Roadway drivable area identification method and system based on multi-sensor fusion

Abstract

The invention relates to a roadway drivable area identification method and system based on multi-sensor fusion, and belongs to the field of automatic driving of engineering machinery. The roadway drivable area identification method based on multi-sensor fusion comprises the following steps: acquiring environmental point cloud data in front of a vehicle head at the current moment by using a laser radar; acquiring synchronous data of an IMU (inertial measurement unit) and the laser radar, and performing motion compensation on the environmental point cloud data acquired by the laser radar at the previous moment through the IMU to acquire environmental point cloud data at the vehicle driving moment; processing the environment point cloud data at the vehicle driving moment to obtain non-ground point cloud data; and using an LOAM algorithm to extract boundary points in the non-ground point cloud data, and obtaining a roadway center line for the vehicle to move forward. The drivable area of the vehicle under the roadway can be intelligently identified through detection of the laser radar and the IMU, the detection and identification precision is high, and more efficient and intelligent construction conditions can be provided for operation under the roadway.

Description

Technical Field

[0001] The present application relates to the field of automatic driving of construction machinery, and in particular to a method and system for identifying drivable areas in lanes based on multi-sensor fusion. Background Art

[0002] Drivable area recognition is the most advanced perception module of intelligent driving. It can detect the entire road surface or obtain partial road information, providing path planning assistance for intelligent driving.

[0003] Currently, safe autonomous driving requires detecting surrounding objects and identifying drivable areas. Current autonomous vehicles typically use a combination of radar, lidar, cameras, high-precision GPS, and prior map information to determine obstacles around the vehicle and safe driving areas on the road.

[0004] In the process of identifying the drivable area, the unmanned driving situation in the alley is complicated, there is no satellite communication signal coverage, and there are multiple problems such as insufficient light, small space, and high humidity. As a result, the radar detection distance is short, and the camera is easily affected by environmental conditions such as light, affecting the accuracy of the recognition process. Summary of the invention

[0005] The embodiments of the present application provide a method and system for identifying drivable areas in alleys based on multi-sensor fusion, so as to solve the problem in the related art that in unmanned driving in alleys, the radar detection distance is short and the camera is easily affected by environmental conditions such as lighting, resulting in low accuracy of the recognition process.

[0006] First aspect

[0007] The present application provides a method for identifying a drivable area in a lane based on multi-sensor fusion, comprising the following steps:

[0008] Use laser radar to collect point cloud data of the environment in front of the vehicle at the current moment;

[0009] Acquire the synchronization data of the IMU and the laser radar, and perform motion compensation on the environmental point cloud data acquired by the laser radar at the previous moment through the IMU to acquire the environmental point cloud data at the time of vehicle driving;

[0010] Processing the environmental point cloud data at the time of vehicle driving to obtain non-ground point cloud data;

[0011] The LOAM algorithm is used to extract boundary points in the non-ground point cloud data to obtain the center line of the lane for the vehicle to travel forward.

[0012] In some embodiments, the laser radar is used to collect point cloud data of the environment in front of the vehicle at the current moment, specifically including:

[0013] Arrange two laser radars on the front of the vehicle so that the two laser radars are symmetrical about the X-axis in the vehicle coordinate system;

[0014] The laser radar is used to collect point cloud data of the environment at the current moment.

[0015] In some embodiments, the emission axis of the laser radar is at an angle of 30° with the X-axis.

[0016] In some embodiments, the synchronization data of the IMU and the laser radar is obtained, and the environmental point cloud data obtained by the laser radar at the previous moment is motion compensated by the IMU to obtain the environmental point cloud data at the time of vehicle driving, specifically including:

[0017] Arrange at least one IMU on the front of the vehicle, and synchronize the IMU data with the laser radar data;

[0018] Obtaining predicted values ​​of environmental point cloud data through the IMU motion equation;

[0019] The Kalman filter is used to process the motion compensation of the laser radar on the predicted value of the environmental point cloud data obtained by the IMU.

[0020] In some embodiments, the Kalman filter processes the motion compensation of the laser radar on the predicted value of the environmental point cloud data acquired by the IMU, specifically including:

[0021] The predicted value is used to remove the distortion of the environmental point cloud data at the previous moment, and the environmental point cloud data at the last moment of vehicle driving is obtained;

[0022] The environmental point cloud data obtained at the time of vehicle driving is updated to the Kalman filter to obtain the posterior pose.

[0023] In some embodiments, the environmental point cloud data at the time of vehicle driving is processed to obtain non-ground point cloud data, specifically including:

[0024] Using a ground estimation algorithm, the environment point cloud data is divided to obtain non-ground point cloud data and ground point cloud data;

[0025] The ground point cloud data is eliminated to obtain non-ground point cloud data.

[0026] In some embodiments, before dividing the environmental point cloud data, the method further includes:

[0027] The environmental point cloud data is preprocessed, wherein the preprocessing includes at least one of denoising and filtering.

[0028] In some embodiments, the LOAM algorithm is used to extract boundary points in the non-ground point cloud data to obtain the center line of the lane for the vehicle to move forward, specifically including:

[0029] A point cloud feature extraction algorithm is used to extract features from the non-ground point cloud data to obtain feature data, wherein the feature data includes a curvature of a wall surface projected on a horizontal plane and a geometric shape of an attached component on the wall surface projected on a horizontal plane;

[0030] The feature tracking algorithm performs feature tracking on the non-ground point cloud data to obtain visible feature data between consecutive frames.

[0031] In some embodiments, a laser radar is arranged at the rear of the vehicle.

[0032] Second aspect

[0033] The embodiment of the present application provides a lane drivable area identification system based on multi-sensor fusion, which is characterized by comprising:

[0034] LiDAR, which is used to collect point cloud data of the environment in front of the vehicle;

[0035] A motion compensation module, which is used to perform motion compensation on the environmental point cloud data acquired by the laser radar to obtain environmental point cloud data at the time of vehicle driving;

[0036] A data processing module, which is used to process the non-ground point cloud data to obtain non-ground point cloud data;

[0037] The centerline acquisition module uses the LOAM algorithm to extract boundary points in the non-ground point cloud data, obtain the centerline of the lane for the vehicle to move forward, and obtain the centerline of the lane for the vehicle to move forward.

[0038] The beneficial effects of the technical solution provided by this application include:

[0039] The present application provides a method and system for identifying a drivable area in a lane based on multi-sensor fusion:

[0040] 1. A method for identifying a drivable area in a lane based on multi-sensor fusion is characterized by comprising the following steps: using a laser radar to collect environmental point cloud data in front of the vehicle at the current moment; obtaining synchronous data between the IMU and the laser radar, and using the IMU to perform motion compensation on the environmental point cloud data obtained by the laser radar at the previous moment to obtain environmental point cloud data at the time of vehicle driving; processing the environmental point cloud data at the time of vehicle driving to obtain non-ground point cloud data; using the LOAM algorithm to extract boundary points in the non-ground point cloud data to obtain the center line of the lane for the vehicle to move forward. Therefore, when the lane is unmanned, the drivable area and center line of the lane can be identified and extracted through the cooperation detection of the laser radar and the IMU, and the advanced data processing algorithm can be used to quickly locate and provide a high-precision route for the vehicle to avoid obstacles under the lane, and efficiently complete the construction in the lane, which is intelligent construction, high practical value, and conducive to promotion.

[0041] 2. Since the laser radar is used to collect the environmental point cloud data in front of the front of the vehicle at the current moment, it specifically includes: arranging two laser radars on the front of the vehicle so that the two laser radars are symmetrical about the X-axis in the vehicle coordinate system; using the laser radar to collect the environmental point cloud data at the current moment, so when the laser radar is arranged on the front of the vehicle to collect the environmental data in front of the front of the vehicle, two laser radars are arranged on the front of the vehicle so that the two laser radars are symmetrical about the X-axis in the vehicle coordinate system; using the laser radar to collect the environmental point cloud data at the current moment, the symmetrical arrangement of the two laser radars helps to reduce blind spots and improve the comprehensiveness and accuracy of environmental perception. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0043] Figure 1 A schematic diagram of the laser radar arrangement state provided in an embodiment of the present application;

[0044] Figure 2 A schematic diagram showing a method for identifying a drivable area in a lane provided in an embodiment of the present application;

[0045] Figure 3 A schematic diagram of a point cloud data motion compensation process provided in an embodiment of the present application;

[0046] Reference numerals:

[0047] 1. Vehicle; 2. LiDAR; 3. IMU. DETAILED DESCRIPTION

[0048] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0049] The embodiments of the present application provide a method and system for identifying drivable areas in alleys based on multi-sensor fusion, which can solve the problem of low accuracy of the identification process caused by short radar detection distance and camera being easily affected by environmental conditions such as lighting during unmanned driving in alleys.

[0050] Example 1

[0051] See also Figures 1 to 3 As shown, an embodiment of the present application provides a method for identifying a drivable area in a lane based on multi-sensor fusion, comprising the following steps: using a laser radar 2 to collect environmental point cloud data in front of the front of the vehicle 1 at the current moment; obtaining synchronous data of the IMU3 and the laser radar 2, and performing motion compensation on the environmental point cloud data obtained by the laser radar 2 at the previous moment through the IMU3 to obtain the environmental point cloud data at the time when the vehicle 1 is driving; processing the environmental point cloud data at the time when the vehicle 1 is driving to obtain non-ground point cloud data; using the LOAM algorithm to extract boundary points in the non-ground point cloud data to obtain the center line of the lane for the vehicle 1 to move forward.

[0052] When the above method is used, specifically through the laser radar 2 emitting a laser beam and receiving the reflected signal, it can accurately measure the three-dimensional position information of objects in the surrounding environment and generate point cloud data. The laser radar 2 collects the environmental data in front of the vehicle in real time, which includes the ground information of the roadway and non-ground information. MU can measure the acceleration, angular velocity and direction change of vehicle 1 in real time. By synchronizing the data of IMU3 with the laser radar 2, the information provided by IMU3 can be used to perform motion compensation on the laser radar 2 data. This is because during the data collection process of the laser radar 2, the vehicle 1 itself may be moving, resulting in dynamic distortion of the point cloud data.

[0053] The purpose of motion compensation is to convert the point cloud data of LiDAR 2 into the reference frame when vehicle 1 is stationary, so as to more accurately reflect the state of the environment. After acquiring the motion-compensated environmental point cloud data, it needs to be processed to remove the ground point cloud data and retain only the non-ground point cloud data. Finally, since the LOAM algorithm is an efficient LiDAR 2 odometer and mapping method, it can simultaneously estimate the motion of vehicle 1 and build a three-dimensional map of the environment. After identifying the non-ground point cloud data, the LOAM algorithm is used to further process these data, especially to extract the boundary points of the lane, which represent the physical limitations of the lane, such as walls, obstacles on the walls, etc. By connecting and analyzing these boundary points, the outline of the lane can be identified. Further, the center line of the lane, that is, the driving path that vehicle 1 should follow, can be calculated using this outline information.

[0054] By combining the data from the LiDAR 2 and IMU 3, as well as advanced data processing and algorithms, the present application can accurately identify the drivable area and center line of the lane, and provide reliable navigation information for the autonomous or unmanned vehicle 1.

[0055] In this application, laser radar 2 is used to collect environmental point cloud data in front of the front of vehicle 1 at the current moment. In the fields of autonomous driving and robot navigation, laser radar 2 is an important sensor that can obtain distance information by emitting laser beams and measuring the time for these laser beams to be reflected back from objects in the surrounding environment, thereby constructing a 3D point cloud map of the surrounding environment. When laser radar 2 is arranged at the front of vehicle 1 to collect environmental data in front of the front of the vehicle, the specific implementation method and configuration are as follows: two laser radars 2 are arranged on the front of vehicle 1 so that the two laser radars 2 are symmetrical about the X-axis in the coordinate system of vehicle 1; laser radar 2 is used to collect environmental point cloud data at the current moment. The coordinate system of vehicle 1 is usually defined as: the X-axis is along the forward direction of vehicle 1, the Y-axis points to the left side of vehicle 1, and the Z-axis is vertically upward. The symmetrical arrangement of two laser radars 2 helps to reduce blind spots and improve the comprehensiveness and accuracy of environmental perception.

[0056] The laser radar 2 is placed at the front of the vehicle 1, as high as possible and away from obstructions such as the hood and front bumper, to ensure that the widest and clearest possible field of view can be obtained. The emission axis of the laser radar 2 is at an angle of 30° to the X-axis, which means that the laser beam not only directly illuminates the area directly in front of the vehicle 1, but also slightly deviates to one side, so that it can cover part of the area on both sides of the front of the vehicle. This design helps to increase the detection capability of the blind spots on both sides of the front of the vehicle.

[0057] Since the laser radar 2 calculates the distance and position of an object by emitting a laser beam and receiving reflected light, these data are initially acquired based on the laser radar 2's own coordinate system. However, in an autonomous driving system, a unified coordinate system is usually required to integrate data from different sensors. This application also requires that the coordinates of the two front laser radars 2 be converted to the vehicle body coordinate system, which can ensure that all sensor data are in the same coordinate frame, facilitating subsequent data processing and analysis.

[0058] In this application, the synchronous data of IMU3 and LiDAR 2 is obtained, and the environmental point cloud data obtained by LiDAR 2 at the previous moment is motion compensated by IMU3 to obtain the environmental point cloud data at the time of vehicle 1 driving. In autonomous driving and robot systems, the synchronous data acquisition and fusion of IMU3 inertial measurement unit and LiDAR 2 is a key technology, which can help the system understand and adapt to the motion state of vehicle 1 or robot more accurately, thereby generating more accurate environmental point cloud data.

[0059] Therefore, by placing at least one IMU3 on the front of vehicle 1, the IMU3 should be able to accurately measure the inertial parameters such as acceleration and angular velocity of vehicle 1. After that, it is necessary to ensure that the IMU3 data is synchronized with the LiDAR 2 data. The synchronization of IMU3 data and LiDAR 2 data is synchronized in time, which is usually achieved through hardware synchronization signals such as trigger signals or software timestamp matching. Time synchronization is the key to ensure that the LiDAR 2 and IMU3 data can be accurately fused. Then the predicted value of the environmental point cloud data is obtained through the IMU3 motion equation; the IMU3 motion equation is based on Newton's second law, and it is mainly used to describe and calculate the motion state of an object such as posture, velocity and position.

[0060] IMU3 mainly includes gyroscopes and accelerometers, and the motion equations are constructed through the measurements of these two sensors. The motion information of vehicle 1 provided by IMU3, including but not limited to position, speed, and attitude, can be used to predict the distance and direction that vehicle 1 may move within the data collection interval of LiDAR 2. Then, based on this predicted information, the point cloud data of LiDAR 2 at the previous moment is transformed accordingly to predict the position of the environmental point cloud at the current moment.

[0061] In addition, a Kalman filter is used to process the motion compensation of the laser radar 2 for the predicted value of the environmental point cloud data obtained by IMU3. The Kalman filter is an efficient recursive filter that can estimate the state of a dynamic system from a series of measurements containing statistical noise. In the scenario where IMU3 and laser radar 2 are fused, the Kalman filter is used to combine the predicted value of IMU3 and the observed value of laser radar 2 to obtain the optimal state estimate. The predicted value of IMU3 is used as the predicted state of the Kalman filter, and the observed value of laser radar 2, that is, the directly acquired environmental point cloud data, is used as the observation input. Based on these inputs, the Kalman filter continuously iterates and optimizes the state estimate through two steps of prediction and update. During the update process of the Kalman filter, the difference between the predicted value of IMU3 and the observed value of laser radar 2 is considered, and the state estimate is adjusted accordingly. This adjustment process is actually the process of motion compensation for the laser radar 2 data, which enables the point cloud data to more accurately reflect the environmental state of the vehicle 1 at the time of driving, and improves the accuracy and real-time performance of the environment perception of the autonomous driving and robot system.

[0062] In this application, since the vehicle 1 is moving during the scanning process of the laser radar 2, the original environmental point cloud data will contain distortion caused by the movement. The Kalman filter processes the motion compensation of the laser radar 2 for the predicted value of the environmental point cloud data obtained by the IMU3, and uses the predicted value to remove the distortion of the environmental point cloud data at the previous moment. By converting each laser point to the coordinate system of a certain moment predicted by the IMU3, this distortion can be removed to obtain the real environmental point cloud data. After that, the environmental point cloud data obtained at the time of vehicle 1 driving is updated to the Kalman filter to obtain the posterior pose. Through the update step of the Kalman filter, the observation data of the laser radar 2 is fused into the prediction of the IMU3 to obtain the optimal estimated pose at the current moment, that is, the position and posture of the vehicle 1. This posterior pose not only contains the prediction information of the IMU3, but also integrates the observation information of the laser radar 2, so it is more accurate and reliable.

[0063] In the present application, the environmental point cloud data at the time when the vehicle 1 is driving is processed to obtain non-ground point cloud data, specifically including: using a ground estimation algorithm to divide the environmental point cloud data to obtain non-ground point cloud data and ground point cloud data; the ground estimation algorithm is used to identify and separate the ground point cloud from the entire environmental point cloud data. This step is the key to obtaining non-ground point cloud data. The ground point cloud data is eliminated to obtain non-ground point cloud data. Once the ground point cloud is identified by the ground estimation algorithm, these points can be eliminated from the original point cloud data to obtain non-ground point cloud data. The elimination process is relatively simple, and usually only the points identified as the ground need to be removed from the data set.

[0064] In the present application, before dividing the environmental point cloud data, the method also includes: preprocessing the environmental point cloud data, and the preprocessing includes at least one of denoising and filtering. It is very important to preprocess the environmental point cloud data before any further processing. The main purpose of preprocessing is to improve the quality of point cloud data, reduce noise and unnecessary interference, so that subsequent processing can be carried out more accurately. Preprocessing generally includes the following steps: Denoising: removing noise points in the point cloud. These noise points may be caused by sensor errors, environmental interference or data transmission problems. Denoising can be achieved by statistical filtering, radius filtering and other methods, which can identify and remove isolated points that are significantly different from the surrounding point clouds. Filtering: Filtering is the process of smoothing point cloud data to reduce random errors and noise in the data. Common filtering methods include Gaussian filtering, bilateral filtering, etc. The filtered point cloud data is smoother, which is conducive to subsequent ground estimation and other processing steps.

[0065] In the present application, the LOAM algorithm is used to extract boundary points in non-ground point cloud data and obtain the center line of the lane for vehicle 1 to move forward, specifically including: a point cloud feature extraction algorithm, which performs feature extraction on non-ground point cloud data to obtain feature data, and the feature data includes the curvature of the wall projected on the horizontal plane and the geometric shape of the attached components on the wall projected on the horizontal plane; a feature tracking algorithm, which performs feature tracking on the non-ground point cloud data to obtain visible feature data between consecutive frames.

[0066] In the LOAM algorithm, point cloud feature extraction is the first step, which aims to identify significant structural features from non-ground point cloud data, which are usually very useful for positioning and mapping. For extracting boundary points to generate the centerline of the lane, we need to pay special attention to the characteristics of the wall. First, the curvature can be estimated by calculating the normal or surface fitting of each point in the point cloud in its local neighborhood. The projection of the wall on the horizontal plane usually appears as a low curvature area, while the corner or boundary of the wall appears as a point where the curvature suddenly changes. By identifying these curvature mutation points, we can preliminarily locate the boundary of the wall. The projection of the attached components that may exist on the wall on the horizontal plane will present a specific geometric shape. By identifying these geometric shapes, we can further verify and refine the identification of the wall boundary. Feature tracking is a key step in connecting consecutive frames of point cloud data. It allows us to track the position changes of the same feature at different time points, thereby estimating the movement of the sensor.

[0067] After obtaining the feature tracking results between consecutive frames, we can use these features to extract the center line of the lane. First, fit the boundary line of the lane according to the boundary points of the wall. This can be achieved through straight line fitting, curve fitting, or more complex geometric models. With the boundary line, the center line of the lane can be obtained by calculating the midpoint between the boundary lines or the least squares midline. If the shape of the lane is complex, a more complex algorithm may be required to ensure the smoothness and accuracy of the center line. Finally, based on the extracted center line, path planning can be performed to provide guidance for vehicle 1 to move forward.

[0068] In this application, when the wall surface monitored by the two front laser radars 2 is relatively smooth, the perception ability of the two front laser radars 2 will decrease, while the laser radar 2 is not affected by light and can stably provide three-dimensional information of the surrounding environment of the vehicle 1. The addition of the rear laser radar 2 enables the vehicle 1 to maintain a high perception ability, realize the all-round perception of the surrounding environment of the vehicle 1, and help improve the driving efficiency and safety of the vehicle 1.

[0069] The implementation principle of a method for identifying a drivable area in a lane based on multi-sensor fusion disclosed in an embodiment of the present application is as follows: when an unmanned vehicle 1 is operating in a lane, a laser radar 2 is used to collect environmental point cloud data in front of the front of the vehicle 1 at the current moment. The laser radar 2 generates point cloud data in three-dimensional space by emitting a laser beam and receiving the reflected signal. These data contain information such as the position and shape of objects in the environment. The IMU 3 can measure the motion state information such as the acceleration and angular velocity of the vehicle 1 in real time. The environmental point cloud data acquired by the laser radar 2 at the previous moment is motion compensated by the IMU 3. Since the vehicle 1 is in motion, the point cloud data collected by the laser radar 2 will have motion distortion, that is, each point in the point cloud data does not completely represent the surrounding environment of the vehicle 1 at the same moment. Therefore, the point cloud data is corrected using the motion information provided by the IMU 3 to obtain the real environmental point cloud data at the time when the vehicle 1 is driving. After obtaining the real environmental point cloud data, the ground point cloud data is removed and the non-ground point cloud data is retained; finally, the non-ground point cloud data is processed using the LOAM algorithm to extract the boundary points therein. The LOAM algorithm is a point cloud registration and map construction algorithm based on LiDAR 2. It can estimate the motion trajectory of vehicle 1 and construct an environmental map by extracting and matching feature points. Boundary points usually represent the edges or obstacles on both sides of the lane. By extracting these boundary points, the center line of the lane can be further calculated. The extraction of the center line is usually based on the geometric relationship of the boundary points, such as fitting the boundary points by the least squares method to obtain the central axis of the lane. Through the fusion of LiDAR 2 and IMU 3, and the application of motion compensation technology, the perception accuracy of the driving environment of vehicle 1 can be significantly improved. This method can cope with complex and changeable lane environments, such as lighting changes, shadows, water accumulation, etc., and provides strong support for the development of autonomous driving technology.

[0070] Example 2

[0071] Reference Figure 1-Figure 3 The embodiment of the present application provides a lane drivable area recognition system based on multi-sensor fusion, including: laser radar 2, motion compensation module, data processing module and centerline acquisition module, laser radar 2 is used to collect environmental point cloud data in front of the vehicle 1; motion compensation module, which is used to perform motion compensation on the environmental point cloud data acquired by laser radar 2 to obtain environmental point cloud data at the time of vehicle 1 driving; data processing module, which is used to process non-ground point cloud data to obtain non-ground point cloud data; centerline acquisition module, which uses LOAM algorithm to extract boundary points in non-ground point cloud data, obtain the centerline of the lane for vehicle 1 to move forward, and obtain the centerline of the lane for vehicle 1 to move forward. When used, it realizes accurate perception of the driving environment of vehicle 1 and effective extraction of the centerline of the lane, providing strong support for the development of autonomous driving technology.

[0072] In the description of the present application, it should be noted that the terms "upper", "lower", etc. indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present application and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present application. Unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be an indirect connection through an intermediate medium, or it can be a connection between the two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present application can be understood according to the specific circumstances.

[0073] It should be noted that, in this application, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0074] The above description is only a specific implementation of the present application, so that those skilled in the art can understand or implement the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest range consistent with the principles and novel features applied for herein.

Claims

1. A method for identifying drivable lane areas based on multi-sensor fusion, characterized in that: The following steps are involved: Using a laser radar (2), collecting point cloud data of the environment in front of the front of the vehicle (1) at the current moment; Acquire synchronous data between the IMU (3) and the laser radar (2), and perform motion compensation on the environmental point cloud data acquired by the laser radar (2) at the previous moment through the IMU (3) to acquire the environmental point cloud data at the time when the vehicle (1) is traveling; Processing the environmental point cloud data of the vehicle (1) at the time of driving to obtain non-ground point cloud data; The LOAM algorithm is used to extract boundary points in the non-ground point cloud data to obtain the center line of the lane for the vehicle (1) to move forward.

2. The method for identifying a drivable lane area based on multi-sensor fusion according to claim 1, characterized in that: The laser radar (2) is used to collect point cloud data of the environment in front of the front of the vehicle (1) at the current moment, specifically including: Arranging two laser radars (2) on the front of the vehicle (1) so that the two laser radars (2) are symmetrical about the X-axis in the coordinate system of the vehicle (1); The laser radar (2) is used to collect point cloud data of the environment at the current moment.

3. The method for identifying a drivable lane area based on multi-sensor fusion according to claim 2, characterized in that: The emission axis of the laser radar (2) forms an angle of 30° with the X-axis.

4. The method for identifying a drivable lane area based on multi-sensor fusion according to claim 1, characterized in that: Acquiring synchronous data between the IMU (3) and the laser radar (2), and performing motion compensation on the environmental point cloud data acquired by the laser radar (2) at the previous moment through the IMU (3) to acquire the environmental point cloud data at the time when the vehicle (1) is driving, specifically comprising: At least one IMU (3) is arranged on the front of the vehicle (1), and data from the IMU (3) is synchronized with data from the laser radar (2); Obtaining predicted values ​​of environmental point cloud data through the IMU (3) motion equation; A Kalman filter is used to process the motion compensation of the laser radar (2) on the predicted value of the environmental point cloud data obtained by the IMU (3).

5. The method for identifying a drivable lane area based on multi-sensor fusion according to claim 4, characterized in that: The Kalman filter processes the motion compensation of the laser radar (2) on the predicted value of the environmental point cloud data acquired by the IMU (3), specifically comprising: The predicted value is used to remove the distortion of the environmental point cloud data at the previous moment, and the environmental point cloud data at the last moment of vehicle (1) driving is obtained; The environmental point cloud data acquired at the time when the vehicle (1) is traveling is updated to the Kalman filter to obtain the posterior pose.

6. The method for identifying a drivable lane area based on multi-sensor fusion according to claim 1, characterized in that: Processing the environmental point cloud data of the vehicle (1) at the time of driving to obtain non-ground point cloud data specifically includes: Using a ground estimation algorithm, the environment point cloud data is divided to obtain non-ground point cloud data and ground point cloud data; The ground point cloud data is eliminated to obtain non-ground point cloud data.

7. The method for identifying a drivable lane area based on multi-sensor fusion according to claim 6, characterized in that: Before dividing the environmental point cloud data, the method further includes: The environmental point cloud data is preprocessed, wherein the preprocessing includes at least one of denoising and filtering.

8. The method for identifying a drivable lane area based on multi-sensor fusion according to claim 1, characterized in that: Using the LOAM algorithm, the boundary points in the non-ground point cloud data are extracted to obtain the center line of the lane for the vehicle (1) to move forward, specifically including: A point cloud feature extraction algorithm is used to extract features from the non-ground point cloud data to obtain feature data, wherein the feature data includes a curvature of a wall surface projected on a horizontal plane and a geometric shape of an attached component on the wall surface projected on a horizontal plane; The feature tracking algorithm performs feature tracking on the non-ground point cloud data to obtain visible feature data between consecutive frames.

9. The method for identifying a drivable lane area based on multi-sensor fusion according to claim 1, characterized in that: A laser radar (2) is arranged at the rear of the vehicle (1).

10. A lane drivable area recognition system based on multi-sensor fusion, characterized by: include: A laser radar (2) for collecting point cloud data of the environment in front of the vehicle (1); A motion compensation module, which is used to perform motion compensation on the environmental point cloud data acquired by the laser radar (2) to obtain environmental point cloud data at the time when the vehicle (1) is traveling; A data processing module, which is used to process the non-ground point cloud data to obtain non-ground point cloud data; A centerline acquisition module uses the LOAM algorithm to extract boundary points in the non-ground point cloud data to obtain the centerline of the lane for the vehicle (1) to move forward.

Images