Method for automatically stacking material cages in carriage of unmanned forklift

Abstract

The invention belongs to the technical field of logistics management, particularly relates to a method for automatically stacking material cages in an unmanned forklift carriage, and is suitable for automatic loading and unloading scenes in the fields of logistics, manufacturing and the like. The core of the method is to solve the problem of high-precision and tight stacking of multiple columns and multiple rows of material cages in a long and narrow compartment space of the unmanned forklift. Through combination of 3D environment perception, laser / visual SLAM navigation, carriage wall fitting and tracking control based on point cloud and differential accurate positioning strategies for side cages and middle cages, unmanned and high-density stacking operation of the whole process is achieved. Wherein for the placement of the middle material cage, a visual auxiliary positioning and error compensation mechanism based on the placed material cage supporting legs is innovatively introduced, the track accumulative error is effectively overcome, and the alignment and compactness of the whole row of material cages are ensured. The automation level, the space utilization rate and the operation safety of loading and unloading operation are remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of logistics management technology, specifically to a method for automatically stacking material cages inside an unmanned forklift truck. Background Technology

[0002] With the rapid development of smart logistics and "lights-out factories," automated guided vehicles (AGVs / AMRs) have been widely used in material handling within structured environments such as warehouses and production lines. The typical process involves loading materials into standardized pallets (cages), which are then quickly transferred by the AGVs. However, achieving fully automated loading and unloading of pallets in external cargo loading scenarios, such as truck beds and shipping containers, still faces significant technical challenges. These scenarios are characterized by: firstly, the long, narrow, and enclosed spaces limit the maneuverability and adjustment space of the AGVs; secondly, the environment is highly unstructured, with potentially uneven floors and a lack of clear path guidance markings; and thirdly, the shielding effect of the metal truck bed completely eliminates external positioning signals such as GPS (i.e., a GPS-denied environment). These characteristics place extremely high demands on the positioning and navigation accuracy, path planning capabilities, and end-effector execution precision of the AGVs.

[0003] In view of this, the present invention proposes a method for automatically stacking material cages in the compartment of an unmanned forklift, which is suitable for high-precision and tight stacking of multiple rows and columns of material cages in narrow loading spaces, thereby improving the automation level, space utilization and operational safety of material cage loading and unloading operations. Summary of the Invention

[0004] The purpose of this invention is to overcome the above-mentioned shortcomings of the prior art and provide a method for automatically stacking material cages in the compartment of an unmanned forklift; it is especially suitable for high-precision and tight stacking of multiple rows and columns of material cages in narrow loading spaces.

[0005] The technical solution adopted in this invention is: a method for automatically stacking material cages inside an unmanned forklift truck, comprising the following steps: S1. Cage stacking layout planning: Obtain the effective loading space dimensions of the car body (or other box-shaped storage space), and combine the size parameters of the material cage (or pallet) and the safety distance to calculate the number of rows N of material cages placed in the car body and the number of material cages M in each row; and plan the global stacking layout of the material cages. S2. Navigation to the compartment: The unmanned forklift autonomously navigates to the loading ramp at the entrance of the truck body based on an environmental map built using laser SLAM or visual SLAM in the area outside the truck body. S3. Driving Plan: Based on the current sequence number of the cages to be placed, the theoretical distance that the unmanned forklift will travel from the entrance of the truck bed to the target placement position is initially calculated; this theoretical distance is used as a rough reference for path tracking. S4. Tracking inside the carriage: After the unmanned forklift enters the truck bed, point cloud data of one side of the truck bed wall is collected in real time, and a straight line of the truck bed side wall is fitted. The vertical distance d from the forklift's rotation center to the fitted straight line of the truck bed side wall is calculated, and the vertical distance d is compared with the expected ideal distance D to obtain the deviation e = d – D. The ideal distance D corresponds to the distance from the target placement position of the current material cage to the truck bed side wall. The forklift's attitude is adjusted through closed-loop control to make the deviation e approach zero. Finally, the unmanned forklift travels along a straight path parallel to the truck bed side wall. S5. Deceleration and Stopping: During the operation of the unmanned forklift, the distance Dist between the forklift and obstacles in front is measured in real time; and the forward movement of the unmanned forklift is controlled in real time based on the value of Dist; the forward movement includes deceleration and stopping. Two thresholds are set for Dist: a first threshold T1 and a second threshold T2. When the distance is less than the first threshold T1, the forklift is controlled to decelerate; when the distance is less than the second threshold T2, the forklift is controlled to stop moving forward. S6. Selection of cage placement mode: Based on the current position type of the material cages to be placed in the stacking layout plan, select the corresponding placement mode; the position type includes side cages and middle cages; the side cages are material cages placed close to the side wall of the carriage in each row; the middle cages are material cages placed between two side cages; the number of middle cages is 1 or 0. If the cage to be placed is a side cage, then execute the side cage side-shifting placement step S7; if the cage to be placed is a middle cage, then execute the middle cage visual-assisted positioning placement step S8. S7. Side cages are moved to the side for placement: After the forklift stops moving forward and the mast is raised to the target height, control the forklift mast to move laterally towards the side wall of the target truck bed until the lateral movement termination condition is met, and then execute the lowering and placement of the material cage. S8. Visual aids for positioning and placement of the medium-sized cage; The system uses visual sensors to identify the positions of the support legs of the cages to be placed on the forklift, as well as the positions of the support legs of adjacent cages already placed in the same row. Based on the identified support leg position information, the system accurately calculates the ideal placement position of the cage to be placed and then executes the placement. S9. Forklift cycle: After placing one cage, the forklift reverses and lowers its forks to return to the truck bed, starting the next cycle of picking up, putting into the truck bed, and placing the cages, until all cages are stacked according to the plan.

[0006] Furthermore, the following steps are performed in step S8: a. Visual inspection: Images are acquired using a vision sensor mounted on the forklift mast; and a visual algorithm is used to identify and locate the positions of four key points in the image: the left and right front outriggers of the cage to be placed; the right front outrigger of the adjacent cage already placed on the left; and the left front outrigger of the adjacent cage already placed on the right. b. Offset calculation: In an ideal scenario, the cages are placed close together at equal intervals. The sum of the centerline positions of the already placed cages on the left and right sides is half of the desired centerline position X_center of the cage to be placed. This can be calculated from the positions of the right and left front legs of the already placed cages on both sides. The actual centerline position Xi of the cage to be placed is calculated from the positions of the left and right front legs. Therefore, the lateral offset Δx between the ideal and actual positions of the cage to be placed is Δx = Xi - X_center. c. Fine-tuning the placement of the gantry: If |△x| ≤ δ, where δ is half of the maximum lateral displacement of the gantry, the controller controls the gantry to laterally displace by a distance of -△x to compensate for the offset; after compensation is completed, the lowering and placement action is executed; and step S8 ends. If |△x|> δ, then proceed to the whole vehicle compensation step d; d. Vehicle compensation: The controller plans a lateral movement command to control the forklift differential drive chassis to perform a lateral translation Δx -sign(Δx)δ', where δ' is a reserved fine-tuning margin; after the forklift chassis completes the lateral translation, step ac is executed again.

[0007] Furthermore, the specific operation of step S1 is as follows: Before the operation begins, the unmanned forklift drives to the front of the truck bed and uses the onboard 3D camera or forward-facing LiDAR to scan the inside of the truck bed to obtain the effective loading length L1 and width L2 of the truck bed; then, combined with the material cage length L3, width L4 and required operating safety distance input by the back-end system, the number of rows N that the material cages can be placed and the number of rows M are generated, thereby generating a global stacking layout diagram; the operating safety distance includes at least the material cage spacing a and the material cage spacing b between the front and rear truck bed walls.

[0008] Furthermore, when placing the cages, start from the first row at the innermost side of the car. When placing each row of cages, always prioritize placing the side cages on both sides, and then place the middle cages. When there are multiple middle cages, the placement order is to start from the middle cage closest to the side cage each time, and place them from the outside inwards towards the center.

[0009] Furthermore, in step S7, the lateral movement termination condition is that the forklift mast lateral movement mechanism reaches its mechanical limit, or the forklift side sensor detects that the distance to the truck bed wall is less than the third threshold T3.

[0010] Furthermore, in step S4, the fitted straight line of the carriage side wall is obtained by processing the side laser point cloud using the Random Sampling Consensus Algorithm (RANSAC) or the least squares method.

[0011] Furthermore, in step S8, the vision sensor is an industrial camera mounted on the gantry or fork arm, which detects by identifying specific features of the cage legs or by template matching.

[0012] Furthermore, the first threshold is greater than the second threshold, and the second threshold is greater than the third threshold.

[0013] Compared with existing technologies, the beneficial effects of this invention are as follows: The proposed method for automatically placing multiple rows of material cages in containers or truck compartments using an unmanned forklift is applicable to automated loading and unloading scenarios in logistics, manufacturing, and other fields. It solves the problem of high-precision and compact placement of multiple rows and columns of material cages by unmanned forklifts in narrow truck compartments. By combining 3D environmental perception, laser / visual SLAM navigation, point cloud-based truck compartment wall fitting and tracking control, and differentiated precise positioning strategies for "side cages" and "middle cages," the invention achieves fully automated, high-density palletizing operations. In particular, for the placement of middle material cages, the invention innovatively introduces a visual-assisted positioning and error compensation mechanism based on the legs of already placed material cages, effectively overcoming track accumulation errors, ensuring the alignment and compactness of the entire row of material cages, and significantly improving the automation level, space utilization, and operational safety of loading and unloading operations. Attached Figure Description

[0014] Figure 1 This is a schematic diagram of the cage stacking layout in Embodiment 1 of the present invention. Detailed Implementation

[0015] To make the objectives, technical solutions, and advantages of this invention clearer, descriptions of well-known structures and techniques are omitted in the following description to avoid unnecessarily obscuring the concepts in this invention.

[0016] Based on the general size design of truck beds, containers and cages (or pallets) in the existing technology, truck beds and containers can generally only hold a maximum of 2-3 cages per row, that is, M can usually only be 1, 2 or 3. Example 1

[0017] like Figure 1 As shown, this embodiment provides a method for automatically stacking material cages inside an unmanned forklift truck, including the following steps: S1. Cage stacking layout planning: Obtain the effective loading space dimensions of the car body, and combine the size parameters of the material cages (or pallets) and the safety distance to calculate the number of rows N of material cages placed in the car body and the number of material cages M in each row; and plan the global stacking layout of the material cages. The specific operation is as follows: Before the operation begins, the unmanned forklift travels to the front of the truck bed and uses its onboard 3D camera or forward-facing LiDAR to scan the interior of the truck bed to obtain the effective loading length L1 and width L2 of the truck bed; then, combined with the material cage length L3, width L4, and required operating safety distance input from the backend system, the operating safety distance includes at least the material cage spacing a and the distance between the material cage and the front and rear truck bed walls b; calculate the number of rows N that can be placed in the truck bed and the number of items M per row, for example: based on the loading length L1, material cage length L3, and operating safety distance, calculate the number of rows N = floor((L1 - 2b) / (L3 + a)) + 1; similarly, based on the truck bed width L2 and material cage width L4, determine the number of items M per row. For ease of description, assume M = 3 in this embodiment; then generate a global stacking layout diagram (e.g. Figure 1 (As shown).

[0018] S2. Navigation to the compartment: After receiving the task, the unmanned forklift activates its laser SLAM (or visual SLAM) navigation function in the platform area. Based on the pre-built factory map, it plans a collision-free path from its current location to the entrance of the target truck's loading ramp and autonomously drives to the destination.

[0019] S3. Driving Plan: Based on the current cage number i (starting from the innermost row 1), the theoretical distance Si = (N - i) * (L3 + a) + b of the unmanned forklift traveling from the truck entrance to the target placement position is initially calculated; this theoretical distance is used as a rough reference for path tracking.

[0020] S4. Tracking inside the carriage: As the forklift drives onto the loading ramp and into the truck bed, the metal structure of the truck bed may interfere with laser SLAM, and the long corridor design could increase positioning errors. Therefore, the system switches to "geometric feature tracking mode." A 2D or 3D LiDAR mounted on the side of the forklift scans one side of the truck bed in real time to obtain point cloud data. The RANSAC algorithm is used to fit the equation of the straight line along the truck bed wall from the point cloud. The controller calculates the vertical distance d from the forklift's rotation center to this fitted line in real time and compares it with a preset ideal distance D (determined by the cage layout plan, i.e., the ideal distance D corresponds to the distance from the target placement position of the cage to the side wall of the truck bed), resulting in a deviation e = d - D. This deviation is input into the lateral controller, which adjusts the speed difference of the forklift's differential drive wheels to control the forklift's heading, bringing the deviation e close to zero, thus enabling the forklift to travel parallel to the truck bed wall in a straight line.

[0021] S5. Deceleration and Stopping: During the operation of the unmanned forklift, the distance Dist between the forklift and obstacles in front (such as the material cages that have been placed in front) is measured in real time; and the forward movement of the unmanned forklift is controlled in real time based on the value of Dist; the forward movement includes deceleration and stopping. Two thresholds are set for Dist: a first threshold T1 (e.g., 1.5 meters) and a second threshold T2 (e.g., 0.3 meters). When the distance Dist is less than the first threshold T1, the forklift is controlled to decelerate; when the distance Dist is less than the second threshold T2, the forklift is controlled to stop moving forward. This step is used to ensure longitudinal parking accuracy.

[0022] S6. Selection of cage placement mode: Based on the current position type of the material cage to be placed in the stacking layout plan, select the corresponding placement mode; the position type includes side cages and middle cages; the side cages are material cages placed close to the side wall of the carriage in each row; the middle cages are material cages placed between two side cages; If the cage to be placed is a side cage, then execute the side cage side-shifting placement step S7; if the cage to be placed is a middle cage, then execute the middle cage visual-assisted positioning placement step S8.

[0023] S7. Side cages are moved to the side for placement: If the cage to be placed is the first or third cage in each row (both "side cages" against the truck bed wall), after the forklift stops moving forward and the mast is raised to the target height, the forklift mast is moved laterally towards the target truck bed side wall until the mast lateral movement mechanism reaches its mechanical limit, or the lateral sensor detects that the distance to the truck bed wall is less than a minimum safety value (i.e., the third threshold T3). At this point, it is assumed that the cage is flush against the truck bed wall. Subsequently, the forks are lowered to place the cage on the bottom surface, thus completing the placement.

[0024] S8. Visual aids for positioning and placement of the medium-sized cage: If the task in this round involves placing the second cage in each row (i.e., the "middle cage"), and there are already cages placed on both sides adjacent to it, then the following specific operations should be performed: a. Visual inspection: Images are acquired using vision sensors mounted on the forklift mast; and visual algorithms are used to identify and locate the positions of four key points in the images: the left and right front legs of the cage to be placed; and the corresponding legs of the cages already placed on both sides (i.e., the right front leg of the cage immediately adjacent to the left and the left front leg of the cage immediately adjacent to the right). b. Offset calculation: Ideally, the cages are placed close together at equal intervals. The sum of the centerline positions of the already placed cages on the left and right sides is the desired centerline position X_center of the cage to be placed. This can be calculated from the positions of the right and left front legs of the already placed cages on both sides. The actual centerline position Xi of the cage to be placed is calculated from the positions of the left and right front legs of the cage to be placed. Let the lateral offset between the ideal and actual positions of the cage to be placed be Δx. Then Δx = Xi - X_center. c. Fine-tuning the placement of the gantry: If |△x| ≤ δ, where δ is half of the maximum lateral displacement of the mast, the controller controls the mast to laterally displace by a distance of -△x to compensate for the offset. After the compensation is completed, the controller controls the forklift to slowly move forward a preset small distance (to compensate for the parking gap), and then performs the lowering and placing action, and ends step S8. If |△x|> δ, then proceed to the whole vehicle compensation step d.

[0025] d. Vehicle compensation: When |△x| > δ, it indicates that the offset △x calculated by vision exceeds the mast's fine-tuning capability, meaning the forklift's lateral position deviation is too large. At this point, the controller plans a lateral movement command, controlling the forklift's differential drive chassis to perform a lateral translation of △x - sign(△x)δ', where δ' is the reserved fine-tuning margin. After the forklift chassis completes the lateral translation, the preceding vision detection, offset calculation, and mast fine-tuning operations are re-executed, i.e., step ac. This mechanism ensures that even with a large positioning error, precise alignment of the intermediate material cage can be achieved through two levels of adjustment: "chassis coarse adjustment + mast fine adjustment."

[0026] S9. Forklift cycle: After placing one cage, the forklift reverses and lowers its forks to return to the truck bed, starting the next cycle of picking up, putting into the truck bed, and placing the cages, until all cages are stacked according to the plan.

[0027] In the process of placing the material cages in this invention, the placement starts from the first row at the innermost side of the carriage. When placing each row of material cages, the side cages located on both sides (the first and third cages in each row in this embodiment) are always placed first, followed by the middle cages (the second cage in each row). The vision sensor in step S8 of this embodiment is an industrial camera installed on the gantry or fork arm, which can detect the material cages by identifying specific features of the cage legs or by using a template matching method. Example 2

[0028] In this embodiment, it is assumed that the number of material cages (or trays) that can be placed in each row is 2. The main difference between this embodiment and embodiment 1 is that there is no intermediate cage in this embodiment, so there is no step S8 in the placement stage.

[0029] The other steps in this embodiment are exactly the same as in Embodiment 1 and will not be repeated here.

[0030] The above are only some embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention can have various combinations and modifications of the aforementioned technical features. Any improvements, modifications, equivalent substitutions, or applications of the structure or method of the present invention to other fields to achieve the same effect without departing from the spirit and scope of the present invention shall fall within the protection scope of the present invention.

Claims

1. A method of automatically stacking cages within a cageless forklift truck bed, characterized by, The method comprises the following steps: S1. Cage stacking layout planning: Obtain the effective loading space size of the carriage, and combine the size parameters of the cage and the safety distance to calculate and plan the number of rows N and the number of cages M in each row of the cage in the carriage; S2. Navigation into the carriage: The unmanned forklift autonomously navigates to the loading bridge position at the entrance of the carriage based on the environment map constructed by laser SLAM or visual SLAM in the external area of the carriage; S3. Travel planning: According to the current to-be-placed cage order number, the theoretical mileage of the unmanned forklift from the carriage entrance to the target placement position is preliminarily calculated; The theoretical mileage is used as a rough reference for path tracking; S4. In-car tracking: After the unmanned forklift enters the carriage, the point cloud data of one side of the carriage wall is collected in real time, and the carriage side wall straight line is fitted; the perpendicular distance d of the forklift rotation center to the fitted carriage side wall straight line is calculated, and the perpendicular distance d and the expected ideal distance D are compared to obtain the deviation e = d-D; wherein the ideal distance D corresponds to the distance from the target placement position of the current to-be-placed cage to the carriage side wall; the forklift posture is adjusted through closed-loop control to make the deviation e tend to zero; and finally the unmanned forklift travels along a straight line path parallel to the carriage side wall; S5. Forward deceleration and stop: In the process of driving the unmanned forklift, the distance Dist between the forklift and the front obstacle is measured in real time; and the forward state of the unmanned forklift is controlled in real time according to the value of Dist; the forward state includes deceleration and stop; Two threshold values are set for Dist, which are a first threshold value T1 and a second threshold value T2; when the distance is less than the first threshold value T1, the forklift is controlled to decelerate; and when the distance is less than the second threshold value T2, the forklift is controlled to stop; S6. Cage placement mode selection: According to the position type of the current to-be-placed cage in the stacking layout planning, a corresponding placement mode is selected; the position type includes an edge cage and an intermediate cage; the edge cage is a cage placed close to the carriage side wall in each row; and the intermediate cage is a cage placed between two edge cages; If the current to-be-placed is an edge cage, the edge cage side shift placement step S7 is performed; and if the current to-be-placed is an intermediate cage, the intermediate cage visual auxiliary positioning placement step S8 is performed; S7. Edge cage side shift placement: After the forklift stops and the gantry is raised to the target height, the gantry of the forklift is controlled to shift to the target carriage side wall direction until the side shift termination condition is met, and then the cage lowering placement action is performed; S8. Intermediate cage visual auxiliary positioning placement: The positions of the to-be-placed cage feet on the forklift and the positions of the adjacent placed cage legs in the same row are recognized by the vision sensor; And according to the position information of the recognized legs, the ideal placement position of the current to-be-placed cage is accurately calculated, and the placement is performed; S9. Forklift cycle: After completing the placement of one cage, the forklift performs actions such as retreating and lowering the forks, drives back to the outside of the carriage, and performs the next cycle of taking, entering and placing, until all the cages are placed according to the planning.

2. The method of automatically palletizing cages within a cage compartment of a driverless fork truck of claim 1, wherein, In step S8, the following steps are performed: a. Visual detection: An image is acquired by a visual sensor mounted on the forklift mast; and a visual algorithm is used to identify and locate the positions of four key points in the image: the left front leg and the right front leg of the to-be-placed container; the right front leg of the closely adjacent placed container on the left side; and the left front leg of the closely adjacent placed container on the right side; b. Offset calculation: In an ideal case, the containers are placed at equal intervals, and the half of the sum of the centerline positions of the placed containers on the left side and the right side is the expected centerline position X_center of the to-be-placed container, which can be calculated from the positions of the right front legs and the left front legs of the placed containers on the two sides; the actual centerline position Xi of the to-be-placed container is calculated from the positions of the left front leg and the right front leg of the to-be-placed container; and the lateral offset △x between the ideal position and the actual position of the to-be-placed container is Xi - X_center; c. Mast fine adjustment placement: If |△x| ≤ δ, δ is half of the maximum lateral displacement of the mast, then the controller controls the lateral displacement of the mast by -△x to compensate for the offset; after the compensation is completed, a lowering placement action is performed; and the step S8 is ended; If |△x| > δ, then the step d of whole vehicle compensation is continued; d. Whole vehicle compensation: The controller plans a lateral movement instruction to control the differential drive chassis of the forklift to move laterally by △x - sign(△x)δ', where δ' is a reserved fine adjustment margin; after the lateral movement of the forklift chassis is completed, the steps a-c are re-executed.

3. The method of automatically palletizing cages within a driverless fork truck bed of claim 1, wherein, The specific operation of step S1 is as follows: before the operation starts, the unmanned forklift drives to the front of the carriage, uses the 3D camera or the front laser radar to scan the inside of the carriage, and obtains the effective loading length L1 and the width L2 of the carriage; then, in combination with the length L3 and the width L4 of the container and the required operation safety distance input by the background system, the number of rows N and the number of containers M in each row that can be placed are calculated, so as to generate a global stacking layout.

4. The method of automatically palletizing cages within a cageless fork truck bed of claim 4, wherein: The operation safety distance includes the container spacing a and the spacing b between the container and the front and rear carriage walls.

5. The method of automatically palletizing cages within a driverless fork truck bed of claim 1, wherein, When placing each row of containers, the side containers on the two sides are always placed first, and then the middle containers are placed.

6. The method of automatically palletizing cages within a driverless fork truck bed of claim 1, wherein, In step S7, the lateral movement termination condition is that the lateral movement mechanism of the forklift mast reaches its mechanical limit, or the lateral sensor of the forklift detects that the distance to the carriage wall is less than a third threshold T3.

7. The method of automatically palletizing cages within a driverless fork truck bed of claim 1, wherein, In step S4, the 2D or 3D laser radar mounted on the side of the forklift is used to scan the carriage wall in real time to obtain point cloud data, and the straight line of the carriage side wall is obtained by processing the side laser point cloud using the random sample consensus algorithm or the least squares method.

8. The method of automatically palletizing cages within a driverless fork truck bed of claim 1, wherein: In step S8, the visual sensor is an industrial camera mounted on the forklift mast or fork arm, which detects by recognizing the specific features of the container legs or by template matching.

9. The method of automatically palletizing cages within a driverless fork truck bed of claim 7, wherein: The first threshold is greater than the second threshold, and the second threshold is greater than the third threshold.

Images