An automatic driving assistance method based on 3D lidar point cloud

Abstract

The present invention discloses an automatic driving assistance method based on 3D laser radar point cloud, which belongs to the field of driving assistance technology, and specifically comprises: collecting point cloud data near the vehicle through a vehicle-mounted laser radar, identifying other vehicles and their distance D from the vehicle after noise reduction processing; i and speed v i ; Extract the shape, size characteristics and vehicle logo of nearby vehicles to determine the model, and then obtain the fuel consumption value O i ; Use the navigation system to calculate the distance L of each nearby vehicle to the next intersection i ; Calculate the emission impact value P of each vehicle on this vehicle according to the formula i , if P i If the value is greater than the preset threshold value P0, the user is prompted to stay away from the vehicle. i are less than or equal to P0, then the absolute value of the filter emission T i The tallest vehicle is marked T max , calculate the safe distance SD, and prompt the car owner to keep the corresponding distance; the present invention recognizes the vehicle models nearby based on the 3D laser radar point cloud, and protects the health of the people in the car by detecting high-emission vehicles.

Description

Technical Field

[0001] The present invention relates to the field of driving assistance technology, and in particular to an automatic driving assistance method based on 3D laser radar point cloud. Background Art

[0002] With the development of autonomous driving technology, the interaction and coordination between vehicles has become particularly important. In order to improve the safety and efficiency of autonomous driving, it is necessary to accurately perceive and analyze the surrounding environment. Traditional autonomous driving assistance methods mainly rely on information obtained by on-board sensors (such as radar, cameras, etc.), but this information is often affected by factors such as weather and lighting, and it is difficult to accurately identify and classify surrounding vehicles.

[0003] In recent years, 3D LiDAR, as a high-precision distance measurement device, has been widely used in the field of autonomous driving. It can provide rich three-dimensional point cloud data to help autonomous driving devices perceive the surrounding environment more accurately.

[0004] Existing automatic driving assistance methods mainly focus on improving driving safety, ignoring the fact that when vehicles are driving on complex urban roads, they will be affected by the exhaust emissions generated by nearby fuel vehicles, which contain harmful substances such as nitrogen oxides and hydrocarbons. Long-term breathing of these harmful substances will have an adverse effect on the breathing of people in the car. Therefore, there is a need for an automatic driving assistance method that can comprehensively consider multiple factors, while paying attention to the performance and safety of automatic driving, also paying attention to the impact of emissions from nearby vehicles on the health of people in the car. Summary of the invention

[0005] The purpose of the present invention is to provide an automatic driving assistance method based on 3D laser radar point cloud to solve the following technical problems:

[0006] Existing automated driving assistance methods mainly focus on improving driving safety, ignoring the fact that vehicles will be affected by exhaust emissions from nearby fuel vehicles when driving on complex urban roads. Long-term breathing of these harmful substances will have an adverse effect on the breathing of people in the car.

[0007] The purpose of the present invention can be achieved through the following technical solutions:

[0008] An automatic driving assistance method based on 3D laser radar point cloud comprises the following steps:

[0009] The vehicle-mounted laser radar collects the original laser radar point cloud data within the set range near the vehicle, and performs noise reduction processing to obtain the laser radar point cloud data. From the point cloud data, other vehicles near the vehicle and the distance D between each nearby vehicle and the vehicle are identified. i and driving speed vi , and extract the shape features and size features of each nearby vehicle from the point cloud data to obtain the outline shape and size information of each nearby vehicle;

[0010] Selecting the rear area or the front area of ​​any nearby vehicle according to the outline shape, taking an optical image of the rear area or the front area based on a vehicle-mounted camera, extracting a vehicle logo from the optical image, and determining the model of any vehicle according to the vehicle logo, outline shape, and size information;

[0011] Obtain the fuel consumption value of each nearby vehicle according to the vehicle model i , obtain the distance from the vehicle to the next intersection through the navigation system, and calculate the distance L of each nearby vehicle to the next intersection based on the relative position of each nearby vehicle and the vehicle i , calculate the emission impact value P of each vehicle on this vehicle according to the formula i , , where V i represents the relative speed coefficient of each nearby vehicle to the vehicle, ρ represents the density of gasoline, Q represents the default net calorific value of gasoline, F represents the net calorific value emission factor, and when there is any nearby vehicle, the emission impact value P i When it is greater than the preset threshold value P0, the vehicle is prompted to move away from the corresponding vehicle;

[0012] When the emission impact value of nearby vehicles P i When both are less than or equal to the preset threshold value P0, the formula Screening emission absolute value T i The tallest vehicle, marked with a T max , through the formula SD=T max / (P0V i ) calculates the corresponding safety distance SD and prompts the owner to keep a safe distance from the vehicle.

[0013] As a further solution of the present invention: the relative velocity coefficient V i The calculation steps are:

[0014] The direction of the vehicle is marked as forward, and the opposite direction is marked as backward. For the vehicle in front of the vehicle, the speed difference between the vehicle and the vehicle is calculated as △v i =v i -v0, v0 represents the speed of the vehicle. For the vehicle behind the vehicle, the speed difference between the vehicle and the vehicle is calculated as △v i =v0-v i , the calculated speed difference is improved and normalized, and the normalized value is marked as the relative speed coefficient V i , the formula is:

[0015] ;

[0016] where △v min Indicates the minimum value of the speed difference, △v max It represents the maximum value of the speed difference, and α is the optimization coefficient, which is used to make the value of the relative speed coefficient not equal to 0.

[0017] As a further solution of the present invention: for a vehicle traveling side by side with the vehicle, the corresponding speed difference is directly v i The absolute value of the difference from v0.

[0018] As a further solution of the present invention: for a vehicle traveling side by side with the present vehicle, the type of the vehicle is determined only by the outline shape and size information.

[0019] As a further solution of the present invention: the process of denoising the original laser radar point cloud data is as follows:

[0020] The point cloud data is divided into several cubic voxel grids. The leafSize parameter is set to determine the side length (x, y, z) of each voxel grid in three dimensions. The original lidar point cloud data is traversed, and each point is assigned to the corresponding voxel grid according to its coordinate position. Each point is added to the point set of the corresponding voxel grid. In each voxel grid, the coordinate average of all points in the grid is calculated to obtain the coordinate of the center point. The center point is used as the representative point, and the point cloud data is downsampled based on the representative point and output.

[0021] As a further solution of the present invention: when any nearby vehicle is an electric car, the emission impact value of the nearby vehicle is not calculated.

[0022] As a further solution of the present invention: the fuel consumption value is the fuel consumption value per 100 kilometers of each nearby vehicle at the current driving speed.

[0023] As a further solution of the present invention, the method of prompting the vehicle to stay away from the corresponding vehicle or maintain a safe distance is:

[0024] Voice prompts are issued through the car audio, or text prompts are displayed on the car display to inform the driver of the actions that need to be taken.

[0025] Beneficial effects of the present invention:

[0026] The present invention utilizes 3D lidar point cloud data in combination with optical images taken by the on-board camera to more accurately identify and classify the models of surrounding vehicles, obtain the corresponding fuel consumption value based on the model, and calculate the emission impact value of each nearby vehicle. When the emission impact value of any nearby vehicle exceeds the standard, the vehicle will be prompted to stay away from the corresponding vehicle, which helps to avoid the impact of exhaust gas from nearby high-emission vehicles on the bodies of people in the vehicle. For high-emission vehicles that are far away, the corresponding safe distance is calculated through a formula, and the owner is prompted to keep a safe distance from the vehicle, which helps to further protect the health of people in the vehicle. While the existing autonomous driving technology ensures the safety and efficiency of driving, the present invention also pays attention to the impact of exhaust gas emitted by nearby vehicles on the health of people in the car, providing drivers with more comprehensive and personalized driving assistance services. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] The present invention will be further described below in conjunction with the accompanying drawings.

[0028] Figure 1 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0029] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0030] See also Figure 1 As shown, the present invention is an automatic driving assistance method based on 3D laser radar point cloud, comprising the following steps:

[0031] Data collection and processing: First, the vehicle-mounted laser radar collects the original laser radar point cloud data within the set range near the vehicle. These data contain the spatial position information of various objects in the surrounding environment. In order to improve the quality and accuracy of the data, the present invention will perform noise reduction processing on the original point cloud data to remove noise and interference, and obtain clearer and more accurate laser radar point cloud data.

[0032] Vehicle identification and feature extraction: From the processed point cloud data, the present invention can identify other vehicles near the vehicle and calculate the distance D between each nearby vehicle and the vehicle. i And the driving speed v i At the same time, the present invention also extracts the shape and size characteristics of each nearby vehicle to obtain its outline shape and size information. This information is crucial for subsequent vehicle model judgment and emission impact assessment.

[0033] Vehicle model determination: Based on the extracted vehicle outline shape and size information, the present invention will select the rear area or front area of ​​any nearby vehicle and take an optical image of the area based on the vehicle-mounted camera. From the optical image, the present invention can extract key information such as vehicle logo, combine the outline shape and size information, and accurately determine the vehicle model by comparing the features of the vehicle models stored in the database.

[0034] Fuel consumption value and emission impact assessment: Based on the vehicle model information, the present invention can obtain the fuel consumption value of each nearby vehicle at the current driving speed. i At the same time, through the navigation system, the present invention can also obtain the distance from the vehicle to the next intersection. Combining the relative position of each nearby vehicle and the vehicle, the present invention can calculate the distance L from each nearby vehicle to the next intersection. i Using these data, the present invention can calculate the emission impact value P of each vehicle on the vehicle according to a specific formula. i :

[0035] ;

[0036] Among them, V i V represents the relative speed coefficient between each nearby vehicle and the current vehicle. That is, when the distance between the nearby vehicles and the current vehicle is getting farther and farther, V i The closer it is to 1, the closer the distance between the nearby vehicles and the current vehicle is. i The closer it is to 0, ρ represents the density of gasoline, Q represents the default net calorific value of gasoline, and F represents the net calorific value emission factor, which is used to calculate the carbon emission value of the vehicle.

[0037] Safety tips and distance recommendations: When there is any nearby vehicle with an emission impact value P i When the emission impact value P of the nearby vehicle is greater than the preset threshold value P0, the present invention will prompt the vehicle to stay away from the corresponding vehicle to avoid being affected by the high-emission vehicle. i When both are less than or equal to the preset threshold value P0, the present invention will i Filter out the vehicles with the highest emissions and mark them as T max Then, the present invention will be based on the formula SD=T max / (P0V i ) calculates the corresponding safety distance SD and prompts the owner to maintain this safety distance from the vehicle.

[0038] In a preferred embodiment of the present invention, the relative velocity coefficient V i The calculation steps are:

[0039] The driving direction of the vehicle is marked as forward, and the opposite direction is marked as backward. For the vehicle in front of the vehicle, since it is in front of the vehicle, when the speed of the vehicle in front is greater than that of the vehicle, it means that the two vehicles are moving away and the pollution impact is decreasing; when the speed of the vehicle in front is less than that of the vehicle, it means that the two vehicles are approaching and the pollution impact is increasing. Therefore, the speed difference between the vehicle in front and the vehicle is calculated as △v i =v i -v0, a positive number indicates moving away, which can indicate a positive trend, and a negative number indicates approaching, which indicates a negative trend. v0 indicates the speed of the vehicle;

[0040] However, for the vehicle behind the vehicle, the situation is the opposite. When the speed of the rear vehicle is greater than that of the vehicle, it means that the two vehicles are approaching each other and the pollution impact is increasing. When the speed of the rear vehicle is less than that of the vehicle, it means that the two vehicles are moving away and the pollution impact is decreasing. Therefore, the speed difference between the rear vehicle and the vehicle is calculated as △v i =v0-v i , so that positive numbers are used to represent distance, which can represent a positive trend, and negative numbers are used to represent approach, which can represent a negative trend, thereby unifying the meanings of the positive and negative signs of the speed difference. Then, the calculated speed difference is improved and normalized to further eliminate the influence of the order of magnitude on the trend judgment, and the positive and negative numbers are unified into numbers between 0 and 1, which is helpful for the calculation of subsequent data. Then, the normalized value is marked as the relative speed coefficient V i , the formula is:

[0041] ;

[0042] where △v min Indicates the minimum value of the speed difference, △v max represents the maximum value of the speed difference, and △v min and △v max It represents the minimum / maximum value of the speed difference between all nearby vehicles and the vehicle. α is the optimization coefficient, which is used to make the value of the relative speed coefficient not equal to 0, so as to avoid invalidating the value of the emission impact value.

[0043] It is worth noting that for the vehicle traveling side by side with the vehicle, the corresponding speed difference is directly v i The absolute value of the difference from v0.

[0044] In a preferred situation of this embodiment, for a vehicle traveling side by side with the present vehicle, the type of the vehicle is determined only by the outline shape and size information.

[0045] In another preferred embodiment of the present invention, the process of performing noise reduction processing on the original laser radar point cloud data is:

[0046] 1. Create a voxel grid:

[0047] The point cloud data is divided into cubic voxel grids. The size of each grid is specified by the user, usually expressed by the side length. For example, in PCL, the size of the voxel can be determined by setting the leafSize parameter. For example, setting it to (0.5f, 0.5f, 0.5f) means that the side length of the voxel in the x, y, and z directions is 0.5.

[0048] 2. Assign points to grids:

[0049] Traverse the input point cloud data and assign each point to the corresponding voxel grid. Specifically, determine the voxel grid to which the point belongs based on its coordinate position, and add each point to the point set of the corresponding voxel grid.

[0050] 3. Select representative points:

[0051] In each voxel grid, select a representative point. This point is usually the center point of all points in the grid, but it can also be other types of points, such as the centroid, minimum / maximum point, etc. A common way to select the center point is to calculate the average coordinates of all points in the grid to get the coordinates of the center point.

[0052] 4. Output results:

[0053] Output downsampled point cloud data, which only contains points represented by representative points. After voxel filtering, the number of point clouds will be greatly reduced while retaining the shape characteristics of the original point cloud as much as possible.

[0054] The voxel filtering algorithm can effectively reduce the amount of point cloud data and improve the speed of subsequent processing algorithms (such as registration, surface reconstruction, shape recognition, etc.), while retaining the shape characteristics of the original point cloud as much as possible. However, it should be noted that voxel filtering may change the position of the points to a certain extent, and if the voxel grid size is not set properly, the data characteristics may be excessively weakened or details may be lost.

[0055] In another preferred embodiment of the present invention, when the system identifies any nearby vehicle as an electric vehicle, the emission impact value of the vehicle will not be calculated. This is because electric vehicles do not produce tail gas emissions like traditional fuel vehicles, so their impact on the environment is mainly measured by electricity consumption and possible carbon emissions generated by the battery production process, rather than by tail gas emissions during driving. Therefore, when conducting emission impact assessments, electric vehicles can be regarded as low-emission or zero-emission vehicles, and are not included in the immediate emission impact calculation.

[0056] In another preferred embodiment of the present invention, the fuel consumption value refers to the fuel consumption value per 100 kilometers of each nearby vehicle at the current driving speed. This data can usually be obtained through the vehicle's onboard computer system, or estimated based on statistical information in the vehicle model database. The fuel consumption value per 100 kilometers is an important indicator for measuring vehicle fuel efficiency, which directly affects the vehicle's operating cost and environmental impact. Therefore, when calculating the impact of nearby vehicles on the vehicle's emissions, the fuel consumption value is a key parameter.

[0057] In another preferred embodiment of the present invention, in order to provide clear operation guidance to the driver, the present invention provides two prompting methods: one is to issue voice prompts through the vehicle audio system, and the other is to display text prompts through the vehicle display screen. These two methods can be selected or used simultaneously according to the driver's habits and preferences. Voice prompts can convey necessary information without distracting the driver's attention, while text prompts can provide more detailed content to help the driver better understand and perform the recommended operations. Whether it is staying away from high-emission vehicles or keeping a safe distance from specific vehicles, these prompts are intended to improve driving safety while reducing the risk of passengers being exposed to harmful emissions.

[0058] The above is a detailed description of an embodiment of the present invention, but the content is only a preferred embodiment of the present invention and cannot be considered to limit the scope of implementation of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. An automatic driving assistance method based on 3D laser radar point cloud, characterized in that: The following steps are involved: The original laser radar point cloud data within the set range near the vehicle is collected by the on-board laser radar, and noise reduction processing is performed to obtain the laser radar point cloud data. Other vehicles near the vehicle and the distance D between each nearby vehicle and the vehicle are identified from the point cloud data. i and driving speed v i , and extract the shape features and size features of each nearby vehicle from the point cloud data to obtain the outline shape and size information of each nearby vehicle; Selecting the rear area or the front area of ​​any nearby vehicle according to the outline shape, taking an optical image of the rear area or the front area based on a vehicle-mounted camera, extracting a vehicle logo from the optical image, and determining the model of any vehicle according to the vehicle logo, outline shape, and size information; Obtain the fuel consumption value of each nearby vehicle according to the vehicle model i , obtain the distance from the vehicle to the next intersection through the navigation system, and calculate the distance L of each nearby vehicle to the next intersection based on the relative position of each nearby vehicle and the vehicle i , calculate the emission impact value P of each vehicle on this vehicle according to the formula i , , where V i represents the relative speed coefficient of each nearby vehicle to the vehicle, ρ represents the density of gasoline, Q represents the default net calorific value of gasoline, F represents the net calorific value emission factor, and when there is any nearby vehicle, the emission impact value P i When it is greater than the preset threshold value P0, the vehicle is prompted to move away from the corresponding vehicle; When the emission impact value of nearby vehicles P i When both are less than or equal to the preset threshold value P0, the formula Screening emission absolute value T i The tallest vehicle, marked with a T max , through the formula SD=T max / (P0V i ) calculates the corresponding safety distance SD and prompts the owner to keep a safe distance from the vehicle.

2. The automatic driving assistance method based on 3D laser radar point cloud according to claim 1, characterized in that: The relative velocity coefficient V i The calculation steps are: The direction of the vehicle is marked as forward, and the opposite direction is marked as backward. For the vehicle in front of the vehicle, the speed difference between the vehicle and the vehicle is calculated as △v i =v i -v0, v0 represents the speed of the vehicle. For the vehicle behind the vehicle, the speed difference between the vehicle and the vehicle is calculated as △v i =v0-v i , the calculated speed difference is improved and normalized, and the normalized value is marked as the relative speed coefficient V i , the formula is: ； where △v min Indicates the minimum value of the speed difference, △v max It represents the maximum value of the speed difference, and α is the optimization coefficient, which is used to make the value of the relative speed coefficient not equal to 0.

3. The automatic driving assistance method based on 3D laser radar point cloud according to claim 2, characterized in that: For the vehicle traveling side by side with this vehicle, the corresponding speed difference is directly v i The absolute value of the difference from v0.

4. The automatic driving assistance method based on 3D laser radar point cloud according to claim 2, characterized in that: For a vehicle traveling side by side with the vehicle, the model of the vehicle is determined only by the outline shape and size information.

5. The automatic driving assistance method based on 3D laser radar point cloud according to claim 1, characterized in that: The process of denoising the original laser radar point cloud data is as follows: The point cloud data is divided into several cubic voxel grids. The leafSize parameter is set to determine the side length (x, y, z) of each voxel grid in three dimensions. The original lidar point cloud data is traversed, and each point is assigned to the corresponding voxel grid according to its coordinate position. Each point is added to the point set of the corresponding voxel grid. In each voxel grid, the coordinate average of all points in the grid is calculated to obtain the coordinate of the center point. The center point is used as the representative point, and the point cloud data is downsampled based on the representative point and output.

6. The automatic driving assistance method based on 3D laser radar point cloud according to claim 1, characterized in that: When any nearby vehicle is a tram, the emission impact value of the nearby vehicle will not be calculated.

7. The automatic driving assistance method based on 3D laser radar point cloud according to claim 1, characterized in that: The fuel consumption value is the fuel consumption value per 100 kilometers of each nearby vehicle at the current driving speed.

8. The automatic driving assistance method based on 3D laser radar point cloud according to claim 1, characterized in that: The method to prompt the vehicle to stay away from the corresponding vehicle or keep a safe distance is: Voice prompts are issued through the car audio, or text prompts are displayed on the car display to inform the driver of the actions that need to be taken.

Images