opening value obtained by the model and the corresponding control data of the real driver within 1 second as the classification feature,其中，判断驾驶员是否受到干扰的函数是利用核函数为高斯函数的支持向量机(SVM)。Wherein, the function for judging whether the driver is disturbed is a support vector machine (SVM) whose kernel function is a Gaussian function.7.根据权利要求6所述的方法，其中，将上述计算的分类特征输入支持向量机模型，如果模型的结果大于0，则可以判断出驾驶员受到干扰，否则，没有受到干扰。 7. The method according to claim 6, wherein the above-mentioned calculated classification features are input into the support vector machine model, if the result of the model is greater than 0, it can be judged that the driver is disturbed, otherwise, it is not disturbed. the

Images