A laser radar-based unmarked parking space detection method for a towed vehicle

Abstract

The application discloses a kind of based on laser radar's with hanging tractor no identification parking stall detection method, comprising the following steps: step 1: the type of parking stall to be detected is obtained by interactive module, and the parameters of selected parking stall type are configured;Step 2: execute the positioning and mapping algorithm based on laser radar, and convert into occupancy grid navigation map;Step 3: during the operation of step 2, the parking stall search method on occupancy grid map is synchronously executed, and the optimal parking stall is found out;Step 4: the final target parking stall is confirmed by interactive module.

Description

Technical Field

[0001] This invention belongs to the field of autonomous parking technology, specifically relating to a method for detecting unmarked parking spaces for trailer-mounted tractors based on lidar. Background Technology

[0002] In recent years, automatic parking space detection has been widely used in passenger vehicles. The parking space detection methods can be roughly divided into two types according to the main sensor type: those based on vehicle cameras and those based on ultrasonic / millimeter-wave radar.

[0003] Ultrasonic / millimeter-wave methods are relatively early fuzzy detection approaches, suffering from low accuracy and numerous scene limitations. Vision-based methods are currently more common, using machine processing and understanding of image data to identify parking signs and markings online, thus determining available parking spaces. However, these methods require clearly marked parking spaces, and the spaces cannot be too far from the observation location. However, parking with trailer-trailers often lacks clear parking space markings, and the length and movement of trailer-trailers restrict driving in these scenarios, requiring parking space detection to begin from a considerable distance. Therefore, vision-based autonomous parking solutions are difficult to widely apply in trailer-trailer parking scenarios.

[0004] LiDAR has advantages such as long detection range, high ranging accuracy, and all-weather operation. It can directly capture three-dimensional point cloud data of the surrounding environment, making it more suitable for autonomous parking applications with trailer tractors. Summary of the Invention

[0005] The purpose of this invention is to provide a method for detecting unmarked parking spaces for trailer-mounted tractors based on lidar, in order to solve the problem of online parking space detection for the automatic parking task of the vehicle preceding the trailer.

[0006] To address the aforementioned problems in the existing technology, the technical solution adopted by this invention is as follows:

[0007] A lidar-based system for detecting unmarked parking spaces for trailer-mounted tractors comprises three modules.

[0008] Module 1: The interaction module for selecting the parking space type for tractor vehicles is mainly used for basic information configuration and final target space selection.

[0009] Module 2: Online positioning and mapping function module for tractor vehicles, used for generating navigation information during online parking space detection.

[0010] Module 3: Online parking space detection module for tractor vehicles, enabling online search for available parking areas during vehicle operation.

[0011] A method for detecting unmarked parking spaces for trailer-mounted tractors based on lidar includes the following steps:

[0012] Step 1: First, obtain the type of parking space to be detected through the interactive module. This includes two main categories: vertical and diagonal, with corresponding left, right, and double-sided parking space types, totaling six types. Then, configure the parking space search parameters for different types through the interactive module. These include the maximum number of parking spaces to be detected (optional), the length and width of the parking space (including the overall length and width of the vehicle's front and trailer), and the higher-order detailed parameters involved in the parking space search algorithm. The parking space detection parameters have different preset search values ​​depending on the parking space type and usually do not need to be changed.

[0013] Step 2: The driver manually starts the vehicle and drives it slowly in a straight line. During the driving process, the localization and mapping (SLAM) algorithm based on LiDAR is executed. During the execution of the SLAM algorithm, the real-time constructed map is converted into an occupied grid map. The coordinate system is constructed with the vehicle's pose at the time of start as the origin.

[0014] Step 3: During the execution of Step 2, the parking space search method on the grid map is executed synchronously to find the optimal parking space.

[0015] Step 4: Confirm your final target parking space through the interactive module.

[0016] Step 3 specifically includes:

[0017] Step 31: First, introduce the complete parking space search parameter configuration. The default parameters are suitable for most standard trailers and applicable to both vertical and lateral parking methods for tow trucks. p3 refers to the vehicle's running distance from finding one parking space to the next search; p5 refers to the vehicle expansion grid number, which is an additional anti-collision expansion grid added to the outer layer after dividing the vehicle body grid into p1 (maximum trailer length) and p2 (maximum trailer width) grids based on p4 (grid map resolution); p6 refers to the initial search deviation angle relative to the current positioning; p7 refers to the initial lateral search distance relative to the current positioning from the center of the target parking space; p8 refers to the distance still required to walk after finding a parking space to meet the reversing conditions; p9 refers to the delayed decision distance to ensure that the laser point cloud can scan complete parking space information; p 10 p 11 This is a lateral threshold set to ensure that the searched parking spaces are centered relative to both sides, p 10 It is the extra search width, p 11 It is the extra search width step size; p 12 p 13 This is a longitudinal threshold to ensure that the searched parking spaces are located relatively far back; p 12 It is the additional search depth, p 13 It is the additional search depth step size; p 14 p 15 This is a rotation threshold to ensure the searched parking space angle is centered, p 14 It is the additional rotation angle, p15 This is the additional rotation angle step size. All the above parameters are preset fixed values, and will be defined as P according to their numbers thereafter. i＝1～15 {p i It should be noted that p8 and p9 are defined with the rear axle center of the vehicle as the reference position, while the last 6 parameters are defined with the parking space center as the reference position.

[0018]

[0019]

[0020] Table 1: Parking Space Search Configuration Parameters

[0021] Step 32: Define the current positioning information obtained by the positioning and mapping module as T(t,R), with the rear axle center as the reference position, where t is the three-dimensional displacement, R is the three-dimensional rotation matrix, and B{B l B r Define a three-dimensional offset and translation vector B composed of p7 and p9 in Table 1. This vector is used when searching for a vehicle moving to a left or right parking space, respectively: B l (-p9,p7,0.0), B r (-p9,-p7,0.0), defines the three-dimensional offset rotation vector Q obtained by transforming parameter 6 in Table 1. lr If the Euler rotation matrices for left and right rotations are -p6 and p6 respectively, then the pre-search position coordinates S on the vehicle side at each moment are:

[0022] S=(t+RB,RQ lr )

[0023] Step 33: In the initial search for parking space S, perform the parking space occupancy determination in a nested manner along the longitudinal, lateral, and rotational directions with their respective step sizes. Define the coordinates of a search space as S. k Based on p1 and p2, expand according to the vehicle body size, and based on p4, find all grid coordinates m occupied by the expanded vehicle body after p5 on the grid map obtained in step 2, and find the grid map data M under that coordinate. m (Unknown is defined as -1; empty as 0; occupied as 100), if all values ​​are 0, then S is considered to be... k These are candidate parking spaces.

[0024] To ensure that the final parking space is centered in orientation and position relative to both sides (when there are obstacles) and has a low depth, a three-level nested loop is used to perform a search from rotation, lateral, and longitudinal directions (the loop order is arranged from the inside to the outside according to priority) to obtain the optimal parking space. The following steps 331 to 335 are described in detail.

[0025] Step 331: The innermost layer of the nested search is a rotation loop, used for centering the parking space angle. Step 3, the occupancy determination, is executed sequentially. After the rotation loop cycle ends, the coordinate sequence of all candidate parking spaces that meet the conditions is stored. r .

[0026] Step 332: The second layer of the nested search is a horizontal loop, used to horizontally center the parking space relative to the left and right obstacles. The loop executes sequentially until the scope cycle ends, then the paired storage is performed. r Sequence size and l r The positional coordinates in the sequence are used to obtain sequence l. l .

[0027] Step 333: The outermost nested loop is a vertical loop to ensure the parking space depth is as low as possible. Each loop execution will store sequence l. l The sequence L is obtained by executing the loop sequentially until the end of the cycle. l .

[0028] Step 334: Based on L l Stored sequence l l l in r The sequence sizes are reordered to obtain a new sequence L. l ',take L l 'Largest size l l The sequence is the median coordinate, serving as the candidate parking space g. k .

[0029] Step 335: As the vehicle travels, it searches sequentially at intervals p3 to obtain the candidate available parking space sequence G. g And regarding G g Perform the condition check to see if p8 is met, until the final available parking space is stored in sequence V. g .

[0030] Steps 331 to 335 above apply to the six different parking space types selected by the interactive module in Process 1. When double-sided parking spaces are required, they can be processed synchronously in parallel or multi-threaded manner according to steps 431 to 435.

[0031] Step 5: Confirm your final target parking space through the interactive module.

[0032] The beneficial effects of this invention are as follows: by proposing parking space detection and parking space search, it solves the common problems faced by trailer-trailer tractors, such as lack of signage and long distances; in particular, this method is designed with adjustable parking space configuration based on the kinematic characteristics of trailer-trailer tractors and for different parking space types and reversing methods, which has strong practical value. Attached Figure Description

[0033] Figure 1 This is a schematic diagram of the parameter definition for the present invention. Figure 1;

[0034] Figure 2 This is a schematic diagram of the parameter definition for the present invention. Figure 2 ;

[0035] Figure 3 These are real-vehicle test cases for the present invention. Detailed Implementation

[0036] The present invention will be further described below with reference to the accompanying drawings and reference numerals.

[0037] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0038] The terms “first,” “second,” “third,” etc., are used only to distinguish descriptions and should not be interpreted as indicating or implying relative importance.

[0039] In the description of this invention, it should also be noted that, unless otherwise explicitly specified and limited, the terms "set," "install," "connect," and "link" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0040] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0041] Example 1:

[0042] like Figure 1-3 As shown, a method for detecting unmarked parking spaces for trailer-mounted tractors based on lidar includes the following steps:

[0043] Step 1: Obtain the type of parking space to be detected through the interactive module;

[0044] Step 2: Execute the LiDAR-based localization and mapping algorithm and convert it into an occupied grid map;

[0045] Step 3: Make the vehicle drive in a straight line. During the driving process, execute the localization and mapping (SLAM) algorithm based on LiDAR. During the execution of the SLAM algorithm, the real-time constructed map will be converted into an occupied grid map. The coordinate system is constructed with the vehicle's pose at the time of start as the origin. Simultaneously execute the parking space search method on the occupied grid map to find available parking spaces.

[0046] Step 4: Confirm your final target parking space through the interactive module.

[0047] Example 2:

[0048] Based on Example 1, step 3 specifically includes:

[0049] Step 31: Obtain relevant parameters;

[0050] Step 32: Using the center of the rear axle as a reference, establish the coordinates S of the pre-search position;

[0051] Step 33: Search for parking spaces. In the initial search for parking space S, the vertical, horizontal and rotational directions are used to perform the parking space occupancy determination in a nested manner with their respective step sizes.

[0052] Example 3:

[0053] Based on Example 2, in step 32, the current positioning information obtained by the positioning and mapping module is T(t,R), where t is the three-dimensional displacement and R is the three-dimensional rotation matrix.

[0054] Define p7 and p9 as a three-dimensional offset translation vector B, which is used when searching for a vehicle moving to a left or right parking space, respectively: B l (-p9,p7,0.0), B r (-p9,-p7,0.0), define p6 to obtain the three-dimensional offset rotation vector Q. lr When the Euler rotation matrices for left and right rotations are -p6 and p6 respectively, the pre-search position coordinates S on the vehicle side at each moment are:

[0055] S=(t+RB,RQ lr )

[0056] Where p6 represents the initial search rotation, p7 represents the lateral distance, p9 represents the delayed search distance, t is the three-dimensional displacement, R is the three-dimensional rotation matrix, and B{B l B r}

[0057] Example 4:

[0058] Based on Example 3, in step 33, a certain search position coordinate in the pre-search position coordinate S is defined as s. kBased on p1 and p2, the vehicle body is unfolded according to its dimensions. Using parameter p4, all grid coordinates m occupied by the unfolded vehicle body (including parameter p5) are located on the grid map obtained in step 2. The grid map data M under these coordinates is then retrieved. m Is it empty? If it is empty, then S is considered empty. k The candidate parking spaces are p1, p2, p4, and p5. The maximum length of the vehicle trailer is p1, the maximum width of the vehicle trailer is p2, p4 is the grid map resolution, and p5 is the number of grid cells for vehicle expansion.

[0059] Example 5:

[0060] Based on Example 4, step 33 involves three nested loops performing searches from rotational, lateral, and longitudinal perspectives to obtain preferred parking spaces, specifically including the following steps:

[0061] Step 331: The innermost layer of the nested search is a rotation loop, storing the coordinate sequence of all candidate parking spaces that meet the conditions. r ;

[0062] Step 332: The second level of nested search is a horizontal loop, with paired storage. r Sequence size and l r The positional coordinates in the sequence are used to obtain sequence l. l ;

[0063] Step 333: The outermost nested loop is a vertical loop. Each loop execution will store sequence l. l The sequence L is obtained by executing the loop sequentially until the end of the cycle. l .

[0064] Step 334: Based on L l Stored sequence l l l in r Reorder the sequences by size and take the new sequence L that is the largest. l 'l l The midpoint coordinates of the sequence are used as the preferred available parking space g. k .

[0065] Step 335: As the vehicle travels, it searches sequentially at intervals p3 to obtain a sequence of candidate available parking spaces G. g ; and regarding G g Perform a condition check to see if p8 is met. If the backward distance is not less than p8, the condition is met; if the backward distance is less than p8, the condition is not met. Change all parking spaces that meet the condition to the final available parking spaces and store them in sequence V. g p3 represents the parking space spacing, and p8 represents the rearward distance.

[0066] This invention is not limited to the above-described optional embodiments. Anyone can derive other various forms of products under the guidance of this invention. However, regardless of any changes made in their shape or structure, any technical solution that falls within the scope of the claims of this invention shall be protected by this invention.

Claims

1. A method for detecting unmarked parking spaces for trailer-mounted tractors based on lidar, characterized in that: Includes the following steps: Step 1: Obtain the type of parking space to be detected through the interactive module; Step 2: Execute the LiDAR-based localization and mapping algorithm and convert it into an occupied grid map; Step 3: During the execution of Step 2, the parking space search method on the grid map is executed synchronously to find the optimal parking space; Step 3 specifically includes: Step 31: Obtain relevant parameters; Step 32: Using the center of the rear axle as a reference, establish the coordinates of the pre-search position. ; Step 33: Search for parking spaces, in the initial parking space search. The determination of whether a parking space is occupied is performed sequentially and nested in the three directions of vertical, horizontal, and rotation, with their respective step sizes; in step 33, a certain search position coordinate in the pre-search position coordinates is defined as... ,according to Expand according to vehicle dimensions and based on parameters Find the raster map containing the parameters in step 2. All grid coordinates occupied by the rear body unfolding And find the raster map data at that coordinate. Is it empty? If it is empty, then it is considered... Candidate parking spaces; among them , , , ; In step 33, the three nested loops perform searches from rotation, lateral, and vertical perspectives respectively to obtain the optimal parking space, specifically including the following steps: Step 331: The innermost layer of the nested search is a rotation loop, which stores the coordinate sequence of all candidate parking spaces that meet the conditions. ; Step 332: The second level of nested search is a horizontal loop, with paired storage. Sequence size and The position coordinates in the sequence are used to obtain the sequence. ; Step 333: The outermost nested loop is a vertical loop, and the sequence is stored in each loop iteration. The sequence is obtained by executing the loop sequentially until the end of the cycle. ; Step 334: Based on Storing sequences In The sequence sizes are reordered to obtain a new sequence L. l ',take L l 'Largest size l l The sequence is the median coordinate, which serves as the preferred parking space. ; Step 335: During the vehicle's operation, according to... By sequentially searching at intervals, a sequence of candidate available parking spaces is obtained. ; and on Does the execution satisfy? The condition is judged until it is changed to the final available parking space and stored in the sequence. ; Indicates the distance between parking spaces. Indicates backward distance; Step 4: Confirm your final target parking space through the interactive module.

2. The method for detecting unmarked parking spaces for trailer-mounted tractors based on lidar according to claim 1, characterized in that: In step 32, the current location information is obtained. ,in For three-dimensional displacement, It is a three-dimensional rotation matrix; definition , Composition of three-dimensional offset and translation vectors The vector is used when searching for a vehicle to park in a left- or right-hand parking space, respectively: , ,definition The transformation yields a three-dimensional offset rotation vector. When rotating left and right using Euler rotation matrices, they are respectively , Then the pre-search position coordinates of the vehicle side at each moment for: in, For three-dimensional displacement, It is a three-dimensional rotation matrix. .

Images