阅读说明：本技术 一种基于驾驶员换道行为预测的自动转向灯控制系统及控制方法 (Automatic steering lamp control system and control method based on driver lane change behavior prediction ) 是由 何友国 龚星 蔡英凤 袁朝春 陈龙 王海 于 2019-08-09 设计创作，主要内容包括：本发明公开了一种基于驾驶员换道行为预测的自动转向灯控制系统及控制方法,由车载信息采集装置、驾驶员信息采集摄像头和视频处理器硬件平台组成。车载信息采集装置负责采集车辆纵向加速度、横向加速度、方向盘转角、转向灯开关信号。驾驶员信息采集摄像头负责采集驾驶员头部运动的图像。视频处理器硬件平台结合车辆运行参数及驾驶员头部朝向信息,基于驾驶员换道行为预测模型对驾驶员换道行为进行预测；并根据预测结果和转向灯的开关情况,进行转向灯自动控制。本发明能够有效避免在换道过程中由于未开转向灯而导致的交通事故,提高车辆驾驶安全性。本发明也能够有效避免在换道完成后由于未关闭转向灯对其他车辆造成的影响,提高交通的流动性。(The invention discloses an automatic steering lamp control system and a control method based on lane changing behavior prediction of a driver. The vehicle-mounted information acquisition device is responsible for acquiring longitudinal acceleration, transverse acceleration, steering wheel turning angles and steering lamp switch signals of the vehicle. The driver information acquisition camera is responsible for acquiring images of the head movement of the driver. The video processor hardware platform predicts the lane changing behavior of the driver based on a driver lane changing behavior prediction model by combining vehicle operation parameters and the head orientation information of the driver; and automatically controlling the turn lights according to the prediction result and the on-off condition of the turn lights. The invention can effectively avoid traffic accidents caused by not turning on the steering lamp in the lane changing process and improve the driving safety of vehicles. The invention can also effectively avoid the influence on other vehicles caused by not turning off the steering lamp after the lane change is finished, and improve the traffic mobility.)

1. An automatic steering lamp control system based on driver lane changing behavior prediction is characterized by comprising a vehicle-mounted information acquisition device, a driver information acquisition camera and a video processor hardware platform;

the vehicle-mounted information acquisition device is responsible for acquiring vehicle operation parameters; the driver information acquisition camera is responsible for acquiring images of the head movement of the driver; the video processor predicts the lane changing behavior of the driver based on a lane changing behavior prediction model of the driver by combining vehicle operation parameters and head orientation information of the driver; and automatically controlling the turn lights according to the prediction result and the on-off condition of the turn lights.

2. The system of claim 1, wherein the vehicle operating parameters comprise vehicle longitudinal acceleration, lateral acceleration, steering wheel angle, and turn signal.

3. The automatic turn signal lamp control system based on the driver lane changing behavior prediction as claimed in claim 2, wherein the vehicle-mounted information acquisition device comprises a vehicle longitudinal acceleration sensor, a lateral acceleration sensor, a steering wheel turning angle sensor, a turn signal lamp switch signal sensor and a signal conditioning circuit.

4. The automatic turn signal lamp control system based on the driver lane changing behavior prediction as claimed in claim 1, wherein the driver information collection camera is a CCD vehicle-mounted camera.

5. The automatic turn signal lamp control system based on the driver lane changing behavior prediction as claimed in claim 1, wherein the video processor hardware platform adopts a high-performance video processor and is responsible for analyzing the head moving image of the driver to obtain the time duty ratios of the head of the driver towards the area right in front of the vehicle, the left side rearview mirror area and the right side rearview mirror area in a fixed time window; meanwhile, vehicle operation information and driver head orientation information are integrated, and the lane changing behavior of the driver is predicted based on a lane changing behavior prediction model of the driver; and then designing an automatic steering lamp control algorithm according to the prediction result and the switching condition of the steering lamp.

6. The automatic turn signal lamp control system based on the driver lane changing behavior prediction as claimed in claim 5, wherein the driver lane changing behavior prediction model is a driver lane changing behavior prediction HMM prediction model λ ═ (N, M, pi, a, B) established based on HMM theory, wherein:

the driver changes lane behavior state S: (S) ═ S₁,S₂,…S_N) The state at the moment t is q_t，q_tE is S, the state number N of the item is 3, wherein S₁For left lane change behavior, S₂For lane keeping behavior, S₃Right lane change behavior;

observation sequence V: v ═ V (V)₁,v₂,…v_M) And the observed event at the time t is O_tThe number of observed values M of this item is 6, where v is₁Is the longitudinal acceleration of the vehicle, v₂As lateral acceleration of the vehicle, v₃Is the steering wheel angle, v₄Duty ratio for driver's head facing the area directly in front of vehicle, v₅Duty cycle, v, for the driver's head towards the left side mirror area₆The driver's head towards left side rearview mirror zone duty cycle;

pi: the probability vector of the initial state of the lane change behavior of the driver, pi ═ pi (pi)₁,π₂,…π_N) In which pi_i＝P(q₁＝S_i)；

A: state transition matrix, i.e. the state transition matrix for the driver to change lanes, a ═ a_ij}_N×NWherein a is_ij＝P(q_t+1＝S_j|q_t＝S_i)，1≤i,j≤N；

B: observation event probability distribution matrix, i.e. probability of occurrence of each observation state under state S of different lane change behaviors of driver, B ═ B_jk}_N×MWherein b is_jk＝P[O_t＝v_k|q_t＝S_j]，1≤j≤N，1≤k≤M。

7. The automatic turn signal lamp control system based on the driver lane change behavior prediction as claimed in claim 5, wherein the turn signal lamp automatic control algorithm:

(a) when the driver has a left lane changing behavior, if the left turn light is not turned on at the moment, the system automatically turns on the left turn light;

(b) when the driver has right lane changing behavior, if the right turn light is not turned on at the moment, the system automatically turns on the right turn light;

(c) when the driver has lane keeping behavior, the system automatically turns off the turn signal if the turn signal is on at this time.

8. An automatic turn signal lamp control method based on the prediction of lane changing behavior of a driver is characterized by comprising the following steps:

vehicle-mounted information acquisition: collecting vehicle operation parameters; the vehicle operation parameters comprise vehicle longitudinal acceleration, transverse acceleration, steering wheel turning angle and steering lamp switch signals;

collecting driver information: collecting images of the head movement of a driver;

information processing and prediction: predicting the lane changing behavior of the driver based on a driver lane changing behavior prediction model by combining vehicle operation parameters and the head orientation information of the driver; automatically controlling the steering lamp according to the prediction result and the switching condition of the steering lamp;

the specific control method of the automatic steering lamp comprises the following steps:

(1) information acquisition: the method comprises the following steps of (1) carrying out longitudinal acceleration on a vehicle, transverse acceleration on the vehicle, steering wheel turning angles, turn signal on/off of a steering lamp and head motion images of a driver;

(2) data processing: analyzing the moving image of the head of the driver to obtain the time duty ratios of the head of the driver to the area right in front of the vehicle, the area of the left side rearview mirror and the area of the right side rearview mirror in a fixed time window;

(3) predicting lane changing behavior of a driver: predicting the lane changing behavior of the driver according to the offline trained lane changing behavior prediction model of the driver;

(4) automatic control of the steering lamp: designing automatic control of the steering lamp according to the prediction result and the switching condition of the steering lamp;

(a) when the driver has a left lane changing behavior, if the left turn light is not turned on at the moment, the system automatically turns on the left turn light;

(b) when the driver has right lane changing behavior, if the right turn light is not turned on at the moment, the system automatically turns on the right turn light;

(c) when the driver has lane keeping behavior, the system automatically turns off the turn signal if the turn signal is on at this time.

9. The automatic turn signal lamp control method based on the driver lane changing behavior prediction as claimed in claim 8, wherein the driver lane changing behavior prediction model design method comprises:

based on the HMM theory, establishing a prediction HMM (HMM prediction model) lambda (N, M, pi, A and B) of the lane changing behavior prediction of the driver, wherein:

the driver changes lane behavior state S: (S) ═ S₁,S₂,…S_N) The state at the moment t is q_t，q_tE is S, the state number N of the item is 3, wherein S₁For left lane change behavior, S₂For lane keeping behavior, S₃Right lane change behavior;

observation sequence V: v ═ V (V)₁,v₂,…v_M) And the observed event at the time t is O_tThe number of observed values M of this item is 6, where v is₁Is the longitudinal acceleration of the vehicle, v₂As lateral acceleration of the vehicle, v₃Is the steering wheel angle, v₄Duty ratio for driver's head facing the area directly in front of vehicle, v₅Duty cycle, v, for the driver's head towards the left side mirror area₆The driver's head towards left side rearview mirror zone duty cycle;

pi: the probability vector of the initial state of the lane change behavior of the driver, pi ═ pi (pi)₁,π₂,…π_N) In which pi_i＝P(q₁＝S_i)；

A: state transition matrix, i.e. the state transition matrix for the driver to change lanes, a ═ a_ij}_N×NWherein a is_ij＝P(q_t+1＝S_j|q_t＝S_i)，1≤i,j≤N；

B: observation event probability distribution matrix, i.e. probability of occurrence of each observation state under state S of different lane change behaviors of driver, B ═ B_jk}_N×MWherein b is_jk＝P[O_t＝v_k|q_t＝S_j]，1≤j≤N，1≤k＝M。

Technical Field

The invention relates to the field of intelligent automobiles, in particular to an automatic steering lamp control system and a control method based on lane changing behavior prediction of a driver.

Background

With the continuous development of national economy, the holding capacity of automobiles and the number of drivers are continuously increased, and the driving level and safety awareness of non-professional drivers are uneven. Road traffic accidents are the first public nuisance of human society, the problem of death and injury caused by road traffic accidents has become one of the key problems concerned by the world, and the occurrence of traffic accidents brings about serious disasters to people and causes huge losses to the nation and the society. According to statistics, 130 million people die of traffic accidents every year around the world, and 2000 to 5000 million people cause non-lethal injury due to the traffic accidents. Wherein 4% -10% of traffic accidents are caused by the lane change behavior of the vehicle. Research aiming at lane changing operation habits of drivers in China shows that the average lane changing frequency is about 68 times/hour during free driving, and the average lane changing frequency is about 127 times/hour during lane changing driving. This means that most chinese drivers are not satisfied with following a preceding slow car, but tend to seek greater driving advantage by changing lanes or overtaking. However, the driver has a low awareness of changing lanes using the turn signal, and the usage rate of the turn signal is less than 50% in the free driving case, and is only about 65% in the lane changing driving case. In the lane changing process, if the turn light can not be turned on in advance, the backward vehicle can not respond to the lane changing behavior of the forward vehicle in time, and then traffic accidents are caused. Meanwhile, some drivers forget to turn off the turn signal lamps after lane changing is completed, so that driving behaviors of surrounding vehicles are influenced, and traffic mobility is influenced. Although automobile manufacturers at home and abroad currently develop a large number of active safety systems, such as a wheel anti-lock Brake System (anti Brake System), an antiskid Brake System (parking Brake System), a Collision prevention System (Collision Avoidance System), a Lane Departure Warning System (Lane Departure Warning System), and the like, no report has been made on the active safety systems in which a driver does not correctly use a turn signal behavior during Lane changing.

By looking up the data, no report is found on the application of the automatic turn signal system based on the prediction of the lane changing behavior of the driver at present.

Disclosure of Invention

Aiming at the problems, the invention provides an automatic steering lamp control system based on the prediction of the lane changing behavior of a driver, which comprises the steps of collecting vehicle parameters through a vehicle-mounted sensor, collecting driving behavior parameters of the driver through a video processor, establishing a prediction model of the lane changing behavior of the driver, and controlling whether to use a steering lamp or not according to a prediction result and the driver.

The invention provides an automatic steering lamp control system based on the prediction of lane changing behavior of a driver. The vehicle-mounted information acquisition device is responsible for acquiring longitudinal acceleration, transverse acceleration, steering wheel turning angles and steering lamp switch signals of the vehicle. The driver information acquisition camera is responsible for acquiring images of the head movement of the driver. The video processor hardware platform predicts the lane changing behavior of the driver based on a driver lane changing behavior prediction model by combining vehicle operation parameters and the head orientation information of the driver; and automatically controlling the turn lights according to the prediction result and the on-off condition of the turn lights. The invention can effectively avoid traffic accidents caused by not turning on the steering lamp in the lane changing process and improve the safety of vehicle driving. Meanwhile, the invention can also effectively avoid the influence on other vehicles caused by not turning off the steering lamp after the lane change is finished, thereby improving the mobility of traffic.

The technical scheme of the invention is as follows:

an automatic turn light control system based on the prediction of lane changing behavior of a driver is composed of a vehicle-mounted information acquisition device, a driver information acquisition camera and a video processor hardware platform.

The vehicle-mounted information acquisition device is composed of a vehicle longitudinal acceleration sensor, a transverse acceleration sensor, a steering wheel corner sensor, a steering lamp switch signal sensor and a signal conditioning circuit, is responsible for acquiring the vehicle longitudinal acceleration, the transverse acceleration, the steering wheel corner and the steering lamp switch signal and sending the signals to a video processor hardware platform through a CAN bus. When the longitudinal acceleration, the transverse acceleration, the steering wheel corner and the steering lamp switch signal of the vehicle CAN be directly analyzed through a vehicle CAN bus communication protocol, the system does not need to be provided with a vehicle-mounted information acquisition device again, and the video processor hardware platform directly analyzes the longitudinal acceleration, the transverse acceleration, the steering wheel corner and the steering lamp switch signal of the vehicle through the CAN bus.

The driver information acquisition camera adopts a CCD vehicle-mounted camera, is arranged on a front windshield of a vehicle and is responsible for acquiring head motion images of a driver.

The video processor hardware platform adopts a high-performance video processor and is responsible for analyzing the head moving images of the driver to obtain the time duty ratios of the head of the driver facing to the area right in front of the vehicle, the left side rearview mirror area and the right side rearview mirror area in a fixed time window. And meanwhile, vehicle operation information and driver head orientation information are integrated, and the lane changing behavior of the driver is predicted based on the lane changing behavior prediction model of the driver. And then designing an automatic steering lamp control algorithm according to the prediction result and the switching condition of the steering lamp.

The invention also provides a control method of the automatic steering lamp control system based on the lane changing behavior prediction of the driver, which comprises the following steps:

vehicle-mounted information acquisition: collecting vehicle operation parameters; the vehicle operation parameters comprise vehicle longitudinal acceleration, transverse acceleration, steering wheel turning angle and steering lamp switch signals;

collecting driver information: collecting images of the head movement of a driver;

information processing and prediction: predicting the lane changing behavior of the driver based on a driver lane changing behavior prediction model by combining vehicle operation parameters and the head orientation information of the driver; and automatically controlling the turn lights according to the prediction result and the on-off condition of the turn lights.

Further, the driver lane change behavior prediction model in the control system and the control method is as follows:

based on the HMM theory, establishing a prediction HMM (HMM prediction model) lambda (N, M, pi, A and B) of the lane changing behavior prediction of the driver, wherein:

the driver changes lane behavior state S: (S) ═ S₁,S₂,…S_N) The state at the moment t is q_t，q_tE is S, the state number N of the invention is 3, wherein S₁For left lane change behavior, S₂For lane keeping behavior, S₃Right lane change behavior;

observation sequence V: v ═ V (V)₁,v₂,…v_M) And the observed event at the time t is O_tThe observation number M of the present invention is 6, wherein v is₁Is the longitudinal acceleration of the vehicle, v₂As lateral acceleration of the vehicle, v₃Is the steering wheel angle, v₄Duty ratio for driver's head facing the area directly in front of vehicle, v₅Duty cycle, v, for the driver's head towards the left side mirror area₆The driver's head towards left side rearview mirror zone duty cycle;

pi: the probability vector of the initial state of the lane change behavior of the driver, pi ═ pi (pi)₁,π₂,…π_N) In which pi_i＝P(q₁＝S_i) (ii) a Wherein q is₁Denotes the initial state, P (q)₁＝S_i) Represents the state S_iAs the probability of the initial state, the sum of all the probabilities is 1;

a: state transition matrix, i.e. the state transition matrix for the driver to change lanes, a ═ a_ij}_N×NWherein a is_ij＝P(q_t+1＝S_j|q_t＝S_i)，1≤i,j≤N；

B: observation event probability distribution matrix, i.e. probability of occurrence of each observation state under state S of different lane change behaviors of driver, B ═ B_jk}_N×MWherein b is_jk＝P[O_t＝v_k|q_t＝S_j]，1≤j≤N，1≤k≤M。

Further, in the control system and the control method, the automatic control algorithm of the turn signal lamp is as follows:

(1) information acquisition: the method comprises the following steps of (1) carrying out longitudinal acceleration on a vehicle, transverse acceleration on the vehicle, steering wheel turning angles, turn signal on/off of a steering lamp and head motion images of a driver;

(2) data processing: and analyzing the moving image of the head of the driver to obtain the time duty ratios of the head of the driver to the area right in front of the vehicle, the area of the left side rearview mirror and the area of the right side rearview mirror in the fixed time window.

(3) Predicting lane changing behavior of a driver: predicting the lane changing behavior of the driver according to the offline trained lane changing behavior prediction model of the driver;

(4) automatic control of the steering lamp: designing automatic control of the steering lamp according to the prediction result and the switching condition of the steering lamp;

(a) when the driver has a left lane changing behavior, if the left turn light is not turned on at the moment, the system automatically turns on the left turn light;

(b) when the driver has right lane changing behavior, if the right turn light is not turned on at the moment, the system automatically turns on the right turn light;

(c) when the driver has lane keeping behavior, the system automatically turns off the turn signal if the turn signal is on at this time.

The invention has the beneficial effects that:

the invention predicts the lane changing behavior of the driver based on the vehicle kinematic parameters and the image of the head movement of the driver. And automatically controlling the steering lamp according to the prediction result and the switching condition of the steering lamp. The invention can effectively avoid traffic accidents caused by not turning on the steering lamp in the lane changing process and improve the safety of vehicle driving. Meanwhile, the invention can also effectively avoid the influence on other vehicles caused by not turning off the steering lamp after the lane change is finished, thereby improving the mobility of traffic.

Drawings

FIG. 1 is a schematic diagram of the system of the present invention.

Fig. 2 is a schematic circuit diagram of the vehicle-mounted information acquisition device according to the present invention.

FIG. 3 is a circuit diagram of a hardware platform of a video processor according to the present invention.

FIG. 4 is a circuit diagram of a hardware platform of a video processor according to the present invention.

FIG. 5 is a flowchart of the off-line training process of the lane change behavior prediction model of the present invention.

FIG. 6 is a flow chart illustrating the prediction of lane change behavior of a driver according to the present invention.

FIG. 7 is a flow chart of an automatic turn signal control algorithm of the present invention.

The accessories in the figure are as follows: the system comprises a driver information acquisition camera 1, a video processor hardware platform 2 and a vehicle-mounted information acquisition device 3.

Detailed Description

The invention will be further explained with reference to the drawings.

Referring to fig. 1, 2, 3 and 4, an automatic turn signal lamp control system based on the lane changing behavior prediction of a driver is composed of a vehicle-mounted information acquisition device 3, a driver information acquisition camera 1 and a video processor hardware platform 2.

The vehicle-mounted information acquisition device 3 is composed of a vehicle longitudinal acceleration sensor, a transverse acceleration sensor, a steering wheel corner sensor, a steering lamp switch signal sensor and a signal conditioning circuit, is responsible for acquiring vehicle longitudinal acceleration, transverse acceleration, a steering wheel corner and a steering lamp switch signal, and sends the signals to a video processor hardware platform through a CAN bus. When the longitudinal acceleration, the transverse acceleration, the steering wheel corner and the steering lamp switch signal of the vehicle CAN be directly analyzed through a vehicle CAN bus communication protocol, the system does not need to be provided with a vehicle-mounted information acquisition device again, and the video processor hardware platform directly analyzes the longitudinal acceleration, the transverse acceleration, the steering wheel corner and the steering lamp switch signal of the vehicle through the CAN bus.

As shown in fig. 2, the vehicle-mounted information acquisition device 3 is provided with 1 vehicle-mounted information acquisition circuit board, and the circuit board mainly comprises a power circuit, a signal conditioning circuit, an information acquisition controller circuit and a CAN bus communication circuit. The power circuit is responsible for supplying power to the vehicle-mounted information acquisition circuit board and the sensor; the signal conditioning circuit is responsible for converting the output signal of the vehicle-mounted information sensor into a voltage signal, filtering, amplifying or reducing the voltage signal to reach the voltage range of the information acquisition controller; the information acquisition controller circuit is responsible for carrying out analog-to-digital conversion on the well-regulated sensor signals, obtaining output values of the sensor signals through calculation, and outputting the output values to a video processor hardware platform through a CAN bus; the CAN bus communication circuit is responsible for CAN bus communication with the video processor hardware platform.

The driver information acquisition camera 1 adopts a CCD vehicle-mounted camera, is arranged on a front windshield of a vehicle and is responsible for acquiring head motion images of a driver.

The video processor hardware platform 2 adopts a high-performance video processor and is responsible for analyzing the head moving images of the driver to obtain the time duty ratios of the head of the driver facing to the area right in front of the vehicle, the left side rearview mirror area and the right side rearview mirror area in a fixed time window. And meanwhile, vehicle operation information and driver head orientation information are integrated, and the lane changing behavior of the driver is predicted based on the lane changing behavior prediction model of the driver. And then designing an automatic steering lamp control algorithm according to the prediction result and the switching condition of the steering lamp.

As shown in fig. 3, 1 video processing circuit board is provided in the video processor hardware platform 2. The video processing circuit board is composed of a power supply circuit, a CAN bus communication circuit, a video decoding circuit, a video processor circuit and an output control circuit. The power supply circuit is responsible for supplying power to the video processing circuit board and the driver information acquisition camera 1; the CAN bus communication circuit is responsible for receiving longitudinal acceleration, transverse acceleration, steering wheel turning angle and steering lamp switch signals of the vehicle, which are sent by the vehicle-mounted information acquisition device 3, through a CAN bus; when the longitudinal acceleration, the transverse acceleration, the steering wheel corner and the turn signal of the turn light of the vehicle CAN be directly analyzed through a CAN bus communication protocol of the vehicle, the video processor directly analyzes the longitudinal acceleration, the transverse acceleration, the steering wheel corner and the turn signal of the turn light of the vehicle through the CAN bus communication circuit; the video decoding circuit is responsible for converting analog image signals of head moving images of the driver, which are acquired by the driver information acquisition camera 1, into digital image signals and sending the digital image signals to the video processor for image processing; the video processor circuit is responsible for executing a driver head motion image processing algorithm, a driver lane changing behavior prediction algorithm and a turn light automatic control algorithm; the output control circuit is responsible for controlling the switch of the turn light.

As shown in fig. 4, a dual-core video processor BF561 of analog devices in the united states of America (ADI) is taken as an example to illustrate the circuit operation principle of the video circuit board of the present invention. The video processing circuit board is composed of a power supply circuit, a CAN bus communication circuit, a video decoding circuit, a video processor circuit and an output control circuit. The power circuit of the video processing circuit board is composed of LM22676, LP38501, LP38693MP and peripheral elements thereof. The LM22676 is responsible for converting the vehicle-mounted 12V power supply into a 5V power supply and supplying power to 5V components of the system. The LP38501 is responsible for converting a vehicle-mounted 5V power supply into a 3.3V power supply and supplying power to 3.3V components of the system. The LP38693MP is responsible for converting a vehicle-mounted 5V power supply into a 1.8V power supply and supplying power to 1.8V components of the system. The CAN bus communication circuit of the video processing circuit board consists of SN65HVD230D and peripheral circuits thereof and is responsible for realizing the physical conversion between the serial communication interface of the video processor and the CAN bus communication interface. The video decoding circuit of the video processing circuit board consists of a video decoding chip ADV7180 and peripheral circuits thereof, and is responsible for decoding an analog video signal, converting the analog video signal into a digital video signal, outputting the digital video signal to a video processor BF561, and processing a video image by the video processor BF 561. The video processor circuit of the video processing circuit board consists of a DDR2 memory MT48LC16M16A2TG, a FLASH memory M29W640D and a video processor BF 561. The DDR2 memory MT48LC16M16A2TG is used to store temporary data when the program is running. FLASH memory M29W640D is used to store the program code for the system. The video processor BF561 is responsible for processing the head moving image information of the driver. Meanwhile, the lane changing behavior of the driver is predicted based on the lane changing behavior prediction model of the driver. And finally, automatically controlling the steering lamp according to the prediction result and the switching condition of the steering lamp.

The driver lane change behavior prediction model is established as follows: the invention establishes a prediction model of the lane changing behavior of a driver, which comprises the following steps: the lane change behavior prediction model (LLC _ HMM), the lane keeping behavior model (LK _ HMM), and the right lane change behavior prediction model (RLC _ HMM).

The driver lane changing behavior prediction model is trained offline: as shown in fig. 5, the off-line training flowchart of the present invention includes the following steps:

(1) and initializing model parameters. The method mainly initializes pi and A, B in the HMM model.

(2) Selecting forward-backward algorithm, and calculating forward frequency alpha by using current sample_t(i) And backward probability beta_t(j)；

(3) Calculating the current new model estimation value by using Baum-Welch algorithm

(4) Calculating likelihood probabilities

(5) If it is notIf the sample is increased progressively, the next estimation is carried out on the sample again by using the new estimation value calculated in the step (3), the step (2) is returned, and iteration is carried out step by step until the sample is subjected to incremental estimationNo longer significantly increased, i.e. converged, at which point the modelThe model is the model sought.

The following describes the training process of the left lane change behavior prediction model according to the present invention, taking the left lane change behavior prediction model (LLC _ HMM) as an example.

(1) Selection of training samples

The observation sequence of the left lane change behavior prediction model selected by the invention comprises the following steps: the vehicle longitudinal acceleration, the vehicle lateral acceleration, the steering wheel angle, the duty ratio of the driver head towards the right front area of the vehicle, the duty ratio of the driver head towards the left side rearview mirror area, and the duty ratio of the driver head towards the left side rearview mirror area are 6 parameters.

The observed sequence of HMMs is described in the form of a vector, as shown in the following equation:

O(t)＝{v₁ v₂ v₃ v₄ v₅ v₆}

wherein v is₁Is the longitudinal acceleration of the vehicle, v₂As lateral acceleration of the vehicle, v₃Is the steering wheel angle, v₄Duty ratio for driver's head facing the area directly in front of vehicle, v₅Duty cycle, v, for the driver's head towards the left side mirror area₆The driver's head towards the left side mirror zone duty cycle.

The number of samples is 100 groups.

(2) Model parameter initialization

The invention adopts a mean value method to obtain the initial values of pi and A.

The invention determines the initial probability distribution of the output probability matrix B according to the prior characteristics of different conversion behaviors.

(3) Left lane changing behavior prediction model

According to the off-line training process shown in fig. 5, the left-turn driving behavior training sample is sent to the initialized left-turn driving behavior prediction model for training, and finally the left-turn driving behavior prediction model is obtained.

The driver lane change behavior prediction process comprises the following steps: the prediction process is shown in fig. 6. And (4) carrying out feature extraction on the original parameters to form a group of observation sequences O. A forward-backward algorithm is applied to calculate the probability P (O/lambda) of each model generating the current observation sequence, and the model with the maximum probability value is the current driving behavior.

The automatic control algorithm of the steering lamp is as follows: as shown in fig. 7, the automatic control algorithm of the turn signal lamp of the present invention is as follows:

(1) information acquisition: the method comprises the following steps of (1) carrying out longitudinal acceleration on a vehicle, transverse acceleration on the vehicle, steering wheel turning angles, turn signal on/off of a steering lamp and head motion images of a driver;

(2) data processing: and analyzing the moving image of the head of the driver to obtain the time duty ratios of the head of the driver to the area right in front of the vehicle, the area of the left side rearview mirror and the area of the right side rearview mirror in the fixed time window.

(3) Predicting lane changing behavior of a driver: predicting the lane changing behavior of the driver according to the offline trained lane changing behavior prediction model of the driver;

(4) automatic control of the steering lamp: designing automatic control of the steering lamp according to the prediction result and the switching condition of the steering lamp;

(a) when the driver has a left lane changing behavior, if the left turn light is not turned on at the moment, the system automatically turns on the left turn light;

(b) when the driver has right lane changing behavior, if the right turn light is not turned on at the moment, the system automatically turns on the right turn light;

(c) when the driver has lane keeping behavior, the system automatically turns off the turn signal if the turn signal is on at this time.

The above-listed detailed description is only a specific description of a possible embodiment of the present invention, and they are not intended to limit the scope of the present invention, and equivalent embodiments or modifications made without departing from the technical spirit of the present invention should be included in the scope of the present invention.