Method for unmanned truck locking station parking

Abstract

The application discloses an unmanned truck locking station parking method, aiming at the locking station interaction scene in the process of loading and unloading containers by unmanned trucks, using a laser radar sensor, through the way of increasing a high reflectivity marker to exclude interference factors in the environment, combining with the shape fitting of the locking station, accurate locking station detection is realized; by constructing a local map of the locking station parking area, the locking station position in the vehicle driving process is updated in real time, even if the sensor receives shielding, accurate parking can still be realized; the scheme effectively solves the problem that the conventional locking station parking scheme depends on the positioning result and cannot be applied in the GPS shielding area by detecting the locking station; at the same time, by constructing the local map of the locking station area, the problem that the locking station cannot be effectively observed due to shielding, sensor field of view limitation and other problems when the vehicle is close to the locking station position is avoided.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of unmanned container truck parking safety, in particular to a method for locking station parking of an unmanned container truck. BACKGROUND

[0002] The locking station automatic parking function is one of the key links for realizing the automation of the operation process of the unmanned container truck, and the core and difficulty thereof lies in how to realize the accurate detection and precise parking of the locking station.

[0003] There are mainly two schemes for the locking station parking function of the existing unmanned container truck.

[0004] The first scheme is based on map information and vehicle self-positioning, and needs to rely on a navigation map to mark the position of the locking station, and requires that the position of the locking station be fixed, which has poor flexibility in application. Since there are a large number of large machines in the container terminal operation environment, the GPS signal is severely blocked, which causes the method to be unable to be applied in the GPS signal blocked area such as a shore crane. Meanwhile, the terminal operation condition is complex, and the position of the locking station needs to be adjusted in real time according to the operation condition. Once the position is adjusted, the position of the locking station needs to be updated by changing the map in the case of the scheme, which is difficult to meet the real-time requirement of the terminal operation.

[0005] The other scheme calculates the parking position by detecting the position of the locking station in real time, which overcomes the problem of relying on the map and positioning in the first scheme, and has a wider application scenario, but still faces many problems: 1) the success rate of locking station detection, since the locking station is usually a frame structure, which is common in the terminal and similar to the structure of many operation related equipment, and is easy to confuse, which puts high requirements on the detection scheme; 2) the blocking problem when the vehicle is close to the locking station, since the distance between the vehicle and the locking station is usually within 0.5 m when the container is loaded or unloaded, and two (or more) locking stations are usually arranged on one side of the vehicle. The vehicle-mounted sensing sensor (usually a laser radar) is usually arranged at the front of the vehicle, and the field of view is often blocked in this working condition, which affects the parking accuracy and effect.

[0006] In summary, the existing locking station parking scheme still has many problems in the application process, which can be summarized as follows:

[0007] 1) It relies on the navigation map and the positioning result, is subject to the operation scene, and has poor flexibility;

[0008] 2) The stability of the detection result is poor, and is easily affected by other elements in the scene and by the blocking. SUMMARY

[0009] To solve the above technical problems, the embodiment of the present application provides a self-driving truck locking station parking method, a scheme for constructing a local map of a locking station area, and local map construction in the area near the locking station through simultaneous localization and mapping.

[0010] The embodiment of the present application provides a self-driving truck locking station parking method, which comprises the following steps:

[0011] S1: After obtaining laser radar point cloud data, screening point clouds in a target region as candidate point clouds;

[0012] S2: Based on the obtained candidate point clouds, applying a locking station shape template to perform data fitting to obtain a locking station pose;

[0013] S3: After recognizing the locking station based on the locking station pose, constructing a local map of the locking station area according to the distance between the vehicle and the locking station;

[0014] S4: Obtaining the detection result of the vehicle pose in the local coordinate system and the locking station, and updating the locking station pose in the local map;

[0015] S5: Based on the locking station position in the map, judging a parking position and guiding the vehicle to park.

[0016] Further, the laser radars are arranged in a horizontal installation mode and at least simultaneously distributed on the top of the truck head, the left side of the truck head and the right side of the truck head.

[0017] Further, the field of view ranges of all the laser radars completely cover the periphery of the truck body when there is no box body on the truck.

[0018] Further, the laser radar point cloud data forms a high reflectivity point cloud according to a high reflectivity point cloud formed by a locking station reflective sticker to obtain a target region for fitting the locking station.

[0019] Further, the reflective sticker is attached to a locking station stand.

[0020] Further, the change of the vehicle pose in the local map of the locking station area is obtained through laser radar point cloud inter-frame matching to avoid GPS position errors caused by GPS occlusion.

[0021] Further, in step S4, the vehicle pose first detected to the locking station is taken as the origin to start constructing the local map.

[0022] Further, in step S4, laser radar odometry is used to obtain point cloud data, and point cloud data inter-frame registration is performed to obtain the vehicle pose in the map; according to the vehicle pose in the local map, the detection result of the locking station associated therewith is projected on the map, and the correlation degree with the existing locking station in the map is calculated to determine whether it is a newly detected locking station for addition or an existing locking station for update.

[0023] Further, if a new lock station is detected in the step S4, a new lock station target is added in the local map; if there is no lock station observation, no operation is needed; if the detected lock station is an existing lock station in the map, the pose of the lock station target in the map is updated according to the detection result to reduce the pose error of the lock station caused by the change of distance and observation angle.

[0024] Further, in the step S5, the relative position relationship between the detected lock station and the vehicle is judged according to the obtained local map of the lock station area, and the parking interval of the vehicle is calculated according to the position of the lock station, so as to guide the vehicle to park by judging the distance between the lock station and the vehicle.

[0025] Effective effect: The application is aimed at the lock station interaction scene in the process of unmanned truck loading and unloading, uses a laser radar sensor, realizes accurate lock station detection by increasing a high reflectivity marker to exclude interference factors in the environment and combining lock station shape fitting, constructs a local map of the lock station parking area, and updates the lock station position in the process of vehicle driving in real time, so that accurate parking can be realized even if the sensor receives shielding; the scheme effectively solves the problem that the conventional lock station parking scheme depends on positioning results and cannot be applied in the GPS shielding area by detecting the lock station; at the same time, by constructing the local map of the lock station area, the problem that the lock station cannot be effectively observed due to shielding, sensor field of view limitation and other problems when the vehicle is close to the lock station position is avoided. BRIEF DESCRIPTION OF DRAWINGS

[0026] In order to more clearly illustrate the technical solutions in the embodiments of the application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description only constitute some embodiments of the application, and for those skilled in the art, other drawings can also be obtained without creative labor based on these drawings.

[0027] Figure 1 is a method flowchart provided by the embodiment of the application;

[0028] Figure 2 is a lock station reflective marker arrangement diagram;

[0029] Figure 3 is a laser radar arrangement diagram;

[0030] Figure 4 (a)-(b) are vehicle entering lock station process schematic diagrams.

[0031] Meaning of reference signs in the drawings:

[0032] 1, 2, 3, 4 - lock standing column, 5 - lock station, 6 - right front laser radar, 7 - roof laser radar, 8 - left front laser radar. DETAILED DESCRIPTION

[0033] In order to make the application purposes, features and advantages of the present application more obvious and easy to understand, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings of the embodiments of the present application. Obviously, the embodiments described below are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor fall within the scope of protection of the present application.

[0034] The present application will be further illustrated below in conjunction with the drawings and specific embodiments.

[0035] In the description of the present application, it should be understood that the terms "upper", "lower", "top", "bottom", "inner", "outer" and the like indicate the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the present application.

[0036] The embodiments of the present application provide a method for unmanned truck to lock station and park, and the present solution mainly uses a scheme of adding high reflectivity markers combined with the detection method of the lock station shape. Specifically, the method comprises the following steps:

[0037] S1: After obtaining the laser radar point cloud data, the point clouds in the target area are selected as candidate point clouds. The laser radar is installed horizontally and is distributed at least simultaneously on the top of the truck head, the left side of the truck head and the right side of the truck head.

[0038] As a specific embodiment, as shown in Figure 4 , three laser radars are installed on the top of the truck head, the left side of the truck head and the right side of the truck head, all of which are installed horizontally. Among them, the top radar of the truck head has a 360-degree horizontal field of view when there is no box, and when there is a box, the rearward field of view is lost due to the blocking of the box; the left and right radars are arranged on the left front of the truck head and the right front of the truck head, and their horizontal fields of view are both 270 degrees due to the blocking of the vehicle body. Through the arrangement of the three radars, the coverage of the vehicle body around (when there is no box) can be realized. The lock station configuration is shown in Figure 2 and Figure 3 . The lock station is an island-shaped structure, which is supported by four columns between the base and the roof. The high reflectivity label is pasted on the column, so that the laser radar can scan the reflective label at all angles. The pasting position of the reflective label depends on the form of the lock station, and the arrangement principle is to facilitate the scanning of the laser radar in all directions.

[0039] S2: Based on the obtained candidate point cloud, the laser radar point cloud data is formed into a high reflectivity point cloud according to the lock station retro-reflective marker, and a rough area of the point cloud for fitting the lock station is obtained as a target area, and the point cloud in the area is intercepted as the input point cloud for lock station fitting. In combination with the lock station shape prior, the lock station shape template is applied, and the corresponding lock station fitting template is used to fit the input point cloud, and the actual pose of the lock station is obtained.

[0040] S3: After recognizing the lock station based on the lock station pose, a local map of the lock station area is constructed according to the distance between the vehicle and the lock station.

[0041] S4: The detection result of the vehicle pose in the local coordinate system and the lock station is obtained, and the lock station pose in the local map is updated. Specifically:

[0042] The vehicle pose where the lock station is first detected is taken as the origin, and the local map is constructed;

[0043] According to the pose of the vehicle in the map, the point cloud inter-frame registration is recursively obtained (laser radar odometry). Although this scheme has a certain cumulative error, it can be ignored considering that the actual application range is only within 50m around the lock station, and its error can be ignored. At the same time, since the lock station map is only used to calculate the relative position relationship between the vehicle and the lock station to guide parking, the cumulative error of the result can also be ignored;

[0044] According to the pose of the vehicle in the local map, the lock station detection result associated therewith is projected on the map, and the correlation degree with the existing lock station in the map is calculated to determine whether it is a newly detected lock station for addition or an existing lock station for update. If a new lock station is detected, a new lock station target is added in the local map; if there is no lock station observation, no operation is needed; if the detected lock station is an existing lock station in the map, the pose of the lock station target in the map is updated according to the detection result, so as to reduce the lock station pose error caused by distance and observation angle change.

[0045] Within the preset mapping range, the above steps are repeatedly executed to obtain a local map of the lock station area.

[0046] S5: Based on the lock station position in the map, the parking position is judged to guide the vehicle to park. According to the obtained local map of the lock station area, the relative position relationship between the detected lock station and the vehicle is judged, and the parking interval of the vehicle is calculated according to the lock station position, and the vehicle is guided to park by judging the distance between the lock station and the vehicle.

[0047] Compared with the prior art, the lock station alignment scheme does not depend on a navigation map and a vehicle positioning result, is not limited by a lock operation point position of a terminal, can realize high-precision alignment parking in an area (an area requiring lock operation) where a lock station is arranged, is flexible in application, uses a laser radar, improves detection accuracy through high reflectivity data, accurately identifies a lock station position in combination with a lock station shape feature, and realizes high-precision parking, uses a local map construction mode, overcomes the problem of unstable results caused by shielding and the like when depending on a single frame detection result, and realizes effective and smooth result output in the entire lock station parking process.

[0048] The preferred embodiments of the application are described in detail above, but the application is not limited to the specific details in the above-described embodiments, and various equivalent transformations (such as quantity, shape, position, etc.) can be made to the technical solutions of the application within the technical concept scope of the application, and these equivalent transformations all belong to the protection of the application.

Claims

1. An unmanned tractor-trailer lock station parking method, characterized by, Comprise: S1: After obtaining the laser radar point cloud data, the point clouds in the target region are screened as candidate point clouds; S2: Based on the obtained candidate point clouds, a lock station shape template is applied for data fitting to obtain the lock station pose; S3: After identifying the lock station based on the lock station pose, a local map of the lock station region is constructed according to the distance between the vehicle and the lock station; S4: The detection result of the vehicle pose in the local coordinate system and the lock station is obtained, the lock station pose in the local map is updated, the point cloud data is obtained through the laser radar odometer, and the point cloud data frame registration is obtained recursively to obtain the vehicle pose in the map; According to the vehicle pose in the local map, the lock station detection result associated with it is projected on the map, and the correlation degree is calculated with the existing lock station in the map to determine whether it is a newly detected lock station for addition or an existing lock station for update; The step S3 starts to construct the local map with the vehicle pose where the lock station is first detected as the origin; S5: Based on the lock station position in the map, the parking position is judged to guide the vehicle to park.

2. The unmanned truck lock station parking method according to claim 1, wherein the laser radar is installed horizontally and distributed at least simultaneously on the top of the truck head, the left side of the truck head and the right side of the truck head.

3. The unmanned truck lock station parking method according to claim 2, wherein the field of view of all the laser radars completely covers the four sides of the truck body when there is no container on the truck.

4. The unmanned truck lock station parking method according to claim 2, wherein the laser radar point cloud data is obtained from the high reflectivity point cloud formed by the lock station reflective sticker to obtain the region of the point cloud for fitting the lock station as the target region. The reflective sticker is attached to the lock station stand. The change of the vehicle pose in the construction of the local map of the lock station region is obtained through the laser radar point cloud frame matching to avoid the GPS position error caused by the GPS occlusion. In the step S4, if a new lock station is detected, a new lock station target is added in the local map; If there is no lock station observation, no operation is needed; If the detected lock station is an existing lock station in the map, the lock station target in the map is updated according to the detection result to reduce the lock station pose error caused by the change of distance and observation angle.

5. The unmanned tractor trailer locking station parking method of claim 4, wherein, In the step S5, the relative position relationship between the detected lock station and the vehicle is judged according to the obtained local map of the lock station region, and the parking interval of the vehicle is calculated according to the lock station position to guide the vehicle to park.

6. The method of claim 1, wherein, ​ 7. The unmanned tractor trailer locking station parking method of claim 1, wherein, ​ 8. The method of claim 1, wherein, ​

Images