An automated loading system

Abstract

This invention discloses an automated loading system, comprising: initializing loading parameters and rules based on orders; a forklift delivering material pallets to the depalletizing station; LiDAR scanning the vehicle and generating point cloud data on-site, and obtaining the vehicle body and loading position coordinates through AI models and algorithms; 3D vision guiding a column-type robot for depalletizing; a telescopic conveyor belt transporting the pallets; scanning and removing abnormal pallets during transport; and 3D vision guiding the loading mechanism to complete loading according to the rules. The system also includes a task dashboard, a step-crossing module, and auxiliary modules such as a driver's cab, achieving automated and intelligent loading, and improving efficiency and quality.

Description

Technical Field

[0001] This invention belongs to the field of cargo loading technology, and particularly relates to an automatic loading system. Background Technology

[0002] Traditional cargo loading operations rely heavily on manual labor to complete the entire process from material handling and vehicle positioning to cargo stacking. This results in low efficiency, high labor intensity, and unstable space utilization. Specifically, manual loading depends on operators' experience to judge the placement of goods, making it difficult to strictly follow scientific stacking rules such as "large to small, heavy to light," which can easily lead to wasted vehicle space or imbalance. Matching vehicle positioning with cargo dimensions relies entirely on manual measurement, which is prone to errors and time-consuming. Furthermore, manual involvement in unstacking and conveying processes can affect operational consistency due to fatigue and differences in operation, increasing the risk of cargo damage.

[0003] As the logistics industry's demand for automation increases, some scenarios are beginning to introduce simple mechanical auxiliary equipment. However, existing technologies still have limitations: for example, vehicle contour recognition often relies on a single sensor, making it difficult to adapt to complex on-site environments; the coordination between depalletizing and loading is insufficient, lacking precise positioning guidance based on 3D vision; and the cargo transportation and anomaly detection processes do not form a closed loop, failing to achieve intelligent management and control throughout the entire process. Therefore, there is an urgent need for a systematic solution integrating vehicle positioning, disordered depalletizing, intelligent transportation, and automated loading to improve loading efficiency and safety. Summary of the Invention

[0004] The purpose of this invention is to provide an automated loading system to solve the problems mentioned in the background section.

[0005] In view of this, the present invention provides an automated loading system, comprising the following steps:

[0006] S1. Loading System Settings: Based on the order task, initialize the loading parameters, collect the length, width, and height dimensions of the goods, measure the weight of the goods, and formulate loading rules;

[0007] S2. Manual forklift delivery: Forklift operators drive forklifts to transport pallets carrying materials to be loaded to the pre-set destacking station, and the forklifts travel along the route planned for the site.

[0008] S3. Vehicle Contour Scanning and Positioning: Using LiDAR, the loading vehicle and its surrounding environment are scanned. The LiDAR emits a laser beam and receives the reflected light signal to generate three-dimensional point cloud data. Through a deep learning-trained AI model, the vehicle contour features are identified. This AI model is trained based on vehicle contour sample data and has the ability to extract features and recognize patterns. The intelligent algorithm outputs the three-dimensional coordinates of the vehicle contour based on the point cloud data and extracts the planar and three-dimensional coordinates of the loading position, providing data support for the subsequent guidance of the loading mechanism.

[0009] S4.3D Vision Disordered Depalletizing: The LiDAR is used to scan the boxes to be depalletized to generate three-dimensional point cloud data. With the help of AI models, the center point and height coordinates of each box are determined, guiding the column robot to perform disordered grasping and depalletizing tasks. Multiple column robots are equipped to work together. The grasping mechanism adopts sponge suction cups and also includes a screw motor lifting mechanism, a cylinder-driven bottom gripper, and a fan-assisted sponge suction cup.

[0010] S5. Telescopic Belt Conveyor: The destacking boxes are transported to the loading station via a telescopic belt. The telescopic belt has a certain bandwidth, height and extension length. During the hydraulic lifting process, the hydraulic system raises the height of the belt. The motor drives the belt to move forward and backward, ensuring that the conveying task can be completed under different terrain conditions.

[0011] S6. Scanning, recording, and rejection of defects: During the conveying process, the scanning equipment is used to scan the appearance and position of the box. The scanning equipment uses image recognition to detect appearance abnormalities such as damage and stains on the surface of the box. At the same time, the sensors monitor the positional deviation of the box on the belt. If an abnormality is detected, the system triggers the alarm mechanism and starts the rejection device to remove the unqualified box from the conveyor line.

[0012] S7 3D Vision-Guided Automated Loading: Based on the vehicle's outline coordinates and loading algorithm planning, the loading mechanism is guided to arrange and push the boxes to the designated positions on the vehicle according to rules, completing the loading. The loading algorithm rules include prioritizing large to small, heavy to light, outside to inside, and order concentration. Horizontal boxes are prioritized for placement on both sides of the vehicle, while vertical boxes are placed in the middle. Simultaneously, information on unloaded products and loading progress are displayed in real time. The loading mechanism is equipped with 3D LiDAR to display the 3D outlines and coordinates of loaded and unloaded positions in real time. It automatically navigates to the loading position based on the coordinates provided by the intelligent system and completes loading according to the planned path. An independent vision system is equipped to perform appearance inspection on each box and issue an alarm.

[0013] In this invention, further, the vehicle contour scanning and positioning in step S3 specifically includes: the lidar scanning the loading site and the vehicle to generate three-dimensional point cloud data of corresponding resolution. The lidar has a horizontal field of view and a vertical field of view, which can cover the loading area.

[0014] The AI ​​model, trained through deep learning, identifies the contour features of the vehicle body. The AI ​​model adopts a corresponding neural network architecture and is trained with vehicle contour sample data from different models and angles.

[0015] Based on point cloud data, the intelligent algorithm outputs the three-dimensional coordinates of the vehicle body outline through corresponding algorithm steps, and extracts the planar and three-dimensional coordinates of the loading position to guide the loading mechanism.

[0016] In this invention, further, the 3D vision disordered depalletizing in step S4 adopts a column-type four-axis robot, whose parameters include: depalletizing speed: adapted to the corresponding operation requirements, multiple robots achieve alternating operation through a collaborative control algorithm; the gripping mechanism is a sponge suction cup, which can bear the corresponding weight of the box; the depalletizing guidance is realized through a 3D vision system, which is applicable to boxes of various sizes. The 3D vision system uses a camera and lens, combined with a specific image processing algorithm, to acquire the three-dimensional information of the box.

[0017] Furthermore, the gripping mechanism further includes: a lead screw motor lifting mechanism: the lead screw motor has a certain stroke range and corresponding lifting accuracy, and can adjust the gripping position according to the height of the box; a cylinder-driven bottom-supporting gripper: the cylinder provides corresponding clamping force to adapt to the lifting and clamping needs of boxes of different sizes; and a fan-assisted sponge suction cup: the fan generates corresponding negative pressure, and through optimized airflow channel design, the negative pressure is evenly distributed on the surface of the sponge suction cup.

[0018] In this invention, further, in step S5, the operating weight is within a certain range, and the drive motor of the belt conveyor can automatically adjust its output power according to different load conditions; it supports hydraulic lifting and motor drive to adapt to step crossing. The hydraulic lifting mechanism can lift the belt conveyor to the corresponding height within a certain time. The lifting height is within a certain range. The motor drive, through the corresponding motor and reducer, realizes the smooth movement of the belt conveyor, which can cross steps at a certain angle.

[0019] In this invention, the loading algorithm rules in step S7 further include: large to small, heavy to light, outside to inside, and order concentration. During the actual loading process, the system prioritizes the goods according to their size, weight, and order information using a corresponding sorting algorithm to ensure a reasonable loading order and stable vehicle center of gravity. Horizontal boxes are placed on the sides of the vehicle first, and vertical boxes are placed in the middle. The system automatically plans the placement of goods by performing three-dimensional modeling of the vehicle's internal space and matching calculations of the goods' dimensions. The system displays information on unloaded products and the loading progress in real time. Through the corresponding software system and display interface, the relevant information on unloaded products and the current loading progress are displayed in an intuitive chart format.

[0020] Furthermore, the system in this invention also includes a task dashboard module, used to obtain order tasks from the customer's warehouse management system. Through corresponding data interfaces and communication protocols, it connects with the customer's warehouse management system to obtain order information, including goods type, quantity, and shipping address. A loading plan is generated using a task optimization algorithm and a hybrid loading algorithm. The task optimization algorithm rationally allocates and sorts order tasks based on corresponding optimization objectives and constraints. The hybrid loading algorithm comprehensively considers factors such as the size and weight of the goods and order attribution to formulate a loading plan, improving loading efficiency and vehicle load factor. The system monitors loading status and anomaly feedback in real time. Through data interaction with various modules of the system, it obtains information such as loading progress and equipment operating status in real time. Once an anomaly is detected, it immediately alerts the operator through appropriate alarm methods for timely handling.

[0021] In this invention, the specific implementation of the 3D vision-guided loading mechanism in step S7 is as follows: the loading mechanism is equipped with a 3D LiDAR to display the three-dimensional contours and coordinates of the loaded and unloaded positions in real time. The 3D LiDAR updates the data in real time at a certain frequency, and the generated three-dimensional contours have corresponding accuracy, providing data support for the path planning of the loading mechanism. The loading mechanism automatically navigates to the loading position according to the coordinates provided by the intelligent system and completes the loading according to the planned path. The intelligent system uses corresponding navigation algorithms and path planning algorithms, combined with vehicle contour coordinates and cargo placement rules, to plan the driving path for the loading mechanism. An independent vision system performs appearance inspection on each box and issues an alarm. The independent vision system uses corresponding cameras and image processing algorithms to complete the appearance inspection of the boxes, detect appearance defects in the boxes, and trigger the alarm mechanism.

[0022] Furthermore, the system in this invention also includes a step-crossing process, specifically: a passive sensor detects the step height; the passive sensor employs a specific model and principle, possessing corresponding detection accuracy, and is capable of acquiring step height information; a hydraulic lifting mechanism adjusts the height of the loading mechanism; based on the step height detected by the sensor, the hydraulic lifting mechanism automatically adjusts the lifting height using a corresponding control algorithm to ensure the loading mechanism can smoothly cross the step; a motor drives the front wheels to touch the ground and propels the mechanism across the step; the motor adjusts its output torque and speed based on the action signal of the hydraulic lifting mechanism, driving the front wheels to touch the ground and propel the mechanism across the step, ensuring a smooth crossing process; and a hydraulic fork assists in completing the up-and-down step movements; the hydraulic fork extends and retracts as needed during the step-crossing process, providing additional support and balance assistance.

[0023] Furthermore, in this invention, the cockpit module integrates a loading machine, a telescopic belt, a side-push mechanism, and a multi-ball control interface;

[0024] The scanning and accounting module records cargo information and generates loading reports. Through data interaction with the scanning equipment, the module automatically records relevant cargo information and generates loading reports based on the completion of loading tasks. The reports are formatted in a standardized way for easy archiving and retrieval. The anomaly rejection module uses 3D vision recognition to identify and remove unqualified containers. Employing a 3D vision recognition algorithm, the module can identify unqualified containers and control the rejection device to remove them from the conveyor line, ensuring that unqualified containers do not enter the loading process and improving loading quality.

[0025] The beneficial effects of this invention are:

[0026] By combining LiDAR scanning with AI models, precise positioning of vehicle outlines and cargo locations is achieved, replacing traditional manual measurement and judgment. The collaborative operation of column-mounted robots and telescopic conveyor belts automates the entire process of depalletizing and conveying, reducing human intervention. For example, a dual-column robot, guided by 3D vision, completes disordered depalletizing, and combined with the stable conveying of the telescopic conveyor belt, significantly improving material handling efficiency.

[0027] Based on the loading algorithm rules of "large first, small second; heavy first, light second", combined with the vehicle's three-dimensional coordinate data, the cargo is scientifically stacked, avoiding the randomness of manual placement; the loading position coordinates are extracted by AI algorithm and the mechanism is guided to ensure the stability of the cargo's center of gravity and reduce the risk of tipping over during transportation.

[0028] The system integrates scanning and accounting functions as well as anomaly rejection. During the transportation of goods, it uses 3D vision to detect appearance defects and positional deviations, automatically rejects unqualified products and records the data. The task dashboard module monitors the loading progress in real time. Combined with the reports generated by the scanning and accounting module, it achieves full traceability from order to loading, reducing the loss rate of goods.

[0029] The lidar's ability to generate 3D point cloud data can adapt to different vehicle models and on-site environments, ensuring vehicle positioning accuracy; the telescopic belt conveyor supports hydraulic lifting and motor drive, and can cope with terrain obstacles such as steps; the step-crossing function improves the system's adaptability to uneven sites through the collaboration of sensors and hydraulic mechanisms.

[0030] Manual forklifts only need to complete the material pallet transfer, while subsequent high-intensity operations such as destacking and loading are handled by robots and automated mechanisms; the cockpit module integrates multiple equipment control interfaces, simplifying the operation process, reducing manual intervention, and reducing reliance on operator experience. Attached Figure Description

[0031] Figure 1 This is a simplified schematic diagram of the steps of the present invention;

[0032] Figure 2 This is a schematic diagram of the steps of the present invention. Detailed Implementation

[0033] In this application, the terms “upper,” “lower,” “left,” “right,” “front,” “back,” “top,” “bottom,” “inner,” “outer,” “middle,” “vertical,” and “horizontal” indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. These terms are mainly for the purpose of better describing this application and its embodiments, and are not intended to limit the indicated device, element, or component to having a specific orientation, or to be constructed and operated in a specific orientation.

[0034] This embodiment provides an automated loading system, including the following steps:

[0035] S1. Loading System Setup: Based on the order task, the loading parameters are initialized, the length, width, and height of the goods are collected, the weight of the goods is measured, and loading rules are formulated. In this step, the size and weight information of the goods are automatically collected by intelligent measuring equipment and uploaded to the system database. The loading rules need to be generated in combination with the characteristics of the goods, vehicle loading limits (including weight distribution and spatial dimensions), and transportation specifications. The rules will be stored synchronously in the system algorithm module as the basis for subsequent loading.

[0036] S2. Manual forklift delivery: Forklift operators drive forklifts to transport pallets carrying materials to be loaded to the pre-set depalletizing station. The forklift travels along the route planned for the site. The depalletizing station is equipped with positioning markers (such as ground QR codes or infrared positioning points). After the forklift arrives, the on-board positioning device is used to confirm the stopping accuracy, ensuring that the deviation between the center of the pallet and the reference point of the depalletizing robot's working range is within the preset range. The route planning must avoid equipment operating areas, personnel passages, and material stacking areas, and an emergency parking point must be set up.

[0037] S3. Vehicle Contour Scanning and Positioning: Using LiDAR, the loading vehicle and its surrounding environment are scanned. The LiDAR emits a laser beam and receives reflected light signals to generate three-dimensional point cloud data. A deep learning-trained AI model identifies the vehicle's contour features. This AI model is trained based on vehicle contour sample data and has feature extraction and pattern recognition capabilities. The intelligent algorithm outputs the three-dimensional coordinates of the vehicle contour based on the point cloud data and extracts the planar and three-dimensional coordinates of the loading position, providing data support for guiding the loading mechanism. The LiDAR is installed on a fixed bracket at the loading station, and the scanning range covers the entire length of the vehicle and a 2-meter area around it. After the point cloud data is generated, irrelevant information such as the ground and obstacles will be automatically filtered out, retaining only the data of the vehicle body and the area related to the loading operation. The coordinate calculation establishes a coordinate system with the rear wheel axle of the vehicle as the reference point.

[0038] S4.3D Vision-Based Disordered Depalletizing: Utilizing LiDAR to scan the boxes to be depalletized, generating 3D point cloud data, and aided by an AI model, determining the center point and height coordinates of each box, guiding the column-mounted robot to perform disordered grasping and depalletizing tasks. Multiple column-mounted robots work collaboratively, with the grasping mechanism employing sponge suction cups. The grasping mechanism also includes a screw motor lifting mechanism, a cylinder-driven bottom-supporting gripper, and a fan-assisted sponge suction cup. The LiDAR and robot are mounted on the same frame, with the scanning frequency matched to the robot's operating rhythm. The AI ​​model can identify the stacking gaps, tilt angles, and other states of the boxes. When the detected tilt exceeds a safety threshold, the robot will adjust its grasping angle to prevent the boxes from slipping.

[0039] S5. Telescopic Belt Conveyor: The destacking boxes are transported to the loading station via a telescopic belt conveyor. This telescopic belt conveyor has a certain bandwidth, height, and extension length. During the hydraulic lifting process, the hydraulic system raises the height of the belt conveyor. The motor drives the belt conveyor to move forward and backward, ensuring that the conveying task can be completed under different terrain conditions. The surface of the belt conveyor is equipped with anti-slip textures, and guide baffles are installed on both sides (the height of which can be adjusted according to the size of the box). The conveying speed can be automatically matched according to the subsequent loading rhythm (such as automatic deceleration when the loading mechanism is on standby). The hydraulic lifting stroke is adapted to the vehicle height to ensure that the height difference between the end of the belt conveyor and the floor of the car body is within the preset range.

[0040] S6. Scanning, Recording, and Removal of Abnormalities: During the conveying process, scanning equipment is used to scan the appearance and position of the boxes. The scanning equipment uses image recognition to detect appearance abnormalities such as damage and stains on the surface of the boxes. At the same time, sensors monitor the positional deviation of the boxes on the belt. If an abnormality is detected, the system triggers an alarm mechanism and starts the removal device to remove the unqualified boxes from the conveyor line. The scanning equipment consists of a top camera and sensors on both sides, which can identify the appearance status of the six sides of the box. Position monitoring uses infrared beam sensors on both sides of the belt. When the edge of the box exceeds the preset range of the belt centerline, it is determined to be an deviation. The removal device is a pneumatic pusher that pushes the abnormal box to the side temporary storage area and automatically records the type and quantity of abnormalities.

[0041] S7 3D Vision-Guided Automated Loading: Based on vehicle outline coordinates and loading algorithm planning, the loading mechanism is guided to arrange and push the containers to the designated positions on the vehicle according to rules, completing the loading. The loading algorithm rules include prioritizing large to small, heavy to light, outside to inside, order concentration, and placing horizontal containers on the sides of the vehicle first, while placing vertical containers in the middle. Simultaneously, it displays information on unloaded products and loading progress in real time. The loading mechanism is equipped with 3D LiDAR, displaying the 3D outlines and coordinates of loaded and unloaded positions in real time. It automatically navigates to the loading position based on coordinates provided by the intelligent system and completes loading according to the planned path. An independent vision system performs appearance inspection on each container and issues an alarm. A pusher plate (covered with cushioning material) is installed at the end of the loading mechanism; the pushing force can be adjusted according to the container material. The outline of the loaded position is updated to the system interface in real time. When the loaded goods exceed the preset stacking height, the system automatically pauses and prompts for adjustment.

[0042] In this invention, further, step S3, vehicle contour scanning and positioning, specifically includes: a LiDAR scanning the loading site and vehicle to generate three-dimensional point cloud data of corresponding resolution. This LiDAR has a horizontal and vertical field of view, capable of covering the loading area. An AI model trained through deep learning identifies the vehicle contour features. The AI ​​model employs a corresponding neural network architecture and is trained using vehicle contour sample data from different vehicle models and angles. An intelligent algorithm, based on the point cloud data and through corresponding algorithmic steps, outputs the three-dimensional coordinates of the vehicle contour and extracts the planar and three-dimensional coordinates of the loading position for guiding the loading mechanism. The LiDAR's horizontal field of view is not less than 120°, and its vertical field of view is not less than 60°, ensuring that a single scan covers the entire vehicle. The AI ​​model training samples include mainstream vehicle models such as trucks, container trucks, and vans, and cover scanning data under different environments such as daytime, nighttime, and rainy days. The intelligent algorithm uses a combination of point cloud clustering and edge extraction to eliminate interference from vehicle accessories (such as rearview mirrors and guardrails) on contour recognition.

[0043] In this invention, furthermore, the 3D vision-based disordered depalletizing in step S4 employs a column-type four-axis robot, whose parameters include: depalletizing speed: adapted to corresponding operational requirements; multiple robots working alternately through a collaborative control algorithm; the gripping mechanism is a sponge suction cup, capable of supporting boxes of corresponding weight; depalletizing guidance is achieved through a 3D vision system, applicable to boxes of various sizes; the 3D vision system uses a camera and lens, combined with a specific image processing algorithm, to acquire the three-dimensional information of the box; the robot's repeatability positioning accuracy is no greater than ±0.5mm; during collaborative operation, a "one-pick-one-place" alternating mode is adopted (i.e., while one robot is gripping, another completes the placement); the sponge suction cup uses sponges of different hardnesses depending on the box material (e.g., 40° hardness for cardboard boxes, 60° hardness for plastic boxes); the 3D vision system sampling frequency is no less than 30 frames / second to ensure real-time coordinates during dynamic gripping.

[0044] Furthermore, the gripping mechanism in this invention further includes: a screw motor lifting mechanism: the screw motor has a certain stroke range and corresponding lifting accuracy, and can adjust the gripping position according to the height of the box; a cylinder-driven bottom-supporting gripper: the cylinder provides corresponding clamping force to adapt to the lifting and gripping needs of boxes of different sizes; and a fan-assisted sponge suction cup: the fan generates corresponding negative pressure, and through an optimized airflow channel design, the negative pressure is evenly distributed on the surface of the sponge suction cup. The screw motor lifting stroke is not less than 500mm, and the positioning accuracy is not greater than ±0.1mm; the cylinder clamping force can be adjusted by a pneumatic valve (range 50-300N), and rubber pads are installed on the inner side of the gripper to prevent damage to the box; the fan negative pressure value can be adjusted according to the weight of the box, and the airflow channel adopts a multi-branch design to ensure that the negative pressure difference in each area of ​​the suction cup does not exceed 5%.

[0045] In this invention, further, in step S5, when the operating weight is within a certain range, the drive motor of the conveyor belt can automatically adjust its output power according to different load conditions; it supports hydraulic jacking and motor drive to adapt to step crossing. The hydraulic jacking mechanism can lift the conveyor belt to a corresponding height within a certain time, and the lifting height is within a certain range. The motor drive, through a corresponding motor and reducer, achieves smooth movement of the conveyor belt, enabling it to cross steps at a certain angle. The drive motor adopts frequency conversion control, and the output power is automatically adjusted according to load changes (power is reduced by more than 30% when unloaded); the hydraulic jacking response time does not exceed 2 seconds, and the lifting height range is 300-1500mm; the motor drive uses a servo motor, in conjunction with a planetary reducer, and the step crossing angle does not exceed 15°.

[0046] In this invention, the loading algorithm rules in step S7 further include: large to small, heavy to light, outside to inside, and order concentration. During the actual loading process, the system prioritizes goods based on their size, weight, and order information using a corresponding sorting algorithm to ensure a reasonable loading order and stable vehicle center of gravity. Horizontal boxes are prioritized for placement on both sides of the vehicle, while vertical boxes are placed in the middle. The system automatically plans the placement of goods by performing 3D modeling of the vehicle's internal space and matching calculations with the goods' dimensions. Information on unloaded products and loading progress are displayed in real time. Through the corresponding software system and display interface, relevant information on unloaded products and the current loading progress are presented in an intuitive chart format. The sorting algorithm uses a multi-factor weighted scoring system, assigning corresponding weights to size, weight, and order urgency. The 3D modeling accuracy error does not exceed 5mm, and a 50mm buffer gap is reserved during space matching. The display interface includes a goods list, a loading progress bar, a vehicle cross-section diagram (showing the loaded area in real time), and the estimated completion time.

[0047] Furthermore, the system in this invention also includes a task dashboard module, used to obtain order tasks from the customer's warehouse management system. Through corresponding data interfaces and communication protocols, it connects with the customer's warehouse management system to obtain order information, including goods type, quantity, and shipping address. A loading plan is generated using a task optimization algorithm and a hybrid loading algorithm. The task optimization algorithm rationally allocates and sorts order tasks based on corresponding optimization objectives and constraints. The hybrid loading algorithm comprehensively considers factors such as the size and weight of the goods and order attribution to formulate a loading plan, improving loading efficiency and vehicle load factor. The system monitors loading status and anomaly feedback in real time. Through data interaction with various modules of the system, it obtains information such as loading progress and equipment operating status in real time. Once an anomaly is detected, it immediately alerts the operator through appropriate alarm methods for timely handling. The data interface supports mainstream formats such as XML and JSON, and the communication protocol uses HTTPS encrypted transmission. Task optimization objectives include the shortest total loading time and the minimum number of vehicles used. Alarm methods include system interface pop-ups, audible and visual alarms, and push notifications to the operator's mobile app. Anomaly information includes the anomaly location, type, and suggested handling solutions.

[0048] In this invention, the specific implementation of the 3D vision-guided loading mechanism in step S7 is as follows: the loading mechanism is equipped with a 3D LiDAR to display the three-dimensional contours and coordinates of the loaded and unloaded positions in real time. The 3D LiDAR updates the data in real time at a certain frequency, and the generated three-dimensional contours have corresponding accuracy, providing data support for the path planning of the loading mechanism. The loading mechanism automatically navigates to the loading position according to the coordinates provided by the intelligent system and completes the loading according to the planned path. The intelligent system uses corresponding navigation algorithms and path planning algorithms, combined with vehicle contour coordinates and cargo placement rules, to plan the driving path for the loading mechanism. An independent vision system performs appearance inspection on each box and issues an alarm. The independent vision system uses corresponding cameras and image processing algorithms to complete the appearance inspection of the boxes, detect appearance defects in the boxes, and trigger the alarm mechanism.

[0049] The 3D LiDAR update frequency is no less than 10Hz, and the contour accuracy error is no more than 3mm; the navigation algorithm adopts AI algorithm, and the path planning avoids loaded goods and vehicle pillars; the independent vision system detects items including box damage (area ≥ 5cm²), deformation (deviation ≥ 10mm), missing labels, etc., and displays the defect location image simultaneously when an alarm is triggered.

[0050] Furthermore, the system in this invention also includes a step-crossing process, specifically: The step height is detected by a passive sensor, which employs a suitable model and principle, possessing appropriate detection accuracy to acquire step height information; a hydraulic lifting mechanism adjusts the height of the loading mechanism, automatically adjusting the lifting height based on the step height detected by the sensor using a corresponding control algorithm to ensure the loading mechanism can smoothly cross the step; a motor drives the front wheels to touch the ground and propels the mechanism across the step, adjusting the output torque and speed according to the action signal of the hydraulic lifting mechanism, ensuring a smooth crossing process; a hydraulic fork assists in completing the up-and-down step movements, extending and retracting as needed during the step-crossing process to provide additional support and balance assistance. The passive sensor is an ultrasonic sensor; the hydraulic lifting control algorithm uses PID regulation to ensure no overshoot during height adjustment; the extension length of the hydraulic fork is automatically matched according to the step depth, maintaining a 5° angle with the ground during extension to reduce impact.

[0051] Furthermore, in this invention, the cab module integrates a loading machine, a telescopic belt, a side-push mechanism, and a multi-ball control interface; the scanning and accounting module records cargo information and generates loading reports. Through data interaction with the scanning equipment, the scanning and accounting module automatically records relevant cargo information and generates loading reports based on the completion of the loading task. The report format is standardized, facilitating archiving and retrieval; the anomaly rejection module uses 3D vision recognition to remove unqualified containers. This module employs a 3D vision recognition algorithm to identify unqualified containers and control the rejection device to remove them from the conveyor line, ensuring that unqualified containers do not enter the loading process and improving loading quality. In this embodiment, 1 is the location of the cab, 2 is the location of the loading machine, 3 is the location of the telescopic belt, 4 is the location of the side push mechanism, and 5 is the location of the destacking. The cab control interface adopts a combination design of touch screen and physical knob, and supports manual / automatic mode switching. The scanning and accounting module records items including cargo ID, size, weight, loading time, operator, etc., and the report can be exported in PDF and Excel formats. The response time of the abnormal rejection device is no more than 1 second, and the cargo inventory data is automatically updated after removal.

[0052] By combining LiDAR scanning with AI models, precise positioning of vehicle outlines and cargo locations is achieved, replacing traditional manual measurement and judgment. The collaborative operation of column-mounted robots and telescopic conveyor belts automates the entire process of depalletizing and conveying, reducing human intervention. For example, a dual-column robot, guided by 3D vision, completes disordered depalletizing, and combined with the stable conveying of the telescopic conveyor belt, significantly improving material handling efficiency.

[0053] Based on the loading algorithm rules of "large first, small second; heavy first, light second", combined with the vehicle's three-dimensional coordinate data, the cargo is scientifically stacked, avoiding the randomness of manual placement; the loading position coordinates are extracted by AI algorithm and the mechanism is guided to ensure the stability of the cargo's center of gravity and reduce the risk of tipping over during transportation.

[0054] The system integrates scanning and accounting functions as well as anomaly rejection. During the transportation of goods, it uses 3D vision to detect appearance defects and positional deviations, automatically rejects unqualified products and records the data. The task dashboard module monitors the loading progress in real time. Combined with the reports generated by the scanning and accounting module, it achieves full traceability from order to loading, reducing the loss rate of goods.

[0055] The lidar's ability to generate 3D point cloud data can adapt to different vehicle models and on-site environments, ensuring vehicle positioning accuracy; the telescopic belt conveyor supports hydraulic lifting and motor drive, and can cope with terrain obstacles such as steps; the step-crossing function improves the system's adaptability to uneven sites through the collaboration of sensors and hydraulic mechanisms.

[0056] Manual forklifts only need to complete the material pallet transfer, while subsequent high-intensity operations such as destacking and loading are handled by robots and automated mechanisms; the cockpit module integrates multiple equipment control interfaces, simplifying the operation process, reducing manual intervention, and reducing reliance on operator experience.

[0057] The embodiments of this application have been described above with reference to the accompanying drawings. Unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other. This application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit and scope of the claims, and all of these forms are within the protection scope of this application.

Claims

1. An automated loading system, characterized in that, Includes the following steps: S1. Loading System Settings: Based on the order task, initialize the loading parameters, collect the length, width, and height dimensions of the goods, measure the weight of the goods, and formulate loading rules; S2. Manual forklift delivery: Forklift operators drive forklifts to transport pallets carrying materials to be loaded to the pre-set destacking station, and the forklifts travel along the route planned for the site. S3. Vehicle Contour Scanning and Positioning: Using LiDAR, the loading vehicle and its surrounding environment are scanned. The LiDAR emits a laser beam and receives the reflected light signal to generate three-dimensional point cloud data. Through a deep learning-trained AI model, the vehicle contour features are identified. This AI model is trained based on vehicle contour sample data and has the ability to extract features and recognize patterns. The intelligent algorithm outputs the three-dimensional coordinates of the vehicle contour based on the point cloud data and extracts the planar and three-dimensional coordinates of the loading position, providing data support for the subsequent guidance of the loading mechanism. S4.3D Vision Disordered Depalletizing: The LiDAR is used to scan the boxes to be depalletized to generate three-dimensional point cloud data. With the help of AI models, the center point and height coordinates of each box are determined, guiding the column robot to perform disordered grasping and depalletizing tasks. Multiple column robots are equipped to work together. The grasping mechanism adopts sponge suction cups and also includes a screw motor lifting mechanism, a cylinder-driven bottom gripper, and a fan-assisted sponge suction cup. S5. Telescopic Belt Conveyor: The destacking boxes are transported to the loading station via a telescopic belt. The telescopic belt has a certain bandwidth, height and extension length. During the hydraulic lifting process, the hydraulic system raises the height of the belt. The motor drives the belt to move forward and backward, ensuring that the conveying task can be completed under different terrain conditions. S6. Scanning, recording, and rejection of defects: During the conveying process, the scanning equipment is used to scan the appearance and position of the box. The scanning equipment uses image recognition to detect abnormal appearances such as damage and stains on the surface of the box. At the same time, the sensors monitor the positional deviation of the box on the belt. If an abnormality is detected, the system triggers an alarm mechanism and starts the rejection device to remove the unqualified box from the conveyor line. S7 3D Vision-Guided Automated Loading: Based on the vehicle's outline coordinates and loading algorithm planning, the loading mechanism is guided to arrange and push the boxes to the designated positions on the vehicle according to rules, completing the loading. The loading algorithm rules include prioritizing large to small, heavy to light, outside to inside, and order concentration. Horizontal boxes are prioritized for placement on both sides of the vehicle, while vertical boxes are placed in the middle. Simultaneously, information on unloaded products and loading progress are displayed in real time. The loading mechanism is equipped with 3D LiDAR to display the 3D outlines and coordinates of loaded and unloaded positions in real time. It automatically navigates to the loading position based on the coordinates provided by the intelligent system and completes loading according to the planned path. An independent vision system is equipped to perform appearance inspection on each box and issue an alarm.

2. The automatic loading system according to claim 1, characterized in that, The vehicle contour scanning and positioning in step S3 specifically includes: the lidar scans the loading site and the vehicle to generate three-dimensional point cloud data of corresponding resolution. The lidar has a horizontal field of view and a vertical field of view, which can cover the loading area. The AI ​​model, trained through deep learning, identifies the contour features of the vehicle body. The AI ​​model adopts a corresponding neural network architecture and is trained with vehicle contour sample data from different models and angles. Based on point cloud data, the intelligent algorithm outputs the three-dimensional coordinates of the vehicle body outline through corresponding algorithm steps, and extracts the planar and three-dimensional coordinates of the loading position to guide the loading mechanism.

3. The automatic loading system according to claim 1, characterized in that, In step S4, the 3D vision-based disordered depalletizing uses a column-type four-axis robot. Its parameters include: depalletizing speed: adapted to the corresponding operational requirements, multiple robots achieve alternating operations through a collaborative control algorithm; the gripping mechanism is a sponge suction cup that can support the corresponding weight of the box; depalletizing guidance is achieved through a 3D vision system, which is suitable for boxes of various sizes. The 3D vision system uses a camera and lens, combined with a specific image processing algorithm, to acquire the three-dimensional information of the box.

4. The automated loading system according to claim 3, characterized in that, The gripping mechanism also includes: a lead screw motor lifting mechanism: the lead screw motor has a certain stroke range and corresponding lifting accuracy, and can adjust the gripping position according to the height of the box; a cylinder-driven bottom support gripper: the cylinder provides corresponding clamping force to adapt to the lifting and clamping needs of boxes of different sizes; and a fan-assisted sponge suction cup: the fan generates corresponding negative pressure, and through the optimized airflow channel design, the negative pressure is evenly distributed on the surface of the sponge suction cup.

5. The automatic loading system according to claim 1, characterized in that, In step S5, the operating weight is within a certain range. According to different load conditions, the drive motor of the belt conveyor can automatically adjust the output power. It supports hydraulic lifting and motor drive to adapt to step crossing. The hydraulic lifting mechanism can lift the belt conveyor to the corresponding height within a certain time. The lifting height is within a certain range. The motor drive realizes the smooth movement of the belt conveyor through the corresponding motor and reducer, and can cross steps at a certain angle.

6. The automated loading system according to claim 1, characterized in that, The loading algorithm rules in step S7 include: large to small, heavy to light, outside to inside, and order concentration. In the actual loading process, the system prioritizes the goods according to their size, weight, and order information using a corresponding sorting algorithm to ensure a reasonable loading order and stable vehicle center of gravity. Horizontal boxes are placed on the sides of the vehicle first, and vertical boxes are placed in the middle. The system automatically plans the placement of goods by performing 3D modeling of the vehicle's internal space and matching calculations of the goods' dimensions. The system displays information on unloaded products and the loading progress in real time. Through the corresponding software system and display interface, the relevant information on unloaded products and the current loading progress are displayed in an intuitive chart format.

7. The automatic loading system according to claim 1, characterized in that, The system also includes a task dashboard module, used to obtain order tasks from the customer's warehouse management system. Through corresponding data interfaces and communication protocols, it connects with the customer's warehouse management system to obtain order information, including goods type, quantity, and shipping address. It generates loading plans through task optimization algorithms and hybrid loading algorithms. The task optimization algorithm rationally allocates and sorts order tasks based on corresponding optimization objectives and constraints. The hybrid loading algorithm comprehensively considers factors such as the size and weight of the goods and order attribution to formulate loading plans, improving loading efficiency and vehicle load rate. It monitors loading status and anomaly feedback in real time. Through data interaction with various modules of the system, it obtains information on loading progress and equipment operating status in real time. Once an anomaly is detected, it immediately reports it to the operator through corresponding alarm methods for timely handling.

8. The automated loading system according to claim 1, characterized in that, The specific implementation of the 3D vision-guided loading mechanism in step S7 is as follows: The loading mechanism is equipped with a 3D LiDAR, which displays the three-dimensional contours and coordinates of the loaded and unloaded positions in real time. The 3D LiDAR updates the data in real time at a certain frequency, and the generated three-dimensional contours have corresponding accuracy, providing data support for the path planning of the loading mechanism. The mechanism automatically navigates to the loading position according to the coordinates provided by the intelligent system and completes the loading according to the planned path. The intelligent system uses corresponding navigation and path planning algorithms, combined with vehicle contour coordinates and cargo placement rules, to plan the driving path for the loading mechanism. An independent vision system performs appearance inspection on each box and issues an alarm. The independent vision system uses corresponding cameras and image processing algorithms to complete the appearance inspection of the boxes, detect appearance defects, and trigger the alarm mechanism.

9. The automatic loading system according to claim 1, characterized in that, The system also includes a step-crossing process, specifically: The step height is detected by a passive sensor, which employs a specific model and principle, possessing appropriate detection accuracy to acquire step height information; a hydraulic lifting mechanism adjusts the height of the loading mechanism, automatically adjusting the lifting height based on the step height detected by the sensor using a corresponding control algorithm to ensure the loading mechanism can smoothly cross the step; a motor drives the front wheels to touch the ground and propels the mechanism across the step, adjusting its output torque and speed according to the action signal from the hydraulic lifting mechanism, driving the front wheels to touch the ground and propelling the mechanism across the step, ensuring a smooth crossing process; The hydraulic fork assists in climbing and descending stairs. During the process of crossing stairs, the hydraulic fork extends and retracts as needed to provide additional support and balance assistance.

10. The automated loading system according to any one of claims 1-9, characterized in that, The system also includes: a cockpit module that integrates a loading machine, a telescopic belt, a side push mechanism, and a multi-ball control interface; The scanning and accounting module records cargo information and generates loading reports. Through data interaction with the scanning equipment, the module automatically records relevant cargo information and generates loading reports based on the completion of loading tasks. The reports are formatted in a standardized way for easy archiving and retrieval. The anomaly rejection module uses 3D vision recognition to identify and remove unqualified containers. Employing a 3D vision recognition algorithm, the module can identify unqualified containers and control the rejection device to remove them from the conveyor line, ensuring that unqualified containers do not enter the loading process and improving loading quality.

Images