Unmanned vehicle positioning method for container terminal scene

Abstract

The invention discloses an unmanned vehicle positioning method for a container terminal scene, and the method comprises the steps: obtaining RTK real-time positioning information through installing RTK client equipment; scanning point cloud information of all bridge cranes, rubber-tyred cranes and rail cranes to create point cloud templates of different types of equipment; the coordinates of the outer edges of the supporting legs of the port machinery are calculated according to the size of each port machinery device, the installation position of the RTK device and the RTK positioning information; laser point cloud is obtained through the unmanned vehicle, point cloud template matching is carried out on the laser point cloud and the point cloud template, and surrounding port machinery equipment is found; and obtaining the coordinates of the scanned port machinery according to the inertial navigation positioning information of the vehicle and the positioning information of the whole port machinery, and reversely calculating the current position of the vehicle according to the relative position relation so as to realize global optimal estimation. According to the method, peripheral port machinery equipment and coordinates thereof are used as positioning reference objects, so that the cost input is effectively reduced, the early-stage preparation conditions of the scheme are reduced, and the operation efficiency of the unmanned vehicle in the port area is improved.

Description

Technical Field

[0001] The present invention relates to a positioning method for an autonomous vehicle, specifically to a method for an autonomous vehicle in a container terminal scenario to assist positioning by using the RTK positioning information of traditional port machinery equipment, belonging to the technical field of autonomous driving of port autonomous vehicles. Background Art

[0002] Currently, there are mainly the following two implementation schemes for the positioning of autonomous vehicles in container terminals:

[0003] First, magnetic nail positioning and navigation are adopted - the positioning of the autonomous vehicle is realized by detecting the magnetic signals of magnetic nails through a magnetic navigation sensor, but its disadvantages are: high investment and deployment cost, usually tens of thousands of magnetic nails need to be laid, and since they are laid underground, it is not easy to change after laying, and rapid deployment cannot be achieved. High environmental requirements, no magnetic field interference is allowed.

[0004] Second, integrated positioning is adopted: a combination of multiple positioning methods, such as the commonly used scheme in the current terminal scenario using laser + vision + GNSS + inertial navigation. Its disadvantages are: in the container terminal scenario, containers and port machinery equipment move frequently, the environment often changes, especially in the quay crane area, there are few environmental features that can be stably detected and have longitudinal constraints, and it is difficult for the laser to match with the high-precision map point cloud.

[0005] Satellite signals are difficult to penetrate metal obstacles. For example, when under the quay crane, due to the huge steel structure of the quay crane, the vehicle often cannot receive satellite signals. In the container yard area, when the containers are stacked too high on both sides, only high-angle satellites available above can be relied on, and there are often situations where satellites cannot be searched.

[0006] Therefore, when the vehicle is under the quay crane or in the container yard area of the container yard, and there is no point cloud matching and the satellite signal is lost, the vehicle can only output the positioning result through inertial navigation, and the error will increase with time. Summary of the Invention

[0007] The technical problem to be solved by the present invention is to overcome the shortcomings of the prior art and provide a positioning method for an autonomous vehicle in a container terminal scenario. Based on the integrated positioning scheme, the positioning information of port machinery equipment is innovatively integrated to assist positioning, and the problems of abnormal positioning accuracy and increased error caused by multiple abnormal positioning results due to environmental reasons are solved.

[0008] To achieve the above purpose, the specific technical solution of the present invention is as follows: A positioning method for an autonomous vehicle in a container terminal scenario includes the following steps,

[0009] Step 1, install an RTK client device in an unobstructed area above the port machinery equipment and obtain RTK real-time positioning information;

[0010] Step 2: Scan the point cloud information of all gantry cranes, rubber-tired gantry cranes, and rail-mounted gantry cranes to create point cloud templates for different types of equipment. Since equipment of the same type has the same appearance, the same template is used. Due to the installation angle limitation of the lidar on the unmanned vehicle, only the four support legs and the girder of the port machinery equipment can be scanned. The four support legs of different types of equipment are quite different, so the point cloud of the four support legs is used as the template.

[0011] Step 3: Based on the dimensions of each port machinery equipment, the installation position of the RTK equipment, and the RTK positioning information in Step 1, calculate the coordinates of the outer edge of the support legs of the port machinery.

[0012] Step 4: The lidar on the unmanned vehicle scans the surrounding area with a range of 100 meters in length and 8 meters in height in real time to obtain the laser point cloud, and performs point cloud template matching with the point cloud template in Step 2 to find the port machinery equipment within 100 meters around.

[0013] Step 5: After the lidar scans the port machinery, based on the inertial navigation positioning information of the vehicle itself and the global port machinery positioning information, obtain the coordinates of the scanned port machinery, and reverse calculate the current position of the vehicle according to the relative position relationship, so as to achieve the global optimal estimation.

[0014] Furthermore, the port machinery equipment in Step 1 includes at least gantry cranes, rubber-tired gantry cranes, and rail-mounted gantry cranes, and these crane equipments usually have four support legs.

[0015] Furthermore, in Step 1, an RTK client device is installed on the steel structure top of the port machinery equipment and communicates with the virtual reference station provided by the port side to form a network.

[0016] Furthermore, in Step 2, use a point cloud acquisition vehicle to start scanning and collecting point cloud data from about 50 meters away from gantry cranes, rubber-tired gantry cranes, and rail-mounted gantry cranes. During the acquisition process, the point cloud acquisition vehicle scans around the side of the port machinery equipment. After the scanning is completed, the point cloud acquisition vehicle scans directly below the equipment to collect the complete surface point cloud data of the port machinery equipment.

[0017] Extract the point cloud data of the four support legs of the steel structure of the port machinery equipment from the laser point cloud, and create a point cloud template of the equipment according to the point cloud information.

[0018] Furthermore, in Step 3, the unmanned vehicle cloud system obtains the coordinate information of the RTK on all port machinery equipment by communicating with the RTK server, and calculates the coordinates of the four support legs according to the installation position of the RTK, the distance between the RTK and the four support legs, and the angle between the direction of the RTK device pointing to the target point and the due north direction. After obtaining the coordinates of the four support legs, the cloud system sends the equipment number, equipment model, and characteristic point coordinates to the unmanned vehicle together.

[0019] The calculation method of the coordinates of the support legs is as follows:

[0020]

[0021] Among them, Longitude target (i) represents the azimuth angle, which is the angle between the direction pointing to the target point and the due north direction with the installation position of the RTK device as the origin;

[0022] Longitude original (i) is the real-time coordinate longitude of the RTK position of the port crane i;

[0023] Latitude original (i) is the real-time coordinate latitude of the RTK position of the port crane i;

[0024] Longitude target (i)(j) is the real-time coordinate longitude of the j-th support leg feature point of the port crane i;

[0025] Latitude target (i)(j) is the real-time coordinate longitude of the j-th support leg feature point of the port crane i;

[0026] angle(i)(j) is the angle between the direction pointing to the j-th door leg feature point of the port crane i and the due north direction with the installation position of the RTK device of the port crane i as the origin;

[0027] distance(i)(j) is the distance from the installation position of the RTK device of the port crane i to the j-th door leg of the port crane i.

[0028] Furthermore, in step 5, if the GNSS positioning and the vehicle's own inertial navigation combined positioning cannot calculate the accurate vehicle positioning, the laser scans and identifies the port crane point cloud. According to the received information of all port crane devices in the field and its own coordinates, the coordinates of the four door leg feature points of the port crane pushed by the cloud are matched to the port crane point cloud, and the current vehicle coordinates are deduced based on the relative position relationship between the vehicle and the port crane point cloud. The specific steps are as follows:

[0029] Step 5.1, when the unmanned vehicle is running, the lidar scans the surrounding environment in real time to collect point cloud information, and identifies various types of port crane devices around by matching with the port crane device point cloud template;

[0030] Step 5.2, according to the current position coordinates of the unmanned vehicle and the coordinate information of each device sent by the cloud system, the perception module of the unmanned vehicle adds the device number tag to the port crane identified from the real-time point cloud and keeps tracking;

[0031] Step 5.3, when GNSS positioning fails, the driverless vehicle will continue to calculate the vehicle position based on inertial navigation positioning, and correct the inertial navigation result by using the vehicle coordinates and speed calculated from the quay crane coordinate information being tracked to replace the GNSS positioning result.

[0032] Further, in the said Step 1, the RTK device client can also be installed and fixed by the roadside. The installation position should ensure good line of sight with the RTK base station as much as possible and avoid being blocked by obstacles to ensure stable signal reception.

[0033] Further, in the said Step 1, a UWB system can also be installed on the quay crane equipment. The UWB system consists of tags, anchors and a positioning engine;

[0034] Tags, installed on the object to be positioned, are used to send and receive UWB signals;

[0035] Anchors, arranged at known positions, are used to receive the UWB signals sent by the tags and perform triangulation; the positioning engine is used to process the received UWB signal data and calculate the position coordinates of the target object;

[0036] By installing UWB tags on the quay crane equipment and deploying the anchors of the UWB system in the port, in order to ensure sufficient coverage between the anchors during triangulation, the service radius of each anchor is 30 - 50 meters.

[0037] Compared with the prior art, the beneficial effects of the present invention are as follows.

[0038] 1. The present invention has low cost investment. For a terminal with driverless vehicles in operation, there is already a need for quay crane positioning in the dispatching itself. Installing RTK equipment will not cause additional investment, killing two birds with one stone. Compared with most other solutions that mostly require new equipment to be added on each driverless vehicle or beside the lane specifically for the positioning of driverless vehicles, the investment is large.

[0039] 2. The present invention is easy to maintain. Since the appearance of the quay crane is very fixed and basically will not change during its life cycle, the vehicle can relatively easily detect and use its feature points and basically does not need to be updated.

[0040] 3. The present invention has high accuracy. The positioning accuracy of the RTK equipment is at the centimeter level, and the accuracy of the laser point cloud is at the millimeter level. Therefore, the theoretical error of this solution is also at the centimeter level, which can fully meet the on-site requirements. Compared with the decimeter-level accuracy of UWB positioning, it has great advantages. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 It is a flowchart of the present application.

[0042] Figure 2 It is a schematic diagram of the bridge crane point cloud information on the quay crane of the present invention.

[0043] Figure 3 This is a schematic diagram for calculating the coordinates of the support legs in the present invention. Detailed implementation manners

[0044] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings:

[0045] An unmanned vehicle positioning method for a container terminal scenario proposed in this embodiment is as Figure 1 shown, and the specific steps are as follows:

[0046] Step 1: Install an RTK client device in the unobstructed area above the port machinery equipment and obtain RTK real-time positioning information. Among them, the port machinery equipment includes at least gantry cranes, rubber-tired gantry cranes and rail-mounted gantry cranes, and these crane equipments usually have four support legs. At the same time, install an RTK client device on the steel structure top of the port machinery equipment and communicate and network with the virtual reference station provided by the port side.

[0047] Moreover, other locators can also be installed on the port machinery equipment, such as a UWB positioning system, which usually consists of tags, anchors and a positioning engine. The tag is installed on the object to be located and is used to send and receive UWB signals. The anchor is arranged at a known position and is used to receive the UWB signals emitted by the tag and perform triangulation. The positioning engine is used to process the received UWB signal data and calculate the position coordinates of the target object.

[0048] By installing UWB tags on the port machinery equipment and deploying the anchors of the UWB system in the port, in order to ensure that there is sufficient coverage between the anchors during triangulation, the service radius of each anchor is 30-50 meters.

[0049] Step 2: Scan the point cloud information of all gantry cranes, rubber-tired gantry cranes and rail-mounted gantry cranes to create point cloud templates for different types of equipment; for the same type of equipment, since the appearance is the same, the same template is used; due to the installation angle limitation of the lidar of the unmanned vehicle, only the four support legs and the girder of the port machinery equipment can be scanned, and the four support legs of different types of equipment are quite different, so the point cloud of the four support legs is used as the template.

[0050] Step 3: Based on the dimensions of each port machine, the installation location of the RTK device, and the RTK positioning information in Step 1, calculate the coordinates of the outer edge of the port machine's support legs. The unmanned vehicle cloud system obtains the RTK coordinate information on all port machines by communicating with the RTK server, and calculates the coordinates of the four support legs according to the installation location of the RTK, the distance between the RTK and the four support legs, and the angle between the direction of the RTK device pointing to the target point and the due north direction. After obtaining the coordinates of the four support legs, the cloud system sends the device number, device model, and feature point coordinates to the unmanned vehicle, as Figure 3 shown.

[0051] The calculation method for the support leg coordinates is as follows:

[0052]

[0053] where Longitude target (i) represents the azimuth angle, which is the angle between the direction pointing to the target point and the due north direction with the installation location of the RTK device as the origin;

[0054] Longitude original (i) is the real-time coordinate longitude of the RTK position of port machine i;

[0055] Latitude original (i) is the real-time coordinate latitude of the RTK position of port machine i;

[0056] Longitude target (i)(j) is the real-time coordinate longitude of the jth door leg feature point of port machine i;

[0057] Latitude target (i)(j) is the real-time coordinate longitude of the jth door leg feature point of port machine i;

[0058] angle(i)(j) is the angle between the direction pointing to the jth door leg feature point of port machine i and the due north direction with the installation location of the RTK device of port machine i as the origin;

[0059] distance(i)(j) is the distance from the installation location of the RTK device of port machine i to the jth door leg of port machine i.

[0060] Step 4: The lidar on the unmanned vehicle scans the surrounding area of 100 meters in length and 8 meters in height in real time to obtain the laser point cloud, and performs point cloud template matching with the point cloud template in Step 2 to find the port machines within 100 meters around;

[0061] Step 5: After the lidar scans the port crane, based on the self-vehicle inertial navigation positioning information and the full-field port crane positioning information, obtain the coordinates of the scanned port crane, and reverse calculate the current position of the vehicle according to the relative position relationship, so as to achieve global optimal estimation. The specific steps are as follows:

[0062] 1. Point cloud preprocessing: First, perform downsampling,

[0063]

[0064] Divide the point cloud into 5cm×5cm×5cm voxel grids and retain the center points of each grid.

[0065] Then perform ground filtering, use the RANSAC algorithm to remove the ground point cloud, and retain the points with a height > 1m.

[0066] 2. Feature extraction: Detect the outer edge of the support leg and use the Hough transform to extract the line features.

[0067] 3. Template matching (ICP algorithm): The point cloud registration formula is:

[0068]

[0069] where, R = rotation matrix, t = translation vector, p i = real-time point cloud, q i = template point cloud. Use SVD decomposition to solve the optimal transformation matrix.

[0070] 4. Matching threshold:

[0071]

[0072] If the GNSS positioning and the self-vehicle inertial navigation combined positioning cannot calculate the accurate positioning of the vehicle, the lidar scans and identifies the port crane point cloud. According to the received full-field port crane equipment information and its own coordinates, match the coordinates of the four portal leg feature points of the port crane pushed by the cloud to the port crane point cloud, and calculate the current coordinates of the vehicle according to the relative position relationship between the vehicle and the port crane point cloud. The specific steps are as follows:

[0073] Step 5.1, when the unmanned vehicle is running, the lidar scans the surrounding environment in real time to collect point cloud information, and identifies various types of port crane equipment around through template matching with the port crane equipment point cloud;

[0074] Step 5.2, according to the current position coordinates of the unmanned vehicle and the coordinate information of each device sent by the cloud system, the perception module of the unmanned vehicle adds the device number tag to the port crane identified from the real-time point cloud and keeps tracking;

[0075] Step 5.3, when GNSS positioning fails, the driverless vehicle will continue to calculate the vehicle position based on inertial navigation positioning, and correct the inertial navigation result by using the vehicle coordinates and speed calculated from the quay crane coordinate information being tracked to replace the GNSS positioning result.

[0076] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above specific embodiments, and the above specific embodiments and the descriptions in the specification are only for further explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements all fall within the scope of the present invention claimed. The scope of the present invention claimed is defined by the claims and their equivalents.

Claims

1. A method for positioning an unmanned vehicle in a container terminal scenario, characterized in that: The following steps are included: Step 1: Install the RTK client device in an unobstructed area above the port machinery equipment and obtain RTK real-time positioning information; Step 2: Scan the point cloud information of all bridge cranes, tire cranes and rail cranes to create point cloud templates for different models of equipment. The same model uses the same template because the equipment has the same appearance. Due to the installation angle limitation of the unmanned vehicle laser radar, only the four supporting legs and beams of the port machinery equipment can be scanned. The four supporting legs of different models of equipment are quite different, so the point cloud of the four supporting legs is used as the template. Step 3, according to the size of each port crane equipment, the installation position of the RTK equipment and the RTK positioning information in step 1, the coordinates of the outer edge of the port crane support leg are calculated; Step 4: The laser radar on the unmanned vehicle scans the surrounding area of ​​100 meters away and 8 meters high in real time to obtain laser point cloud, and matches the point cloud template with the point cloud template in step 2 to find the port machinery equipment within 100 meters; Step 5: After the laser radar scans the port crane, the coordinates of the scanned port crane are obtained based on the inertial navigation positioning information of the vehicle and the positioning information of the port cranes in the entire field, and the current position of the vehicle is reversely calculated based on the relative position relationship, thereby achieving global optimal estimation.

2. The unmanned vehicle positioning method for a container terminal scene according to claim 1 is characterized in that: The port machinery equipment in step 1 at least includes a bridge crane, a tire crane and a rail crane, and these crane equipment usually have four supporting legs.

3. The unmanned vehicle positioning method for a container terminal scene according to claim 1 is characterized in that: In step 1, an RTK client device is installed on the top of the steel structure of the port machinery equipment, and communicates with the virtual reference station provided by the port party.

4. The unmanned vehicle positioning method for a container terminal scene according to claim 1 is characterized in that: In step 2, a point cloud collection vehicle is used to scan and collect point cloud data starting from about 50 meters away from the bridge crane, tire crane and rail crane equipment; during the collection process, the point cloud collection vehicle scans around the side of the port machinery equipment, and after the scanning is completed, the point cloud collection vehicle scans directly below the equipment to collect point cloud data of the complete surface of the port machinery equipment; Extract the point cloud data of the four supporting legs of the steel structure of the port machinery equipment from the laser point cloud, and create a point cloud template for the equipment based on the point cloud information.

5. The unmanned vehicle positioning method for container terminal scene according to claim 1 is characterized by: In step 3, the cloud system of the unmanned vehicle obtains the coordinate information of the RTK on all port machinery equipment by communicating with the RTK server, and calculates the coordinates of the four supporting legs according to the installation position of the RTK, the distance between the RTK and the four supporting legs, and the angle between the direction of the RTK device pointing to the target point and the north direction. After obtaining the coordinates of the four supporting legs, the cloud system sends the device number, device model and feature point coordinates to the unmanned vehicle.

6. The unmanned vehicle positioning method for container terminal scene according to claim 5 is characterized by: The calculation method of the support leg coordinates is: Among them, Longitude target (i) Azimuth, which is the angle between the direction pointing to the target point and the true north direction, with the RTK device installation location as the origin; Longitude original (i) is the real-time coordinate longitude of the RTK position of port machinery i; Latitude original (i) is the real-time coordinate latitude of the RTK position of port machinery i; Longitude target (i)(j) is the real-time coordinate longitude of the feature point of the ith support leg of port machinery i; Latitude target (i)(j) is the real-time coordinate longitude of the feature point of the ith support leg of port machinery i; angle(i)(j) is the angle between the j-th gantry leg feature point pointing to the port crane i and the true north direction, with the RTK equipment installation position of the port crane i as the origin; distance(i)(j) is the distance from the RTK equipment installation location of port crane i to the jth door leg of port crane i.

7. The unmanned vehicle positioning method for container terminal scenes according to claim 5 is characterized by: In step 5, if the combined positioning of GNSS positioning and vehicle inertial navigation cannot calculate the precise positioning of the vehicle, the laser scans and identifies the port machinery point cloud, matches the coordinates of the four door legs of the port machinery pushed by the cloud to the port machinery point cloud according to the received information of the entire port machinery equipment and its own coordinates, and calculates the current coordinates of the vehicle according to the relative position relationship between the vehicle and the port machinery point cloud. The specific steps are as follows: Step 5.1: When the unmanned vehicle is running, the laser radar scans the surrounding environment in real time to collect point cloud information, and identifies various types of surrounding port machinery equipment by matching it with the point cloud template of the port machinery equipment; Step 5.2: Based on the current position coordinates of the unmanned vehicle and the coordinate information of each device sent by the cloud system, the unmanned vehicle perception module will add the equipment number tag to the port machinery identified from the real-time point cloud and keep tracking it; Step 5.3, when GNSS positioning fails, the unmanned vehicle will continue to calculate the vehicle position based on the inertial navigation positioning, and correct the inertial navigation results by using the vehicle coordinates and speed calculated from the coordinate information of the port machinery being tracked instead of the GNSS positioning results.

8. The unmanned vehicle positioning method for container terminal scenes according to claim 1 is characterized by: In step 1, the RTK device client can also be installed and fixed on the roadside to ensure that there is a good line of sight between the RTK client and the RTK base station without any obstacles.

9. The unmanned vehicle positioning method for a container terminal scene according to claim 1, characterized in that: In step 1, a UWB system may also be installed on the port machinery equipment, and the UWB system is composed of a tag, an anchor point and a positioning engine; A tag, mounted on the object to be located, for sending and receiving UWB signals; Anchor points are placed at known locations to receive UWB signals from tags and perform triangulation. Positioning engines process received UWB signal data and calculate the location coordinates of target objects. By installing UWB tags on port machinery equipment, anchor points of the UWB system are deployed in the port. In order to ensure sufficient coverage between the anchor points when performing triangulation, the service radius of each anchor point is 30 to 50 meters.

Images