CN118163803A - Road surface type recognition method and vehicle autonomous emergency braking control method - Google Patents

Abstract

The invention relates to the technical field of automobile driving safety, in particular to a pavement type identification method and a vehicle autonomous emergency braking control method and device based on the pavement type identification method, and the method comprises the following steps: acquiring current vehicle state information and current front target state of the vehicle through a vehicle camera and a sensor, and transmitting the acquired information to an automobile domain controller; the method comprises the steps of identifying the type of a road surface in front of a vehicle running through obtaining the reflection intensity characteristics of point cloud by means of a laser radar, so as to obtain the attachment coefficient of the road surface; judging whether potential risks exist or not by carrying out risk assessment on the running state of the automobile in the transverse and longitudinal directions, and carrying out braking measures according to the response of a driver; the type of the road surface on which the vehicle runs is identified in advance, so that the accuracy of estimating the road surface attachment coefficient is improved; the real-time change of the road surface state in front is fully considered, the braking is dynamically adapted and adjusted, the braking of the vehicle can be realized more moderately, and the driving safety of the road and the vehicle is effectively improved.

Description

Road surface type recognition method and vehicle autonomous emergency braking control method

Technical Field

The present invention relates to the technical field of automobile driving safety, and in particular to a road surface type recognition method based on point cloud data and a vehicle autonomous emergency braking control method and device.

Background technique

Automobile driving safety has always been a research focus in the field of vehicles and transportation. In the process of automobile intelligence development, in order to improve the intelligence and safety performance of automobiles, more and more scientific research institutions and automobile manufacturers have increased the perception accuracy of automobiles by adding on-board sensors, and obtained more driving environment information around the car, so as to better realize vehicle control. At present, the vehicle autonomous emergency braking system, as an important function in advanced driver assistance, has reached a bottleneck period. It is mainly reflected in the fact that the adhesion conditions of the road ahead are not detected and the road adhesion coefficient is not considered during the autonomous emergency braking process, resulting in the problem that the vehicle cannot dynamically adapt to the braking. The ultimate goal of studying this problem is to reduce the incidence of traffic accidents by improving the safety, reliability and efficiency of the vehicle autonomous emergency braking system.

The research and defects of road condition detection in the prior art are as follows: (1) Model method is adopted. This method estimates the road adhesion coefficient by establishing a vehicle dynamics model related to the road adhesion coefficient and integrating the state observer design; for example, the tire dynamics response is combined with the μ-s curve to estimate the road adhesion coefficient through the tire dynamics response and the μ-s curve characteristics, wherein the tire and tire mechanical response are estimated by Kalman filtering or real-time detection by sensors; however, this method has the problem of time lag, and a more accurate estimate can only be obtained when the tire is under sufficiently large excitation; at the same time, this method cannot identify the road condition in advance when the road condition ahead suddenly changes, resulting in the inability to accurately estimate the road adhesion coefficient, thereby making it impossible for the autonomous driving car to take countermeasures in advance, which is dangerous. (2) The sensor method uses sensors to obtain data for subsequent processing, thereby further estimating the road adhesion coefficient; for example, an acoustic wave sensor is used, the acoustic wave sensor is installed on the wheel, and the road type is identified by analyzing the noise between the tire and the road surface; among them, the most common method is to use a visual sensor, with the help of a camera, to obtain photos or videos, and then use a machine learning method to manually extract RGB values, textures, grayscale and other features, and use support vector machines or K nearest neighbors and other algorithms to identify the road type; or use a deep learning algorithm to realize the recognition of different types of roads; however, the use of visual sensors is easily affected by lighting conditions. In the case of poor lighting environment, it has a greater impact on the recognition result, and in order to achieve better recognition effect, multiple cameras are usually fused for recognition, which requires greater computing power.

With the increasing research on unmanned driving related equipment, LiDAR sensors have low cost, good perception advantages in environmental perception, high perception accuracy, and are not easily affected by light. They have good application prospects in identifying the road conditions ahead. Therefore, real-time and accurate identification of the road conditions ahead of the vehicle based on LiDAR data is of great significance to AEB control, improving driving comfort and driver safety.

Summary of the invention

To this end, the technical problem to be solved by the present invention is to overcome the problem in the prior art that during the process of autonomous emergency braking, the vehicle cannot identify the road surface condition in advance, resulting in the inability to accurately estimate the road adhesion coefficient, causing the vehicle to be unable to take countermeasures in advance and unable to dynamically adapt to brake adjustments, and at the same time the vehicle driving risk is high; the road surface type cannot be accurately identified due to environmental influences, resulting in low accuracy in estimating the road adhesion coefficient.

In order to solve the above technical problems, the present invention provides a road surface type recognition method based on point cloud data, comprising:

S1: The vehicle-mounted laser radar is connected to the vehicle domain controller through a network cable and transmits data; the vehicle-mounted laser radar obtains point cloud data information around the vehicle and transmits the point cloud data information to the vehicle domain controller;

S2: In the vehicle domain controller, the point cloud data information is preprocessed by the Ransac algorithm; the preprocessed point cloud data information is used as a feature to identify the type of road surface on which the vehicle is currently traveling;

S3: According to the features, using a machine learning or deep learning method, identifying the type of the road surface the vehicle is currently traveling on;

S4: According to the identified road surface type and in combination with the mapping relationship between adhesion coefficients of different road surface types, the adhesion coefficient of the road surface on which the vehicle is currently traveling is obtained.

Preferably, the point cloud data includes time series information of the current vehicle in the front area, three-dimensional coordinate information under laser radar coordinates, and reflection intensity information value.

Preferably, the method using machine learning or deep learning is a CNN deep learning model.

The present invention also provides a vehicle autonomous emergency braking control method, comprising:

Through the vehicle camera and sensors, the vehicle status information and the status of the target in front of the vehicle are obtained, and the obtained information is transmitted to the vehicle domain controller;

The vehicle domain controller calculates the lateral distance between the current vehicle and the target ahead, and determines whether the lateral distance between the current vehicle and the target ahead is less than a preset critical state threshold. If the lateral distance between the current vehicle and the target ahead is less than the preset critical state threshold, it is determined that the current vehicle has a potential collision risk in the lateral direction, otherwise it does not exist;

Based on the safety distance model and the safety time model, a kinematic expression is established, and the kinematic expression is solved to obtain the remaining time of the longitudinal collision of the current vehicle; the road type identification method based on point cloud data described above is used to identify the road type of the current vehicle and obtain the adhesion coefficient of the road type of the current vehicle; the longitudinal safe collision avoidance time is calculated according to the adhesion coefficient of the road type of the current vehicle and the sum of the time from the brake pedal being stepped on to the braking taking effect and the braking force growth time; through the vehicle domain controller, it is determined whether the remaining time of the longitudinal collision is greater than the longitudinal safe collision time. If the remaining time of the longitudinal collision is greater than the longitudinal safe collision time, the current vehicle is in a safe state in the longitudinal direction; otherwise, the current vehicle is in a potential collision risk in the longitudinal direction;

If the current vehicle is in danger of collision in the lateral and/or longitudinal direction, an early warning will be issued through the human-machine interface to remind the driver of the potential collision risk in the corresponding direction; if the driver does not take corresponding countermeasures, the vehicle domain controller will calculate the throttle opening and brake pressure, and based on the adhesion coefficient of the current vehicle driving road type, an emergency braking behavior control signal will be obtained, and the control signal will be transmitted to the vehicle's braking system through the CAN bus communication protocol;

In the braking system, the vehicle's braking control is achieved through the sound and light alarm device, the braking system controller and the throttle controller.

Preferably, the obtaining of the current vehicle state information and the current state of the target in front of the vehicle through the vehicle camera and sensor includes:

The vehicle's own status information is directly or indirectly obtained through the speed sensor, angle sensor, and GPS/IMU sensor; among them, the speed sensor is installed on the vehicle's axle, and the vehicle speed information is obtained through the wheel speed; the angle sensor is installed on the vehicle's steering wheel column, and the vehicle's heading angle information is obtained through the steering wheel angle; the GPS/IMU sensor is installed on the vehicle's front bumper to obtain the vehicle's position information, vehicle acceleration and acceleration;

The current vehicle front target state information is obtained through the camera, and the current vehicle target state information includes: the position and speed of the vehicle, pedestrian, and object in front of the current vehicle; wherein one or more cameras are installed in front of the vehicle and connected to the vehicle domain controller through the GMSL harness; the video information obtained by the camera is transmitted to the vehicle domain controller through the GMSL harness, and the vehicle domain controller detects and identifies the vehicle front target state information through a clustering algorithm or a deep learning algorithm;

Among them, by detecting and identifying the front target, a detection frame will be placed around the front target.

Preferably, judging, by the vehicle domain controller, whether the lateral distance between the current vehicle and the front target is less than a preset critical state threshold comprises:

According to the state information of the target ahead, it can be known that the target ahead is a vehicle that is adjacent to the current vehicle, and the vehicle that is adjacent to the current vehicle is taken as the target vehicle;

Calculate the lateral distance between the current vehicle and the target vehicle, and its expression is:

Among them, S represents the lateral distance between the current vehicle and the target vehicle; R ₁ represents the rated direct distance from the laser radar to the nearest point of the target vehicle detection frame; β ₁ represents the angle between any beam of laser radar in the vertical field of view and the horizontal; α ₁ represents the angle between any beam of laser radar in the horizontal field of view and the front; l represents the width of the current vehicle;

Compare the lateral distance S between the current vehicle and the target vehicle with the preset critical state threshold γ to determine whether the current vehicle is in a potential collision risk in the lateral direction; that is, if S < γ, there is a potential collision risk; if S ≥ γ, there is no potential collision risk.

Preferably, judging, by the vehicle domain controller, whether the remaining longitudinal collision time is greater than the longitudinal safety collision time comprises:

According to the state information of the front target, it can be known that the front target is a moving vehicle next to the current vehicle, and the moving vehicle next to the current vehicle is taken as the target vehicle;

Based on the safety distance model and safety time model, the kinematic expression is established as follows:

Where a _f represents the acceleration of the target vehicle; a represents the acceleration of the current vehicle; v _f represents the speed of the target vehicle; v represents the speed of the current vehicle; TTC represents the remaining time to collision; D represents the relative distance between the current vehicle and the target vehicle;

By solving the kinematic expression, the remaining time TTC of the current vehicle collision is obtained;

Calculate the safe collision avoidance time, the expression is:

Among them, TTA represents the safe collision avoidance time; μ represents the adhesion coefficient of the current vehicle driving road type; δ represents the slope; g represents the acceleration of gravity; v represents the current vehicle speed; T represents the sum of the time from the brake pedal being pressed to the braking effect and the braking force growth time;

Through the vehicle domain controller, the size relationship between the remaining collision time and the safe collision event is compared to determine whether the current vehicle is in potential collision danger in the longitudinal direction; if TTC＞TTA, the current vehicle is in a safe state in the longitudinal direction; if TTC≤TTA, the current vehicle is in a dangerous state in the longitudinal direction.

Preferably, a detector installed at the brake pedal of the vehicle is used to detect whether the driver has taken corresponding countermeasures.

Preferably, the calculation of the valve opening and the brake pressure includes:

Calculate the throttle opening value, the formula is:

Among them, ψ represents the throttle opening; T _exp represents the expected engine torque; ω _ε represents the real-time engine speed, which can be obtained through the sensor installed on the vehicle crankshaft; r represents the tire rolling radius, which can be obtained through the wheel speed sensor installed on the wheel hub; ω _t represents the output speed of the torque converter turbine, which can be obtained through the speed sensor installed near the turbine or pump wheel; T _ε represents the output torque of the engine, which can be obtained through the torque sensor installed on the engine; v represents the current vehicle speed, which can be indirectly obtained through the wheel speed sensor installed on the wheel hub; R _g represents the transmission ratio of the transmission; R _m represents the transmission ratio of the main reducer; the wheel speed can be measured by the wheel speed sensor to deduce the transmission ratio of the transmission and the main reducer;

The formula for calculating the brake pressure is:

Among them, F _bf represents the braking torque of the front wheel; F _br represents the braking torque of the rear wheel; and they can be obtained by a torque sensor installed on the transmission shaft.

The present invention also provides a vehicle autonomous emergency braking control device, comprising:

Environmental perception module: detects the state of the target ahead with the help of the camera, millimeter-wave radar and lidar sensor installed on the car; detects the state of the vehicle itself with the help of speed sensor, angle sensor, GPS/IMU sensor system, etc.; transmits the acquired information to the central processing module;

Central processing module: With the help of the vehicle domain controller, it analyzes and processes the data obtained by the environmental perception module, determines whether the car will have a potential collision risk, detects whether the measures taken by the driver are correct, and determines the countermeasures to be implemented based on the judgment results; transmits the control signal corresponding to the countermeasure to the control execution module;

Control execution module: executes the response of the central processing module and realizes the braking control of the vehicle through the sound and light alarm device, brake system controller and throttle controller.

The above technical solution of the present invention has the following beneficial effects compared with the prior art:

The vehicle autonomous emergency braking control method described in the present invention obtains the reflection intensity characteristics of the point cloud by means of a laser radar to identify the road type ahead of the vehicle, and obtains the road adhesion coefficient according to the road type; it can accurately identify the road type and identify the road type of the vehicle in advance, thereby improving the accuracy of estimating the road adhesion coefficient, and at the same time, taking into account the road friction coefficient during the emergency braking process, so that the vehicle braking process will be relatively gentle;

The vehicle autonomous emergency braking control method described in the present invention takes into account the real-time perception of the road surface state in front of the vehicle in the vehicle emergency braking scenario, focuses on studying the more realistic traffic environment of the vehicle during driving, obtains more detailed and accurate surrounding traffic data with the help of on-board sensors, and takes more comprehensive consideration of the surrounding environmental information; by performing a risk assessment on the vehicle's driving state in the lateral and longitudinal directions, it is determined whether there is a potential risk, and braking measures are taken according to the driver's response, and the real-time changes in the road surface state in front are fully considered. The braking can be dynamically adapted and adjusted, and the vehicle braking can be achieved more gently, which can effectively improve the safety of road driving and the active safety of the vehicle; the vehicle can take countermeasures in advance, reducing the risk of vehicle driving.

BRIEF DESCRIPTION OF THE DRAWINGS

In order to make the content of the present invention more clearly understood, the present invention is further described in detail below according to specific embodiments of the present invention in conjunction with the accompanying drawings, wherein

FIG1 is a flow chart of a method for identifying road surface types based on point cloud data provided by the present invention;

FIG2 is a curve diagram of point cloud intensity values per unit area of different road surface types provided by the present invention;

FIG3 is a confusion matrix diagram of classification results of different road surface types provided by the present invention;

FIG4 is a flow chart of a vehicle autonomous emergency braking control method provided by the present invention;

FIG5 is a schematic diagram of horizontal and vertical state detection provided by the present invention;

FIG. 6 is a schematic diagram of a vehicle autonomous emergency braking control device provided by the present invention.

Explanation of the accompanying symbols in the specification: L represents a laser radar; l represents the width of the car; W represents the width of the lane; D represents the relative distance between the own car and the target vehicle, that is, the longitudinal distance between the two; bbox represents the detection box BoundingBox obtained by target detection; R ₁ and R ₂ represent the direct distance from the laser radar to the nearest point of bbox, α ₁ and α ₂ represent the angle between a certain beam of the laser radar and the front in the horizontal field of view, β ₁ and β ₂ represent the angle between a certain beam of the laser radar and the horizontal in the vertical field of view; S represents the lateral distance between the own car and the target vehicle.

Detailed ways

The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments so that those skilled in the art can better understand the present invention and implement it, but the embodiments are not intended to limit the present invention.

Referring to FIG. 1 , FIG. 1 is a flow chart of a method for identifying road types based on point cloud data provided by the present invention; specifically, the method comprises:

S1: The vehicle-mounted laser radar is connected to the vehicle domain controller through a network cable and transmits data. The vehicle-mounted sensor laser radar obtains point cloud data information around the vehicle. The point cloud data mainly includes the timing information of the current vehicle in the front area, the x, y, z and reflection intensity information values under the laser radar coordinates;

S2: In the vehicle domain controller, the Ransac algorithm is used to pre-process the acquired point cloud data information to achieve the segmentation of the ground point cloud; the method used for data pre-processing includes but is not limited to deep learning methods based on point cloud data, clustering algorithms, etc.; the time series information of the current vehicle in the front area, the three-dimensional coordinate information under the laser radar coordinates, and the reflection intensity information value in the pre-processed point cloud data are used as important features for identifying different road types;

In order to verify whether the point cloud reflection intensity information of different road types can be used as a feature, the application implements the following steps: driving a car equipped with a laser radar on different types of road surfaces, and testing four types of road surfaces: dry asphalt, wet asphalt, dry concrete, and wet concrete during the verification process; the laser radar is connected to the domain controller through a network cable, and the car domain controller is used to obtain the point cloud data detected by the laser radar; the point cloud data of a fixed area in front of the vehicle is selected as the research object, and a 5m*5m area directly in front of the vehicle is selected during the verification process, and all intensity information values of the area are taken, and the average value of the reflection intensity per unit area is used as the measurement indicator;

As shown in FIG2 , the average values of reflection intensity per unit area of different road surface types are all different. Therefore, in the embodiment of the present application, the reflection intensity information in the point cloud data can be used as a classification feature to identify different road surface types.

S3: using a machine learning or deep learning classification method to classify different types of roads according to the features in step S2, so as to accurately identify the road type on which the vehicle is located; wherein, in a specific embodiment of the present invention, a CNN deep learning model is used to classify different types of roads according to the features in step S2;

S4: After the ground types are separated by S3, the adhesion coefficients of different road surfaces are obtained by combining the mapping relationship of the adhesion coefficients of different road surface types; since the adhesion coefficients of different road surface types are within a range, the average value of the range is selected as the adhesion coefficient of the current road surface.

Based on the above embodiment, in this embodiment, the road surface type recognition method based on the reflection intensity in the point cloud data has the following specific steps:

Step 1: Drive a car equipped with a lidar on different types of roads. Here, four types of roads are tested: dry asphalt, wet asphalt, dry concrete, and wet concrete.

Step 2: The laser radar is connected to the domain controller via a network cable, and the car domain controller is used to obtain the point cloud data detected by the laser radar;

Step 3: In this embodiment, the point cloud data of a fixed area in front of the vehicle is selected as the research object. Here, an area of 5m*5m in front of the vehicle is selected;

Step 4: In the selected fixed range, randomly select the reflection intensity values of 1024 points in each frame of the data point cloud as the feature values of the area. Then generate a 32*32 feature matrix based on these 1024 reflection intensity values. Then, for four types of pavement: dry asphalt, wet asphalt, dry concrete, and wet concrete, select 500 frames of data for each type of pavement, and finally obtain 2000 data sets;

Step 5: Divide the above data set into training set, test set and validation set, setting 70% of the data as training set, 15% of the data as test set, and 15% of the data as validation set;

Step 6: For the data set in step 5, send it to the basic CNN network for classification, set the number of iterations to 300, and then iterate. The learning rate is set to 0.0001, the Adam optimizer is used, and the activation function uses the ReLU activation function. The overall structure is mainly: input layer, convolution layer 1, maximum pooling layer, convolution layer 2, maximum pooling layer, fully connected layer, output layer;

Step 7: After iterative training, the classification results of the four types of pavement are summarized through the confusion matrix, and the results are shown in Figure 3. In Figure 3, Class0, Class1, Class2, and Class3 represent dry concrete pavement, wet concrete pavement, dry asphalt pavement, and wet asphalt pavement, respectively. The final result shows that the classification effect is quite significant.

Step 8, after the classification is completed, the adhesion coefficient of the dry concrete pavement is combined with the empirical value through the table lookup method. According to "GA/T643-2006 Technical Appraisal of Vehicle Speed in Typical Traffic Accident Forms", the adhesion coefficient range of the dry concrete pavement is [0.8, 1], the adhesion coefficient of the wet concrete pavement is [0.5, 0.8], the adhesion coefficient of the dry asphalt pavement is [0.6, 0.8], and the adhesion coefficient of the wet asphalt pavement is [0.45, 0.7]. Here, the average value of the value range is selected as the adhesion coefficient. Therefore, the adhesion coefficient of the dry concrete pavement is 0.9, the adhesion coefficient of the wet concrete pavement is 0.65, the adhesion coefficient of the dry asphalt pavement is 0.7, and the adhesion coefficient of the wet asphalt pavement is 0.58. Finally, the adhesion coefficients of different road surface types are determined.

Referring to FIG. 4 , FIG. 4 is a flow chart of a vehicle autonomous emergency braking control method provided by the present invention; specifically comprising:

Step 1: directly or indirectly obtain the vehicle's own state information through the speed sensor, angle sensor, and GPS/IMU sensor; among them, the speed sensor is installed on the axle of the vehicle, and obtains the vehicle speed information through the wheel speed; the angle sensor is installed on the steering column of the vehicle's steering wheel, and obtains the vehicle's heading angle information through the steering wheel angle; the GPS/IMU sensor is installed on the front bumper of the vehicle to obtain the vehicle's position information, the vehicle's acceleration and acceleration;

The current vehicle front target state information is obtained through the camera, and the current vehicle target state information includes: the position and speed of the vehicle, pedestrian, and object in front of the current vehicle; wherein one or more cameras are installed in front of the vehicle and connected to the vehicle domain controller through the GMSL harness; the video information obtained by the camera is transmitted to the vehicle domain controller through the GMSL harness, and the vehicle domain controller detects and identifies the vehicle front target state information through a clustering algorithm or a deep learning algorithm;

Among them, by detecting and identifying the front target, a detection frame will be placed around the front target;

The on-board laser radar is connected to the vehicle domain controller through a network cable and transmits data; the on-board laser radar obtains point cloud data information around the vehicle and transmits the point cloud data information to the vehicle domain controller; the point cloud data information includes time series information, coordinates in the x, y, and z axes under the laser radar coordinates, and reflection intensity information;

Step 2: In the vehicle domain controller, the data is preprocessed by the Ransac algorithm; the coordinates of the x, y, and z axes under the preprocessed laser radar coordinates and the reflection intensity information are used as features for identifying the type of the road surface on which the current vehicle is traveling; based on the features, the CNN deep learning model is used to identify the type of road surface on which the current vehicle is traveling; based on the identified road surface type, the adhesion coefficient of the road surface on which the current vehicle is traveling is obtained by using a table lookup method and combining the mapping relationship of adhesion coefficients of different road surface types;

Referring to FIG. 5 , FIG. 5 is a schematic diagram of the horizontal and vertical state detection provided by the present invention; according to FIG. 5 , it can be seen that there are two targets in front of the current vehicle, and there are detection boxes BoundingBox for the two detected targets; the moving vehicle next to the current vehicle in FIG. 5 is taken as the target vehicle, and the potential collision risk in the horizontal and vertical directions when the current vehicle gradually approaches the target vehicle is considered;

Calculate the lateral distance between the current vehicle and the target vehicle, and its expression is:

Among them, S represents the lateral distance between the current vehicle and the target vehicle; R ₁ represents the rated direct distance from the laser radar to the nearest point of the target vehicle detection frame; β ₁ represents the angle between any beam of laser radar in the vertical field of view and the horizontal; α ₁ represents the angle between any beam of laser radar in the horizontal field of view and the front; l represents the width of the current vehicle;

Compare the distance S between the current vehicle and the target vehicle in the lateral direction with the preset critical state threshold γ to determine whether the current vehicle is in a potential collision risk in the lateral direction; that is, if S < γ, there is a potential collision risk; if S ≥ γ, there is no potential collision risk;

Based on the safety distance model and safety time model, the kinematic expression is established as follows:

Wherein, a _f represents the acceleration of the target vehicle; a represents the acceleration of the current vehicle; v _f represents the speed of the target vehicle; v represents the speed of the current vehicle; TTC represents the remaining time before collision; represents the relative distance between the current vehicle and the target vehicle;

By solving the kinematic expression, the remaining time TTC of the current vehicle collision is obtained;

Calculate the safe collision avoidance time, the expression is:

Among them, TTA represents the safe collision avoidance time; μ represents the adhesion coefficient of the current vehicle driving road type; δ represents the slope; g represents the acceleration of gravity; v represents the current vehicle speed; T represents the sum of the time from the brake pedal being pressed to the braking effect and the braking force growth time;

The vehicle domain controller compares the size relationship between the remaining collision time and the safe collision event to determine whether the current vehicle is in a potential collision risk in the longitudinal direction; if TTC>TTA, the current vehicle is in a safe state in the longitudinal direction; if TTC≤TTA, the current vehicle is in a dangerous state in the longitudinal direction;

According to the judgment result of whether the current vehicle has potential collision danger in the lateral and longitudinal directions, when it is known that there is potential collision danger in the lateral direction, or there is potential collision danger in the longitudinal direction, or there is potential collision danger in both the lateral and longitudinal directions, a warning is issued through the human-machine interaction interface and the vehicle buzzer to remind the driver to follow or change lanes and pay attention to the potential collision danger in the corresponding direction; if the current vehicle does not have potential collision danger in both the lateral and longitudinal directions, the driver keeps the current vehicle driving normally;

If the current vehicle is in a potential collision risk, the detector installed at the brake pedal will detect whether the driver has stepped on the brake pedal; if the driver has stepped on the brake pedal, the detector will detect that the driver has taken corresponding countermeasures; if the driver has not stepped on the brake pedal, the detector will detect that the driver has not taken corresponding countermeasures; the detector will transmit the detection results to the vehicle domain controller;

According to the detection results, the vehicle domain controller knows that the driver has not taken corresponding countermeasures, and then calculates the throttle opening and brake pressure through the vehicle domain controller; the formula for calculating the throttle opening value is:

Among them, ψ represents the throttle opening; T _exp represents the expected engine torque; ω _ε represents the real-time engine speed, which can be obtained through the sensor installed on the vehicle crankshaft; r represents the tire rolling radius, which can be obtained through the wheel speed sensor installed on the wheel hub; ω _t represents the output speed of the torque converter turbine, which can be obtained through the speed sensor installed near the turbine or pump wheel; T _ε represents the output torque of the engine, which can be obtained through the torque sensor installed on the engine; v represents the current vehicle speed, which can be indirectly obtained through the wheel speed sensor installed on the wheel hub; R _g represents the transmission ratio of the transmission; R _m represents the transmission ratio of the main reducer; the wheel speed can be measured by the wheel speed sensor to deduce the transmission ratio of the transmission and the main reducer;

The formula for calculating the brake pressure is:

Wherein, F _bf represents the braking torque of the front wheel; F _br represents the braking torque of the rear wheel; it can be obtained by the torque sensor installed on the transmission shaft;

Based on the adhesion coefficient of the current road surface type on which the vehicle is traveling, an emergency braking behavior control signal is obtained, and the control signal is transmitted to the vehicle's braking system via the CAN bus communication protocol;

Step 3: In the braking system, the vehicle's braking control is achieved through the sound and light alarm device, the braking system controller and the throttle controller.

Referring to FIG. 6 , FIG. 6 is a schematic diagram of a vehicle autonomous emergency braking control device provided by the present invention; specifically comprising:

Environmental perception module: detects the state of the target ahead with the help of the camera, millimeter-wave radar and lidar sensor installed on the car; detects the state of the vehicle itself with the help of speed sensor, angle sensor, GPS/IMU sensor system, etc.; transmits the acquired information to the central processing module;

Central processing module: With the help of the vehicle domain controller, it analyzes and processes the data obtained by the environmental perception module, determines whether the car will have a potential collision risk, detects whether the measures taken by the driver are correct, and determines the countermeasures to be implemented based on the judgment results; transmits the control signal corresponding to the countermeasure to the control execution module;

Control execution module: executes the response of the central processing module and realizes the braking control of the vehicle through the sound and light alarm device, brake system controller and throttle controller.

The device of this embodiment is used to implement the aforementioned vehicle autonomous emergency braking control method. Therefore, the specific implementation method of the vehicle autonomous emergency braking control device can be seen in the embodiment part of the vehicle autonomous emergency braking control method in the previous text. For example, the environmental perception module 100, the central processing module 200, and the control execution module 300 are respectively used to implement steps 1 to 3 in the above-mentioned vehicle autonomous emergency braking control method. Therefore, its specific implementation method can refer to the description of the corresponding various parts of the embodiment, which will not be repeated here.

The present invention discloses an autonomous emergency braking control device for a vehicle. The device studies and optimizes the method from three aspects: an environmental perception module, a central processing module, and a control execution module. By optimizing the three modules, the autonomous emergency braking control of the vehicle in an emergency braking scenario is realized, and the road ahead is detected in real time. Changes in the road ahead status are taken into consideration, laying a solid foundation for achieving L3 and higher levels of autonomous driving.

Obviously, the above embodiments are merely examples for clear explanation and are not intended to limit the implementation methods. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the implementation methods here. The obvious changes or modifications derived from these are still within the protection scope of the invention.

Claims (10)

1. The road surface type identification method based on the point cloud data is characterized by comprising the following steps of:

S1: the vehicle-mounted laser radar is connected with the automobile domain controller through a network cable and transmits data; acquiring point cloud data information around a vehicle through a vehicle-mounted laser radar, and transmitting the point cloud data information to an automobile domain controller;

S2: in an automobile domain controller, preprocessing point cloud data information through Ransac algorithm; taking the preprocessed point cloud data information as a characteristic for identifying the type of the current vehicle driving road surface;

s3: according to the characteristics, identifying the type of the current vehicle driving road surface by adopting a machine learning or deep learning method;

s4: and according to the identified road surface types, combining the mapping relation of the adhesion coefficients of different road surface types to obtain the adhesion coefficient of the current vehicle running road surface.

2. The method for identifying a road surface type based on point cloud data as claimed in claim 1, wherein the point cloud data includes time series information of a current vehicle in a front area, three-dimensional coordinate information under laser radar coordinates, and reflection intensity information values.

3. The method for identifying the road surface type based on the point cloud data according to claim 1, wherein the method adopting machine learning or deep learning is a CNN deep learning model.

4. A method of autonomous emergency brake control of a vehicle, comprising:

acquiring current vehicle state information and current front target state of the vehicle through a vehicle camera and a sensor, and transmitting the acquired information to an automobile domain controller;

Calculating the distance between the current vehicle and the front target in the transverse direction through the automobile domain controller, judging whether the distance between the current vehicle and the front target in the transverse direction is smaller than a preset critical state threshold value, if the distance between the current vehicle and the front target in the transverse direction is smaller than the preset critical state threshold value, judging that the potential collision risk exists in the transverse direction of the current vehicle, otherwise, judging that the potential collision risk does not exist;

Based on the safe distance model and the safe time model, a kinematic expression is established, and the kinematic expression is solved to obtain the current longitudinal collision residual time of the vehicle; identifying the current vehicle driving road surface type by using the road surface type identification method based on the point cloud data according to any one of claims 1-3 to obtain an attachment coefficient of the current vehicle driving road surface type; calculating longitudinal safety collision avoidance time according to the attachment coefficient of the current vehicle running road surface type and the sum of the time from the depression of a brake pedal to the brake effect and the brake force increasing time; judging whether the residual time of the longitudinal collision is longer than the longitudinal safety collision time or not through the automobile domain controller, and if the residual time of the longitudinal collision is longer than the longitudinal safety collision time, enabling the current vehicle to be in a safety state in the longitudinal direction; otherwise, the current vehicle has potential collision danger in the longitudinal direction;

If the current vehicle has collision danger in the transverse direction and/or the longitudinal direction, warning the driver through a human-computer interaction interface to pay attention to the potential collision danger in the corresponding direction; if the driver does not take corresponding countermeasures, calculating the opening degree of an accelerator and the braking pressure through an automobile domain controller, obtaining an emergency braking behavior control signal based on the attachment coefficient of the current vehicle driving road surface type, and transmitting the control signal to a braking system of the vehicle through a CAN bus communication protocol;

in the braking system, the braking control of the vehicle is realized through an audible and visual alarm device, a braking system controller and a throttle valve controller.

5. The method for controlling autonomous emergency braking of a vehicle according to claim 4, wherein the acquiring, by means of the vehicle camera and the sensor, the current vehicle own state information and the current vehicle front target state includes:

The method comprises the steps that the state information of a vehicle is directly or indirectly obtained through a rotation speed sensor, a rotation angle sensor and a GPS/IMU sensor; the rotation speed sensor is arranged at an axle of the vehicle, and obtains vehicle speed information through the rotation speed of wheels; the steering angle sensor is arranged on a steering column of a steering wheel of the vehicle, and the course angle information of the vehicle is obtained through the steering angle of the steering wheel; the GPS/IMU sensor is arranged at the front bumper of the vehicle to acquire vehicle position information, vehicle acceleration and acceleration;

Acquiring current front target state information of a vehicle through a camera, wherein the current target state information of the vehicle comprises: the position and speed of the vehicle, pedestrian and object in front of the current vehicle; one or more cameras are arranged in front of the automobile and connected with the automobile domain controller through gmsl wire harnesses; video information acquired by the camera is transmitted to an automobile domain controller through gmsl wire harnesses, and the automobile domain controller detects and identifies the state information of the target in front of the automobile through a clustering algorithm or a deep learning algorithm;

The front target is detected and identified, and a detection frame is arranged around the front target.

6. The method according to claim 5, wherein determining, by the vehicle domain controller, whether the distance between the current vehicle and the front target in the lateral direction is smaller than a preset critical state threshold value comprises:

According to the state information of the front target, the front target is known to be the running vehicle which is immediately adjacent to the current vehicle, and the running vehicle which is immediately adjacent to the current vehicle is taken as the target vehicle;

Calculating the distance between the current vehicle and the target vehicle in the transverse direction, wherein the expression is as follows

Wherein S represents the distance between the current vehicle and the target vehicle in the transverse direction; r ₁ represents the nominal direct distance from the nearest point of the laser radar to the target vehicle detection frame; beta ₁ represents the included angle between any beam of laser radar in the view angle field and the horizontal in the vertical state; α ₁ represents the angle between any beam of lidar and the right front in the horizontal view field; l represents the width of the current vehicle;

Comparing the magnitude relation between the distance S between the current vehicle and the target vehicle in the transverse direction and a preset critical state threshold gamma, and judging whether the current vehicle has potential collision danger in the transverse direction; if S < gamma, then there is a potential collision risk; if S is larger than or equal to gamma, potential collision danger does not exist.

7. The method of autonomous emergency braking control of a vehicle of claim 5, wherein determining, by the automotive domain controller, whether the longitudinal crash remaining time is greater than the longitudinal safe crash time comprises:

According to the state information of the front target, the front target is known to be the running vehicle which is immediately adjacent to the current vehicle, and the running vehicle which is immediately adjacent to the current vehicle is taken as the target vehicle;

Based on the safe distance model and the safe time model, a kinematic expression is established as

Where a _f represents the acceleration of the target vehicle; a represents the acceleration of the current vehicle; v _f denotes the speed of the target vehicle; v represents the speed of the current vehicle; TTC represents the collision remaining time; representing a relative distance between the current vehicle and the target vehicle;

Obtaining the current vehicle collision remaining time TTC by solving the kinematic expression;

calculating the safe collision avoidance time, wherein the expression is

Wherein TTA represents a safe collision avoidance time; mu represents the adhesion coefficient of the current vehicle driving road surface type; delta represents gradient; g represents gravitational acceleration; v represents the speed of the current vehicle; t represents the sum of the time from when the brake pedal is depressed to when the brake is in effect and the brake force increasing time;

Comparing the magnitude relation between the residual time of collision and the safe collision event through the automobile domain controller, and judging whether the current vehicle has potential collision danger in the longitudinal direction; if TTC is greater than TTA, the current vehicle is in a safe state in the longitudinal direction; if TTC is less than or equal to TTA, the current vehicle is in a dangerous state in the longitudinal direction.

8. The method for autonomous emergency braking control of a vehicle according to claim 4, wherein a detector mounted at a brake pedal of the vehicle is used to detect whether the driver takes corresponding countermeasures.

9. The method of claim 4, wherein calculating the valve opening and the brake pressure comprises:

Calculating the throttle opening value, wherein the formula is

Wherein, psi represents the throttle opening; t _exp represents the desired engine torque; omega _ε represents the real-time engine speed, which can be obtained by a sensor arranged on the crankshaft of the vehicle; r represents the rolling radius of the tire, and can be obtained through a wheel speed sensor arranged on a wheel hub of the wheel; omega _t represents the torque converter turbine output speed, which can be obtained by a speed sensor mounted near the turbine or impeller; t _ε represents the output torque of the engine, which can be obtained by a torque sensor mounted on the engine; v represents the current vehicle speed and can be indirectly obtained through a wheel speed sensor arranged on a wheel hub; r _g represents the gear ratio of the transmission; r _m represents the transmission ratio of the main speed reducer; the rotation speed of the wheels can be measured through a wheel speed sensor, and the transmission ratio of the speed changer and the main speed reducer can be obtained;

The braking pressure is calculated, and the formula is

Wherein F _bf represents the braking torque of the front wheel; f _br represents the braking torque of the rear wheel; can be obtained by a torque sensor arranged on the transmission shaft.

10. An autonomous emergency brake control device for a vehicle, comprising:

An environment sensing module: detecting the front target state by means of a camera, a millimeter wave radar and a laser radar sensor which are arranged on the automobile; detecting the state of the vehicle by means of a rotation speed sensor, a rotation angle sensor, a GPS/IMU sensor system and the like; transmitting the acquired information to a central processing module;

And the central processing module is used for: the method comprises the steps of analyzing and processing data acquired by an environment sensing module by means of an automobile domain controller, judging whether an automobile has potential collision danger or not, detecting whether measures taken by a driver are correct or not, and determining the executed countermeasures according to a judging result; transmitting a control signal corresponding to the countermeasure to a control execution module;

The control execution module: and executing the response of the central processing module, and realizing the braking control of the vehicle through the audible and visual alarm device, the braking system controller and the throttle valve controller.