Automatic hoisting operation system and method for hook arm type garbage truck

Abstract

The invention discloses an automatic hoisting operation system and method for a hook arm type garbage truck, and relates to the technical field of intelligent control. Comprising an image detection module, an intelligent central computing gateway ICG, and an automatic parking area controller APDC, a chassis system, a vehicle control unit VCU, a power system, a vehicle body area controller BDC, an intelligent cabin area controller ICDC and a hoisting operation area controller which are connected to the intelligent central computing gateway ICG. According to the invention, the whole lifting operation process is intelligently and automatically completed by carrying out overall electronic and electrical architecture design on the hook arm type garbage truck, adding a vehicle-mounted monitoring camera, an automatic parking area controller and an intelligent lifting operation area controller and coordinating with an intelligent central computing gateway, a chassis power-assisted steering system, a braking system and the like; human intervention is not needed in the whole process.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent control, and specifically, to an automatic hoisting operation system and method for a hook-arm garbage truck. Background Art

[0002] There are many fixed domestic waste collection points in the city. The hook-arm garbage truck goes to these waste collection points, pulls out the detachable garbage compactor (the garbage inside may weigh several tons), hoists it onto the rear frame, and then transports it. For a traditional hook-arm garbage truck, a large amount of manual operation is required when hoisting a detachable garbage compactor. Generally, the driver parks the vehicle in place according to experience, manually starts and adjusts the hoisting device (there are remote control operation buttons and handles in the cockpit). The maintenance personnel at the waste point observe whether the vehicle hook and the lifting ring of the detachable garbage compactor are accurately aligned, and let the driver drive the vehicle and the hook-arm device to make various pose adjustments, and finally let the hook correctly hook the lifting ring so as to pull out the detachable garbage compactor and load it onto the vehicle. The maintenance personnel at the waste point observe whether the middle position between the garbage compactor and the two guide wheels on the rear frame is centered / deviated, and whether the box body is centered, stable and non-deviated during the hoisting process. If there is an inclination or deviation, the maintenance personnel at the waste point will immediately notify the driver to let the vehicle and the hook-arm make pose adjustments.

[0003] There are problems with this operation method:

[0004] 1) Limited by the lack of standardized parking spaces and hoisting operation areas at the waste collection points, the whole process of vehicle parking in place, vehicle pose adjustment, and hoisting operation requires an experienced driver to operate the vehicle and the hoisting device. Moreover, the maintenance personnel at the waste point have to closely observe whether the hook and the lifting ring are aligned and hooked at all times. During the process of pulling the garbage compactor onto the rear frame slide rail, whether the box body is tilted or collides with the vehicle body, and whether the bottom of the garbage compactor is completely attached to the rear frame slide rail after the garbage compactor is loaded onto the vehicle. The maintenance personnel at the waste point also have to communicate with the driver in real time to let the driver continuously adjust the pose of the vehicle and the hoisting device. This results in very low operation efficiency and low success rate.

[0005] 2) After the detachable garbage compactor is hoisted onto the vehicle, the driver may be negligent in operation and fail to lock it to the rear frame in time, or the rear outriggers are not retracted. Therefore, the maintenance personnel at the waste point have to help the driver check the locking and retraction status to avoid dangers such as the box body falling. Summary of the Invention

[0006] The purpose of the present invention is to provide an automatic hoisting operation system and method for a hook-arm garbage truck, which is used to solve the problems in the prior art that manual adjustment of the vehicle pose, communication and confirmation of the hoisting device pose between the driver and the maintenance personnel, and inspection of the locking status after hoisting are required, resulting in cumbersome operation and low efficiency.

[0007] The present invention solves the above problems through the following technical solutions:

[0008] An automatic hoisting operation system for a hook-arm garbage truck, comprising an image detection module, an intelligent central computing gateway ICG, and an automatic parking domain controller APDC, a chassis system, a vehicle control unit VCU, a power system, a body domain controller BDC, an intelligent cockpit domain controller ICDC, and a hoisting operation domain controller connected to the intelligent central computing gateway ICG, wherein:

[0009] The image detection module is used to obtain the video data of the line parking space of the hook-arm garbage truck and the video data of the surrounding environment of the hoisting operation, and input them into the automatic parking domain controller APDC; it is also used to collect the video data of the hook of the hook-arm garbage truck and the lifting ring of the garbage compression box, and input them into the hoisting operation domain controller LODC;

[0010] The intelligent central computing gateway ICG is used to implement protocol stack encapsulation and data routing and forwarding among the automatic parking domain controller APDC, the chassis system, the vehicle control unit VCU, the power system, the body domain controller BDC, the intelligent cockpit domain controller ICDC, and the hoisting operation domain controller LODC;

[0011] The automatic parking domain controller APDC is used to identify the images input by the image detection module, communicate with the vehicle control unit VCU, and the chassis system realizes automatic parking by the vehicle control unit VCU, or the automatic parking domain controller APDC directly issues control instructions to the chassis system to realize automatic parking; the automatic parking domain controller APDC is built-in with an AI deep learning neural network to train and learn a large number of images to realize the detection, judgment, and control of the automatic parking process;

[0012] The power system is used to provide transmission control, speed change control, and provide power;

[0013] The vehicle control unit VCU is used to control the chassis system and the power system;

[0014] The body domain controller BDC is used to check and give early warnings about the tire pressure status of the vehicle;

[0015] The lifting operation domain controller LODC is communicatively connected to the image detection module and is used to determine whether the hook is aligned with and hooked to the sling and whether the hook is secure on the sling based on the input video data. The lifting operation domain controller LODC has an AI deep learning neural network built-in, which trains and learns from a large amount of images to realize the detection, judgment, and control of the automatic lifting operation process. It is also connected to the detection module and the electric control module through a private CAN. The detection module is used to detect the hydraulic oil volume, hydraulic pipeline, and quick connectors of the hook-arm type garbage truck, and the electric control module is used to control the power take-off, rear outriggers, main boom lock hook, and hook-arm telescoping of the hook-arm type garbage truck to realize the status detection before the lifting operation and start the lifting operation.

[0016] Further, the image detection module includes a first wide-angle camera arranged at the hook of the hook-arm type garbage truck and AVM surround cameras respectively arranged at the front, rear, left, and right sides of the hook-arm type garbage truck; it also includes a second wide-angle camera arranged at the top frame of the garbage house door, a third wide-angle camera on the left side in the middle of the line parking space, and a fourth wide-angle camera on the right side in the middle of the line parking space. The AVM surround cameras are used to obtain video data of the surrounding environment of the lifting operation, the first wide-angle camera is used to collect video data of the hook and the sling of the garbage compactor, the line parking space is a rectangular frame, the side of the rectangular frame close to the garbage house is N meters away from the garbage house, the width of the rectangular frame = the inner width of the garbage house exit, and the length of the rectangular frame ≥ the length of the hook-arm type garbage truck.

[0017] Further, it also includes a field multi-access edge computing MEC system and an intelligent terminal. The field multi-access edge computing MEC system is communicatively connected to the second wide-angle camera, the third wide-angle camera, and the fourth wide-angle camera, and is used to process and judge the input video data of the hook-arm type garbage truck line parking space, and communicate with the automatic parking domain controller APDC and the lifting operation domain controller LODC through the intelligent central computing gateway ICG;

[0018] The intelligent terminal communicates with the intelligent central computing gateway ICG through Bluetooth or WiFi and is used to start the lifting operation of the hook-arm type garbage truck with one key.

[0019] Further, the field multi-access edge computing MEC system includes:

[0020] An image acquisition and preprocessing module, which is used to parallel-process the video data collected by the second wide-angle camera, the third wide-angle camera, and the fourth wide-angle camera, and obtain high-quality image data frames through one or more algorithms such as noise suppression, data standardization, color space transformation, and geometric distortion correction;

[0021] An environment perception and implementation reasoning module, which is used to perform target recognition and tracking on the input high-quality image data frames;

[0022] A logic calculation and judgment warning module, which is used to judge whether the hoisting operation area is safe according to the target recognition and tracking results, and during the hoisting process, judge:

[0023] Whether the garbage compactor maintains balance and stability and whether there is central deviation;

[0024] When the garbage compactor is completely hoisted onto the rear frame, whether its bottom is completely fitted with the rear frame slide rail;

[0025] Whether the vehicle's rear outriggers are deployed when the hoisting operation starts and whether the vehicle's rear outriggers are restored and retracted when the hoisting operation is completed;

[0026] The logic calculation and judgment warning module needs to use an AI deep learning neural network to train and learn a large amount of images to realize the detection, judgment and control of the safety of the hoisting operation area and the hoisting operation process;

[0027] A V2I communication module, which is used to actively broadcast the above judgment results to the intelligent central computing gateway ICG.

[0028] Furthermore, the chassis system includes an electronic parking brake EPB, an electro-mechanical brake EMB and an electric power steering system EPS. The automatic parking domain controller APDC is communicatively connected to the chassis system through the chassis domain CAN, and the automatic parking domain controller APDC is also communicatively connected to the electro-mechanical brake EMB and the electric power steering system EPS through a private CAN.

[0029] Furthermore, the electro-mechanical brake EMB adopts a wire-controlled brake dual-redundancy setting.

[0030] Furthermore, the intelligent cockpit domain controller ICDC is connected to a central control screen CSD. The automatic parking domain controller APDC forwards the spliced image of the video data input by the image detection module to the intelligent cockpit domain controller ICDC through in-vehicle Ethernet and the intelligent central computing gateway ICG, and the intelligent cockpit domain controller ICDC displays it in real time through the central control screen CSD. So that the driver can set parameters and monitor the real-time status of the hoisting operation process in the vehicle.

[0031] Furthermore, the power system includes a power domain module group, a rear motor controller integrated rotation MCRT and a rear transmission control unit TCUR. The rear motor controller integrated rotation MCRT and the rear transmission control unit TCUR are communicatively connected to the vehicle control unit VCU and the intelligent central computing gateway ICG through the power domain CAN-1; the power domain module group is communicatively connected to the vehicle control unit VCU and the intelligent central computing gateway ICG through the power domain CAN-2.

[0032] Further, the body domain controller BDC is communicatively connected to the main tire pressure monitoring module TPMM and the slave tire pressure monitoring module TPMS through the body domain CAN.

[0033] Further, the detection module includes a hydraulic oil quantity detection module, a hydraulic pipeline detection module, and a quick connector detection module; the electronic control module includes a power take-off electronic control module, a rear outrigger electronic control module, a main boom hook electronic control, and a hook arm telescopic electronic control module.

[0034] The environmental perception and recognition detection of intelligent lifting operations are realized by training and reasoning using a deep learning model.

[0035] (1) The deep learning model (using the AVM surround-view camera) built in the vehicle-end automatic parking domain controller APDC needs to identify and detect scenarios including: line parking spaces, pedestrians in the lifting operation area, and other dynamic obstacles.

[0036] (2) The deep learning model (using the wide-angle camera at the vehicle hook) built in the vehicle-end lifting operation domain controller LODC needs to identify and detect scenarios including: the vehicle hook aligning with and hooking the lifting ring of the garbage compactor and whether it is firm.

[0037] (3) The deep learning model built in the field-side MEC (using the camera at the top frame of the garbage house door and the cameras on the left and right sides in the middle of the field-side line parking space) needs to assist the vehicle-end in identifying and detecting scenarios including: pedestrians in the lifting operation area, and other dynamic obstacles; among them: using the camera at the top frame of the garbage house door, the scenarios that need to be identified and detected include: during the lifting process, the middle position alignment / deviation of the garbage compactor and the two guide wheels on the rear frame; when the lifting operation starts / ends, the vehicle's rear outriggers unfold / restore and retract; using the cameras on the left and right sides in the middle of the field-side line parking space, the scenarios that need to be identified and detected include: during the process of loading the box body, the bottom of the garbage compactor is completely attached to the rear frame slide rail; when the lifting operation starts / ends, the vehicle's rear outriggers unfold / restore and retract.

[0038] A hook-arm type garbage truck automatic lifting operation method implemented by using the hook-arm type garbage truck automatic lifting operation system described above includes:

[0039] Step S1, the automatic parking process, includes:

[0040] A1. Start automatic parking.

[0041] A2. The automatic parking domain controller APDC performs line parking space detection and recognition, surrounding environment perception of the operation area, and obstacle identification and positioning based on the video data input by the image detection unit.

[0042] A3. The automatic parking domain controller APDC generates a parking path, starts parking, and performs collision detection, trajectory tracking control, and exception handling.

[0043] The A4 hook-arm garbage truck stops at the central position of the on-street parking space, and the automatic parking is completed;

[0044] Step S2. Start the status check before the hoisting operation, including:

[0045] (1) The hoisting operation domain controller LODC determines whether the status of the hydraulic oil quantity, hydraulic pipeline and quick connector reported by the detection module is normal;

[0046] (2) The hoisting operation domain controller LODC determines whether the status reported by the electric control module is normal;

[0047] (3) The vehicle body domain controller BDC determines whether the tire pressure data is normal;

[0048] (4) The automatic parking domain controller APDC confirms whether the status of the electric power steering system EPS and the electro-mechanical brake EMB of the chassis system is normal;

[0049] (5) The automatic parking domain controller APDC confirms that there are no personnel or obstacles in the hoisting operation area;

[0050] If any one of (1)-(4) is abnormal or there are personnel or obstacles in the hoisting operation area, an alarm is issued; otherwise, the status check before the hoisting operation is completed;

[0051] Step S3. Hoisting operation process, including:

[0052] B1. Start the unmanned hoisting operation;

[0053] B2. The vehicle control unit VCU controls the electronic parking brake EPB of the chassis system to brake the vehicle handbrake; the vehicle control unit VCU controls the rear transmission control unit TCUR of the power system to put the vehicle in neutral gear;

[0054] B3. The hoisting operation domain controller LODC controls the rear outrigger electric control module in the electric control module to lower the rear outriggers of the vehicle;

[0055] B4. The vehicle control unit VCU controls the power supply and power domain module group of the built-in hoisting operation hydraulic system to start the vehicle engine and turn on the hydraulic system power supply, and the vehicle control unit VCU controls the MCRT to provide power;

[0056] B5. The vehicle control unit VCU controls the rear transmission control unit TCUR to drive the synchronizer, and the hoisting operation domain controller LODC controls the power take-off electric control module in the electric control module, and the two cooperate to complete the power take-off action of the hydraulic pump;

[0057] B6. The main boom hook electronic control module in the electronic control module is controlled by the lifting operation domain controller LODC to keep the main boom hook in a fully released state;

[0058] B7. The boom telescoping electronic control module in the electronic control module is controlled by the lifting operation domain controller LODC to continuously extend the main boom cylinder until the auxiliary boom hook flips to a position slightly lower than the hook center of the garbage compactor lifting ring;

[0059] B8. The synchronizer is driven by the rear transmission control unit TCUR controlled by the vehicle control unit VCU, and the power take-off electronic control module in the electronic control module is controlled by the lifting operation domain controller LODC. The two cooperate to close the hydraulic pump power take-off;

[0060] B9. The synchronizer is driven by the rear transmission control unit TCUR controlled by the vehicle control unit VCU to shift the vehicle into reverse gear. The integrated rotation MCRT of the rear motor controller of the power system is controlled by the vehicle control unit VCU to reverse the vehicle backward to the position where the hook is at the hook center of the garbage compactor lifting ring; The connection state between the hook and the lifting ring is monitored in real time, and after the lifting operation domain controller LODC fully confirms that it is firmly hooked, the vehicle control unit VCU controls the EMB to brake the vehicle;

[0061] B10. Perform hydraulic power take-off again: The synchronizer is driven by the rear transmission control unit TCUR controlled by the vehicle control unit VCU, and the power take-off electronic control module in the electronic control module is controlled by the lifting operation domain controller LODC. The two cooperate to complete the hydraulic pump power take-off action;

[0062] B11. The boom telescoping electronic control module is controlled by the lifting operation domain controller LODC to continuously contract the main boom cylinder, tow the garbage compactor to the slide rail on the rear frame, and continuously lift it until the box is completely suspended and parallel to the frame until the box is completely pulled onto the rear frame;

[0063] B12. Monitor the centering of the middle position between the box and the two guide wheels on the rear frame. The lifting operation domain controller LODC performs real-time analysis and judgment. If it is determined that the box deviates from the centering position, an early warning message will be immediately sent to notify the vehicle to adjust the position and angle;

[0064] B13. During the process of loading the box, monitor whether the bottom of the garbage compactor is fully attached to the slide rail of the rear frame;

[0065] B14. Receive the message of the attachment state between the box and the slide rail of the rear frame. If it is fully attached, the main boom hook electronic control module is controlled by the lifting operation domain controller LODC to keep the main boom hook in a fully locked state and enter the next step; If it is prompted that it cannot be attached, the vehicle needs to adjust the position and angle. After receiving the message that it cannot be fully attached multiple times, the vehicle will suspend the lifting operation;

[0066] B15. After the vehicle hoisting operation is completed normally, the vehicle control unit VCU controls the rear transmission control unit TCUR to drive the synchronizer, and the hoisting operation domain controller LODC controls the power take-off electronic control module. The two cooperate to close the hydraulic pump power take-off and release the hydraulic pressure;

[0067] B16. The vehicle control unit VCU controls the power domain module group to turn off the power supply of the hoisting operation hydraulic system; the hoisting operation domain controller LODC controls the rear outrigger electronic control module to restore and retract the rear outriggers of the vehicle, and the hoisting operation is completed.

[0068] A method for automatic hoisting operation of a hook-arm type garbage truck implemented by using the described automatic hoisting operation system for a hook-arm type garbage truck. When the system further includes a field multi-access edge computing MEC system and an intelligent terminal, the field multi-access edge computing MEC system is communicatively connected to the image detection module, and is used for processing and judging the input video data of the hook-arm type garbage truck's line parking space, and communicating with the automatic parking domain controller APDC and the hoisting operation domain controller LODC through the intelligent central computing gateway ICG; the intelligent terminal communicates with the intelligent central computing gateway ICG through Bluetooth or WiFi, and is used for starting the hoisting operation of the hook-arm type garbage truck with one key; the method is as follows:

[0069] Step S1. The automatic parking process includes:

[0070] A1. The intelligent terminal establishes a communication connection with the hook-arm type garbage truck and starts automatic parking with one key on the intelligent terminal;

[0071] A2. The automatic parking domain controller APDC performs line parking space detection and recognition, surrounding environment perception of the operation area, and obstacle recognition and positioning according to the video data input by the image detection unit;

[0072] A3. The automatic parking domain controller APDC generates a parking path, starts parking, and performs collision detection, trajectory tracking control and exception handling;

[0073] A4. The hook-arm type garbage truck stops at the central position of the line parking space, and the automatic parking is completed;

[0074] Step S2. The intelligent terminal starts the status check function before the hoisting operation with one key, and the hook-arm type garbage truck establishes a V2I short-range wireless connection with the field multi-access edge computing MEC; including:

[0075] (1) The hoisting operation domain controller LODC judges whether the status of the hydraulic oil quantity, hydraulic pipeline and quick connector reported by the detection module is normal;

[0076] (2) The hoisting operation domain controller LODC judges whether the status reported by the electronic control module is normal;

[0077] (3) The Body Domain Controller (BDC) determines whether the tire pressure data is normal;

[0078] (4) The Automatic Parking Domain Controller (APDC) confirms whether the states of the Electric Power Steering System (EPS) and the Electro-Mechanical Brake (EMB) of the chassis system are normal;

[0079] (5) The Automatic Parking Domain Controller (APDC) confirms that there are no personnel or obstacles in the hoisting operation area;

[0080] (6) The Field Edge Computing (MEC) confirms that there are no personnel or obstacles in the hoisting operation area;

[0081] If any of (1)-(4) is abnormal or there are personnel or obstacles in the hoisting operation area, an alarm is issued and sent to the intelligent terminal; otherwise, the status detection before hoisting operation is completed;

[0082] Step S3, the hoisting operation process, includes:

[0083] B1. The intelligent terminal starts the unmanned hoisting operation with one key, and the operation instruction is forwarded by the ICG of the hook-arm garbage truck to the Vehicle Control Unit (VCU) and the Hoisting Operation Domain Controller (LODC);

[0084] B2. The Vehicle Control Unit (VCU) controls the Electronic Parking Brake (EPB) of the chassis system to brake the vehicle's handbrake; the Vehicle Control Unit (VCU) controls the Rear Transmission Control Unit (TCUR) of the power system to put the vehicle in neutral gear;

[0085] B3. The Hoisting Operation Domain Controller (LODC) controls the rear outrigger electric control module in the electric control module to lower the vehicle's rear outriggers;

[0086] B4. The Vehicle Control Unit (VCU) controls the power supply and power domain module group of the built-in hoisting operation hydraulic system to start the vehicle engine and turn on the power of the hydraulic system, and the Vehicle Control Unit (VCU) controls the MCRT to provide power;

[0087] B5. The Vehicle Control Unit (VCU) controls the rear transmission control unit (TCUR) to drive the synchronizer, and the Hoisting Operation Domain Controller (LODC) controls the power take-off electric control module in the electric control module. The two cooperate to complete the power take-off action of the hydraulic pump;

[0088] B6. The Hoisting Operation Domain Controller (LODC) controls the main boom locking hook electric control module in the electric control module to control the main boom locking hook to be in a fully released state;

[0089] B7. The Hoisting Operation Domain Controller (LODC) controls the hook-arm telescopic electric control module in the electric control module to continuously extend the main boom cylinder until the sub-boom hook flips to a position slightly lower than the hook center of the garbage compactor;

[0090] B8. The vehicle control unit (VCU) controls the rear transmission control unit (TCUR) to drive the synchronizer, and the lifting operation domain controller (LODC) controls the power take-off electronic control module in the electronic control module. The two cooperate to close the hydraulic pump power take-off.

[0091] B9. The vehicle control unit (VCU) controls the rear transmission control unit (TCUR) to drive the synchronizer, shifts the vehicle into reverse gear, and the vehicle control unit (VCU) controls the integrated rotation of the rear motor controller of the power system, MCRT, to reverse the vehicle backward to the hook of the hook and the eye of the lifting ring of the garbage compactor; the connection state between the hook and the lifting ring is monitored in real time, and after the lifting operation domain controller (LODC) fully confirms that it is firmly hooked, the vehicle control unit (VCU) controls the EMB to brake the vehicle.

[0092] B10. Perform hydraulic power take-off again: The vehicle control unit (VCU) controls the rear transmission control unit (TCUR) to drive the synchronizer, and the lifting operation domain controller (LODC) controls the power take-off electronic control module in the electronic control module. The two cooperate to complete the hydraulic pump power take-off action.

[0093] B11. The lifting operation domain controller (LODC) controls the hook arm telescopic electronic control module to continuously contract the main arm cylinder, tow the garbage compactor to the slide rail of the rear frame, and continuously lift it until the box is completely suspended and parallel to the frame until the box is completely pulled onto the rear frame.

[0094] B12. Monitor the centering of the middle position between the box and the two guide wheels on the rear frame. The field-side multi-access edge computing (MEC) performs real-time analysis and judgment. If it is determined that the box deviates from the centering position, an early warning message will be sent immediately to notify the vehicle to adjust its position and angle.

[0095] B13. During the process of loading the box onto the vehicle, monitor whether the bottom of the garbage compactor is completely attached to the slide rail of the rear frame. The field-side multi-access edge computing (MEC) performs real-time analysis and judgment and sends the attachment message to the lifting operation domain controller (LODC).

[0096] B14. After receiving the attachment status message of the box and the slide rail of the rear frame, if it is completely attached, the lifting operation domain controller (LODC) controls the main arm locking hook electronic control module to make the main arm locking hook in a fully locked state and proceed to the next step; if it is prompted that it cannot be attached, the vehicle needs to adjust its position and angle. After receiving the message that it cannot be completely attached multiple times, the vehicle will suspend the lifting operation.

[0097] B15. After the vehicle's lifting operation is completed normally, the vehicle control unit (VCU) controls the rear transmission control unit (TCUR) to drive the synchronizer, and the lifting operation domain controller (LODC) controls the power take-off electronic control module. The two cooperate to close the hydraulic pump power take-off and release the hydraulic pressure.

[0098] Replace B16 with: The vehicle control unit (VCU) controls the power domain module group to turn off the power supply of the hoisting operation hydraulic system; the hoisting operation domain controller (LODC) controls the rear outrigger electronic control module to retract the vehicle's rear outriggers and send a message to the intelligent terminal, indicating that the hoisting operation is completed.

[0099] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0100] (1) Through the overall electronic and electrical architecture design of the hook-arm garbage truck, the present invention adds on-vehicle monitoring cameras, an automatic parking domain controller, and an intelligent hoisting operation domain controller, which work in coordination with the intelligent central computing gateway, the chassis power steering system, the braking system, etc., to complete the entire hoisting operation process intelligently and automatically without any human intervention.

[0101] (2) By adding cameras for detection, the present invention confirms whether the hoisting operation area is safe, that is, there are no pedestrians and other dynamic moving obstacles (vehicle-end AVM surround-view cameras, field-end wide-angle cameras (the second wide-angle camera, the third wide-angle camera, and the fourth wide-angle camera)). The hook of the hook-arm truck aligns with and hooks the lifting ring of the garbage compactor (the first wide-angle camera). During the hoisting process, the hook is firm (the first wide-angle camera), and the garbage compactor maintains balance and stability (the field-end wide-angle camera). The bottom of the garbage compactor is fully attached to the rear frame slide rail (the field-end wide-angle camera), and the vehicle's rear outriggers are deployed / retracted (the field-end wide-angle camera). These key points determined by visual detection are trained through image training of a deep learning neural network and are respectively placed in the vehicle-end controller and the field-end controller. During the hoisting operation process, the field-end controller assists the vehicle-end controller in real-time inference to identify and detect the correctness of the operation, realizing unmanned hoisting operations.

[0102] (3) The present invention establishes a communication connection among the intelligent terminal, the hook-arm garbage truck, and the field-edge MEC to achieve one-key triggering of automatic hoisting operations and automatic sending of warnings to the intelligent terminal.

[0103] (4) Through the SDM stable diffusion large model and technologies such as image-to-image and text-to-image, the present invention expands the high-quality data set of the training samples to solve the problem of insufficient image learning samples, enabling the intelligent hoisting recognition and detection model to be quickly trained, effectively improving the generalization ability and inference accuracy. Additionally, by fine-tuning the parameters of the YOLO neural network, the requirements for the industrial detection level of unmanned hoisting operations are met.

[0104] (5) The entire process of hoisting the detachable garbage bin can be completed with high accuracy and efficiency through the coordination of components such as sensors, controllers, and actuators on the vehicle, as well as field sensors and controllers, avoiding the traditional inefficient manual method (garbage point maintenance personnel observe whether the vehicle hook and the lifting ring of the detachable garbage bin are accurately aligned, and let the driver drive the vehicle and the hook arm device to make various pose adjustments); the entire hoisting operation process is fully unmanned, improving efficiency and avoiding the risk of personal injury in the operation area. Brief Description of the Drawings

[0105] Figure 1 It is the electronic and electrical architecture diagram of the hook-arm type garbage truck of the present invention;

[0106] Figure 2 It is the communication interaction schematic diagram of the field multi-access edge computing MEC system with the hook-arm type garbage truck and the mobile phone APP;

[0107] Figure 3 It is the automatic parking flowchart of the hook-arm type garbage truck;

[0108] Figure 4 It is the status inspection flowchart before the hoisting operation;

[0109] Figure 5 It is the hoisting operation flowchart;

[0110] Figure 6 It is the schematic diagram of expanding and generating a high-quality data set based on the SD stable diffusion AI large model;

[0111] Figure 7 It is the schematic diagram of fine-tuning training based on the YOLO-v11 model;

[0112] Figure 8 It is for Figure 7 The structural diagram of the C2PSA module composition in

[0113] Figure 9 It is for Figure 7 The structural diagram of the Detect module composition in Detailed Embodiment

[0114] The present invention will be further described in detail below in conjunction with embodiments, but the embodiments of the present invention are not limited thereto.

[0115] Before introducing the embodiments of the present invention, the following abbreviations involved in this article are first explained as follows:

[0116] 5G-V2X: 5G Vehicle to Everything, communication between vehicles and all things;

[0117] AB: Attention Block, an attention block;

[0118] AHD: Analog High Definition, an analog high-definition video transmission technology;

[0119] APDC: Auto Parking Domain Controller, an automatic parking domain controller;

[0120] ASN.1: Abstract Syntax Notation One, an ISO / ITU-T standard for abstract syntax notation 1;

[0121] AVM: Around View Monitor, a panoramic imaging system;

[0122] AWS: Amazon Web Service, Amazon Web Services;

[0123] BDC: Body Controller, a body domain controller;

[0124] BiFPN-XL: An extended version of BiFPN (Bidirectional Feature Pyramid Network) designed to further optimize the feature fusion process;

[0125] C2PSA: Combines the CSP (Cross Stage Partial) structure and the PSA (Pyramid SqueezeAttention) attention mechanism to enhance multi-scale feature extraction capabilities;

[0126] CAN: Controller Area Network, a controller area network;

[0127] CAN-FD: Controller Area Network Flexible Data-Rate, a controller area network with flexible data transfer rate;

[0128] CIoU: Combined Intersection over Union, a combined intersection over union;

[0129] CLIP: Constrastive Language-Image Pre-training, a pre-training method or model based on contrastive text-image pairs;

[0130] CLSLoss: Classification Loss, which is an important loss term in the object detection model and is used to measure the accuracy of the model's prediction for each object category;

[0131] CNN: Convolutional Neural Network, a convolutional neural network;

[0132] Conv2d: Convolution of 2 Dimensions, a two-dimensional convolution;

[0133] CSD: Central Screen Display, the central control screen;

[0134] CSP: Cross Stage Partial Network, a cross-stage partial network;

[0135] DWConv: Depthwise Separable Convolution, a depthwise separable convolution;

[0136] EC2: Elastic Compute Cloud, an elastic compute cloud;

[0137] EEA: Electrical / Electronic Architecture, an electrical / electronic architecture;

[0138] EMB: Electro-Mechanical Brake, an electro-mechanical brake;

[0139] EPB: Electrical Park Brake, an electronic parking brake;

[0140] EPS: Electrical Power Steering, an electric power steering system;

[0141] Ethernet: In-vehicle Ethernet;

[0142] FP16: Half-Precision Floating-Point, a half-precision floating-point number;

[0143] FPS: Frames Per Second, the number of frames per second;

[0144] GDPR: General Data Protection Regulation, the General Data Protection Regulation;

[0145] GPU: Graphics Processing Unit, a graphics processing unit;

[0146] GSC: GroupNorm + Swish + Conv combined component, group normalization, Swish smooth activation function, convolutional layer

[0147] GT: Ground Truth, ground truth;

[0148] HIPAA: Health Insurance Portability and Accountability Act, the Health Insurance Portability and Accountability Act;

[0149] HSV: Hue / Saturation / Value, hue / saturation / brightness, a non - linear model for representing colors in a color space

[0150] IAM: Identity and Access Management, identity and access management;

[0151] ICDC: Intelligent Cockpit Domain Controller, intelligent cockpit domain controller;

[0152] ICG: Intelligent Central Gateway, intelligent central gateway;

[0153] INF2: AWS Inferentia2, the first inference - optimized instance in Amazon EC2;

[0154] IPM: Inverse Perspective Mapping, inverse perspective transformation;

[0155] ITU - T: International Telecommunication Union - Telecommunication Standardization Sector, the International Telecommunication Union Telecommunication Standardization Sector;

[0156] KMS: Key Management Service, key management service

[0157] LODC: Lifting Operation Domain Controller, lifting operation domain controller;

[0158] LN: Layer Normalization, layer normalization;

[0159] LVDS: Low-Voltage Differential Signaling, low voltage differential signal;

[0160] MCRT: Motor Control Unit Rear Transmission, rear motor controller integrated transmission;

[0161] MEC: Multi-access Edge Computing;

[0162] MLP: Multi-Layer Perceptron, multi-layer perceptron;

[0163] MPC: Model Predictive Control, model predictive control;

[0164] P4D: AWS P4D is an instance based on the AMD EPYC processor, with specific models such as P4d.24xlarge;

[0165] P5: AWS P5 is a high-performance computing instance based on NVIDIA H100 Tensor Core GPU;

[0166] PSA Block: Pyramid Squeeze Attention Block, pyramid squeeze attention block;

[0167] RB: ResNet Block, residual network block;

[0168] ReLU: Rectified Linear Unit, linear rectification function;

[0169] RSM: Roadside Safety Message, roadside safety message;

[0170] ResNet:Residual Network,Residual Network;

[0171] RSU: Road Side Unit, road side unit;

[0172] S3: Amazon Simple Storage Service, Amazon Simple Storage Service;

[0173] SAE: Society of Automotive Engineers;

[0174] SDM: Stable Diffusion Model, a stable diffusion model;

[0175] SILU: Sigmoid Linear Unit, also known as the Swish activation function, is a non - linear function that multiplies the input by its own sigmoid value;

[0176] Swin Transformer: A computer vision model based on the transformer architecture, aiming to solve the problems of high computational cost and low efficiency in traditional transformers when applied to image processing;

[0177] SW - MSA: Shifted Window Multi - Head Self Attention, window - shifted multi - head self - attention;

[0178] TBox: Telematic Box, an intelligent in - vehicle terminal in the vehicle - to - everything system;

[0179] TCUR: Transmission Control Unit Rear, the rear transmission control unit;

[0180] TMC: Thermoplastic Composite Material, self - lubricating oil - containing nylon;

[0181] TPMM: Tire Pressure Monitor System Main, the main tire pressure detection module;

[0182] TPMS: Tire Pressure Monitor System Slave, the slave tire pressure detection module;

[0183] Transformer: A transformer, a sequence model based on the attention mechanism;

[0184] U - NET: A U - shaped architecture based on convolutional neural networks for image segmentation tasks;

[0185] V2I: Vehicle to Infrastructure, communication technology between vehicles and infrastructure;

[0186] VAE: Variational Autoencoder, variational autoencoder;

[0187] VCU: Vehicle Control Unit, vehicle control unit;

[0188] W-MSA: Window-Based Multi-Head Self Attention

[0189] YOLO: You Only Look Once, an end-to-end object detection algorithm that can complete recognition with a single scan.

[0190] Example 1:

[0191] An automatic hoisting operation system for a hook-arm garbage truck, comprising:

[0192] An image detection module, which is composed of newly added cameras, including:

[0193] 1 camera deployed on the top frame of the garbage house door at the garbage collection point, used to monitor in real time whether the hook is firm on the sling ring, the balance and stability of the garbage compactor, and whether the rear outriggers of the vehicle are deployed / retracted during the hoisting operation of the hook-arm garbage truck;

[0194] 1 camera is respectively deployed on the left and right sides in the middle of the line parking space. When the garbage compactor body gets on the vehicle, it is used to monitor in real time whether the bottom of the garbage compactor is fully attached to the rear frame slide rail and whether the rear outriggers of the vehicle are deployed / retracted;

[0195] The above 3 cameras also assist in real-time perception and monitoring of the vehicle's surrounding environment during the hoisting operation to ensure that there are no pedestrians or dangerous obstacles within a radius of 5 meters of the operation area; and detect: the garbage compactor maintains balance and stability during the hoisting process; the bottom of the garbage compactor is fully attached to the rear frame slide rail; the rear outriggers of the vehicle are deployed / retracted;

[0196] The hook-arm garbage truck is provided with 4 AVM surround-view cameras, which are respectively set at the front, rear, left and right of the vehicle, used to detect whether the hoisting operation area is safe, that is, there are no pedestrians and other dynamic moving obstacles;

[0197] 1 wide-angle camera is set at the hook of the hook-arm garbage truck, used to detect that the hook of the hook-arm truck is aligned with and hooked to the sling ring of the garbage compactor, and the hook is firm during the hoisting process;

[0198] The setting of the line parking space is as follows: The dedicated line parking space for the hook-arm garbage truck starts to be set at a position 2.5 meters away from the door of the garbage collection point. The width of the line parking space (the two side boundary lines) is aligned with the inner side of the garbage house exit, and the length of the line parking space only needs to cover the length of the hook-arm garbage truck.

[0199] Such as Figure 1As shown in the figure, the existing hook-arm garbage truck is redesigned with an electronic and electrical architecture, mainly including ICG, APDC, chassis system, VCU, power and module group, BDC, ICDC, LODC, where:

[0200] (1) The intelligent central computing gateway ICG integrates the in-vehicle gateway, Tbox, and 5G-V2X functions, supports software protocol stack encapsulation and data routing and forwarding of in-vehicle Ethernet, CAN, CAN-FD, wireless cellular network, V2X communication (the V2I capability is mainly reflected in this invention), Bluetooth, WIFI, etc.;

[0201] (2) The automatic parking domain controller APDC uses LVDS to connect 4-channel AVM surround-view cameras to realize the detection and recognition of the dedicated line parking spaces for hook-arm garbage trucks at urban fixed garbage collection points, as well as the perception and recognition of the surrounding environment (pedestrians, obstacles, etc.) during the hoisting operation. APDC supports connection methods such as in-vehicle Ethernet, CAN-FD, and CAN for external associated systems. The deep learning model (using AVM surround-view fisheye cameras) built into the vehicle-end APDC needs to identify and detect scenarios including: line parking spaces, pedestrians within the hoisting operation area, and other dynamic obstacles;

[0202] As an optional method, if someone is in the vehicle during the hoisting operation, APDC can forward the image after AVM video stitching to the intelligent cockpit domain controller ICDC through the in-vehicle Ethernet via the ICG gateway for real-time display on the central control screen CSD;

[0203] (3) There are electronic parking brake EPB, electro-mechanical brake dual-redundancy system EMB, and electric power steering system EPS on the chassis domain CAN bus;

[0204] (4) Since the entire process of automatic parking and hoisting operation is unattended, a certain degree of redundancy setting needs to be considered for vehicle-end key systems to ensure functional safety. Specifically, it includes:

[0205] Communication network dual redundancy: The control instructions of APDC for the chassis system can be forwarded to each module of the chassis system through ICG. In addition, APDC has a dedicated private CAN network that directly connects to the EMC and EPS system modules of the chassis system. If the network connected to ICG is abnormal, APDC can directly send instructions to the chassis system for control through the dedicated private CAN network;

[0206] EMB: EMB requires dual redundancy to ensure that once vehicle EMB#A fails when someone is outside the vehicle, EMB#B can actively take over and brake and stop the vehicle normally;

[0207] (5) The power domain has two CAN buses. On the power domain CAN-1, there are the rear motor controller integrated transmission MCRT and the rear transmission control unit TCUR to achieve automatic gear shifting and speed change of the vehicle; on the power domain CAN-2, there is a power domain integrated module group, which mainly can include a DC-DC converter, a steering DC-AC converter, a braking DC-AC converter, a high-voltage battery management module, a high-voltage power distribution unit, a power supply module for the hoisting operation hydraulic system, etc., not listed one by one;

[0208] (6) The vehicle control unit VCU is connected to the chassis domain CAN, the power domain CAN-1, and the power domain CAN-2 at the same time to achieve the control of the vehicle chassis system and power system;

[0209] (7) The body domain controller BDC is connected to the ICG gateway through the body domain CAN. Below the BDC, the main tire pressure monitoring module TPMM and the slave tire pressure monitoring module TPMS are mounted through the CAN bus. The main function of the BDC is to check and give early warnings about the tire pressure status of the vehicle before the hoisting operation is executed;

[0210] (8) The hoisting operation domain controller LODC is connected to the ICG gateway through the CAN bus. In addition, a wide-angle camera at the vehicle hook is connected below through LVDS. 1) During the hoisting operation, the camera collects and monitors the video data of the hook of the hook-arm vehicle and the lifting ring of the garbage compactor in real time, and the LODC makes real-time inferences to confirm whether the hook is aligned and hooked with the lifting ring, and whether the hook is firm on the lifting ring, etc.; 2) The deep learning model built in the vehicle-end LODC (using the wide-angle camera at the vehicle hook) needs to identify and detect scenarios including: the vehicle hook is aligned and hooked with the lifting ring of the garbage compactor, and whether it is firm;

[0211] (9) The LODC is connected to the detection modules and electronic control modules required for the hoisting operation through a private CAN, including: a hydraulic oil quantity detection module (check whether the hydraulic oil quantity of the vehicle hydraulic system is normal), a hydraulic pipeline detection module (check whether there is leakage of the lubricating oil in the vehicle hydraulic system), a quick connector detection module (check whether there is damage to the quick connectors for disconnecting and connecting the oil circuit), a power take-off electronic control module (start / stop the power take-off in an electronic control manner), a rear outrigger electronic control module (lower / restore and retract the rear outriggers), a main boom lock hook electronic control module (lock / unlock the main boom lock hook), a hook-arm telescopic electronic control module (operate the telescopic length of the hook-arm);

[0212] As an optional method, the intelligent cockpit domain controller ICDC is mounted with the central control screen CSD through LVDS so that the driver can set parameters and monitor the real-time status of the hoisting operation in the vehicle. The ICDC supports connection methods such as in-vehicle Ethernet and CAN-FD for external associated systems.

[0213] Embodiment 2:

[0214] Based on Embodiment 1, combined with Figure 2 As shown, a field-edge multi-access edge computing MEC system communicates interactively with a hook-arm garbage truck and intelligent terminals such as mobile phones:

[0215] (1) The field sensors include a wide-angle camera at the top frame of the garbage house door, a wide-angle camera on the left side in the middle of the line parking space, and a wide-angle camera on the right side in the middle of the line parking space. They are connected to the field MEC system through an AHD analog high-definition video transmission link;

[0216] (2) The main components of the field MEC system include: an image acquisition and preprocessing module, an environmental perception real-time inference module, a logic calculation and judgment warning module, and a V2I communication module; among them:

[0217] Image acquisition and preprocessing module: Process the video images input by the camera in parallel in a multi-threaded manner. In the preprocessing stage, optimize the data quality through algorithms such as noise suppression, data normalization, color space transformation, and geometric distortion correction;

[0218] Environmental perception real-time inference module: Based on the high-quality image data frames input in the previous step, realize the recognition, detection, and tracking of dynamic targets such as pedestrians and moving obstacles. Specific algorithms can consider One Stage detectors YOLO-v11, NanoDet, or optimized models based on attention mechanisms such as MobileViT, EfficientFormer, etc.;

[0219] Logic calculation and judgment warning module: Based on the results of environmental perception real-time inference, perform necessary logical calculations and analyze and judge:

[0220] ① Is the hoisting operation area safe (i.e., there are no pedestrians and other dynamic moving obstacles)?

[0221] ② During the hoisting process, does the garbage compactor maintain balance and stability? If there is a deviation from the center, what is the deviation distance?

[0222] ③ When the garbage compactor is completely hoisted onto the rear frame, does its bottom fit perfectly with the rear frame slide rail?

[0223] ④ When the hoisting operation is started, are the vehicle's rear outriggers deployed? When the hoisting operation is completed, are the vehicle's rear outriggers retracted?

[0224] The deep learning model built into the field MEC (using the camera at the top frame of the garbage house door, the scenarios that require auxiliary vehicle-end recognition and detection include: pedestrians and other dynamic obstacles in the hoisting operation area;

[0225] The deep learning model built into the edge MEC (using the camera on the top frame of the garbage room door) needs to identify and detect the following scenarios: pedestrians and other dynamic obstacles in the hoisting operation area; during hoisting, the centering / deviation of the middle position between the garbage compactor and the two guide wheels on the rear vehicle frame; when the hoisting operation starts / ends, the deployment / retraction of the vehicle's rear outriggers.

[0226] The deep learning model built into the edge MEC (using the cameras on the left and right sides in the middle of the field-side line parking space) needs to identify and detect the following scenarios: during the process of loading the box onto the vehicle, the bottom of the garbage compactor is fully attached to the slide rail of the rear vehicle frame; when the hoisting operation starts / ends, the deployment / retraction of the vehicle's rear outriggers.

[0227] V2I communication module: Based on the 5G-V2X PC5 interface, it enables direct communication between the RSU and the vehicle, supports low-latency and high-reliability broadcasting, with a coverage range of 1 km+. Through wireless broadcasting, it can actively broadcast the above calculation and analysis results to the hook-arm garbage truck periodically (low latency < 100 ms). The vehicle's ICG gateway integrates 5G-V2X functions and can passively receive the V2I messages broadcasted. The V2I communication protocol can use the SAE J2735 core standard, ASN.1 encoding, and RSM (Road Side Safety Message) data structure.

[0228] The in-vehicle intelligent central computing gateway ICG (Gateway + Tbox 5G-V2X) realizes V2I communication with the edge MEC and communicates with the driver's mobile phone APP via Bluetooth; the broadcast messages received by the ICG from the MEC are routed and forwarded to the automatic parking domain controller APDC, and APDC decides whether to stop the vehicle (forward the vehicle control instruction to the chassis system through the ICG) and whether to stop the hoisting operation (forward the vehicle control instruction to the hoisting operation domain controller LODC through the ICG to control various operation modules).

[0229] The driver's mobile phone APP connects to the vehicle via Bluetooth. After the driver gets out of the vehicle, they can start the unmanned hoisting operation with one key on the mobile phone. During the hoisting operation, the vehicle's running status information, the operation information of each module of the hoisting operation, danger warning information, etc. can all be sent to the mobile phone APP via the Bluetooth protocol for real-time display and warning.

[0230] Optionally, the mobile phone APP and the vehicle can also establish a connection through the WIFI hotspot. The vehicle sends video streams such as the AVM panoramic stitching video stream and the video stream of the hook aligning with the sling to the mobile phone for real-time monitoring.

[0231] Optionally, if the driver is in the vehicle cab, the vehicle can send its own operating status information, operation information of each module of the lifting operation, danger warning information, AVM surround view stitching video stream, video stream such as the hook and the sling hook alignment and engagement, etc. to the intelligent cockpit domain controller ICDC via the CAN bus and in-vehicle Ethernet for real-time display, monitoring and warning on the central control screen CSD.

[0232] Embodiment 3:

[0233] Based on Embodiment 2, the AI deep learning network built into the vehicle-end control and yard-end controller needs to be trained based on the visual image data of the industrial environment perception scenarios involved in the lifting operation. The deep learning model trained based on a small dataset is prone to overfitting, and in actual use, the recognition and detection accuracy is not high during training. The present invention designs a technology based on the SD Stable Diffusion AI large model such as image-to-image and text-to-image to expand the training samples into a high-quality dataset, greatly improving the inference accuracy of the intelligent lifting recognition and detection models at the vehicle end and yard end, that is, the built-in AI deep learning network.

[0234] As Figure 6 shown, expanding and generating a high-quality dataset based on the SD Stable Diffusion AI large model includes:

[0235] (1) The composition structure of the AI large model based on SD Stable Diffusion includes a ClipText text-image encoder, an Image Information creator reconstructed based on U-Net and a 50-step scheduler, and a VAE model image decoder;

[0236] (2) Input the visual image and the user prompt text into ClipText, input the visual image into the ResNetEncoder residual network encoder, and output the image embedding Embedding; input the user prompt text into the TransformerEncoder conversion encoder, and output a 77ⅹ768 (77 token embeddings vectors, each token vector has 768 dimensions) text embedding Embedding;

[0237] (3) The SDM Stable Diffusion model image information creator consists of a U-NET model with an attention mechanism, and uses the Scheduler scheduler to complete 50 repeated iterations (running multiple steps Steps) of scheduling calculations for the U-NET model to generate a predicted noise image tensor Tensor;

[0238] (4) U-NET is a conditional diffusion model that is responsible for gradually denoising in the latent space to generate the target image. Its input is a latent variable with noise (64x64x4 tensor), timestep embedding, and text condition (ClipText text encoding); its output is the predicted noise residual (with the same dimension as the input); a U-NET contains 18 ResNet Blocks residual network blocks and 9 Attention attention blocks; the U-NET of Stable Diffusion achieves efficient latent space denoising through a fine combination of residual modules and attention blocks, and its design balances computational efficiency and generation quality;

[0239] (5) The composition structure of U-NET includes a downsampling path encoder, an upsampling path decoder, and a bottleneck layer Bottleneck; the encoder consists of 4 downsampling stages, and each stage contains: 2 ResNet Blocks residual network blocks (responsible for feature extraction and temporal condition fusion), 1 Spatial Transformer (Attention) spatial transformer based on the attention mechanism (responsible for introducing text conditions and spatial attention), and 1 DownSample downsampling block (reducing the resolution through strided convolution or pooling); the decoder consists of 4 upsampling stages, and each stage contains: 2 ResNet Blocks residual network blocks (restoring details by combining residual connections), 1 Spatial Transformer (Attention) spatial transformer based on the attention mechanism (fusing conditional information), and 1 UpSample upsampling block (increasing the resolution through interpolation or transposed convolution); the bottleneck layer Bottleneck connects the middle part of the encoder and the decoder, and it contains: 2 ResNet Blocks residual network blocks (used for further extracting global features), 1 Spatial Transformer (Attention) spatial transformer based on the attention mechanism (enhancing long-range dependence modeling);

[0240] (6) The internal composition of ResNet Block includes: GroupNorm normalization layer to improve training stability; SILU activation to replace the traditional ReLU and enhance non-linearity; 3x3 convolutional layer to process spatial features; timestep embedding injection to integrate temporal information into features through a fully connected layer; residual connection to retain input features and alleviate gradient vanishing;

[0241] (7) The core mechanisms of Spatial Transformer (Attention) include: Self-Attention, which models the relationships between internal regions of the image; Cross-Attention, which injects text conditions (CLIP embeddings) into the image generation process; and positional encoding, which preserves spatial position information through sine encoding.

[0242] (8) An Attention module is added after each ResNet Block that transforms latent space data in U-NET. As a text conditioning mechanism, it incorporates the input TokenEmbedding into each processing stage and then sends it to the next ResNet Block, forming a cascaded structure. Not all of the input to each Attention module is sent to the next module for processing. Instead, a portion of it is directly sent to the final stage of the processing process in the form of an Attention module.

[0243] (9) Latent Seed: The latent variables are initialized using random probabilities based on Gaussian distribution noise. A noise amount intensity regulator is used to adjust the noise intensity through a proportional variance scaling algorithm, predict the noise residuals based on the time step timestep, and update the noise amount embedding (the latent variable of the noise intensity level) according to the scheduler.

[0244] (10) The VAE image decoder gradually reconstructs an image (512×512) from the latent variable through upsampling as the output. Its composition structure includes: a latent space input layer (64×64×4 tensor); a convolutional upsampling path decoder, which includes 1 CNN convolutional module, 1 MidBlock intermediate block, 3 UpBlock upsampling blocks, 1 ResNet Block residual network block, and 1 GSC (Group Normalization + Swish activation function + Conv convolutional layer) module in the network building order. The MidBlock intermediate module includes in the network building order: 1 ResNet Block residual network block + 1 SelfAttention self-attention module + 1 ResNet Block residual network block; The UpBlock block includes in the network building order: 1 Interpolate interpolation module (upsamples the image, usually using bilinear interpolation to magnify the spatial resolution of the image) + 1 Conv convolutional layer (performs convolution operations on the upsampled image to restore image details); The ResNet Block residual network block introduces skip connections. After the input features are processed through several layers of convolution and activation, the original input features are superimposed to enhance the learning ability and depth of the network, thus solving the problem of gradient disappearance in deep networks; Each element in the SelfAttention self-attention module (such as pixels or features in an image) interacts with other elements through attention weights to obtain global features, solving the problem of capturing long-range dependencies; The GSC module combines (GroupNorm normalizes features, stabilizes the training process, and ensures high-quality normalization even in the case of small batches; The Swish activation function provides smooth and differentiable activation, enhancing the non-linear expression ability of the model and being suitable for deep networks and generative models; Conv extracts local features of the image and is the basic operation unit of CNN);

[0245] (11) The SDM adopts mature AI large models and cloud computing resources, and generates a large amount of high-quality datasets for fine-tuning the YOLO-v11 through text-to-image and image-to-image mechanisms. The well-trained and widely used amazon stability.ai (Stable Diffusion XL) model can be considered. This model is accessed through the managed API provided by Amazon Bedrock. AWS has performed hardware acceleration on the model using Inferentia2 chips and the SageMaker inference toolchain, which can greatly reduce inference latency and costs. The model is deployed on EC2 Inf2 instances (optimized for inference) or GPU instances (such as P4d / P5), and the generated content is stored in S3, supporting version control and lifecycle management. The access rights to the model are controlled through IAM roles, and static data is encrypted using AWS's KMS key management service. User data is encrypted during transmission through AWS PrivateLink, meeting standards such as GDPR and HIPAA;

[0246] As Figure 7 shown, fine-tuning training and inference are performed based on the YOLO-v11 model, specifically including:

[0247] (1) The pre-trained models from YOLO-v8 to YOLO-v11 (such as trained on the COCO dataset) are mainly for common objects in natural scenes (such as pedestrians, vehicles, animals, etc.). As a general object testing framework, there are significant differences in image features between dedicated industrial job quality inspection and other fields and the general dataset. Therefore, in specific professional applications, additional fine-tuning and retraining are usually required to solve problems such as the non-existence of the COCO dataset in specific job scenarios, the need to redefine category labels, and the high-precision positioning of small targets (hook and ring alignment), symmetry detection of medium targets (guide wheel centering status), contact surface judgment of large targets (box and slide rail fitting detection), and binary classification anomaly detection (rear leg status classification) during the operation process;

[0248] (2) High-quality datasets generated based on the expansion of the stable diffusion AI large model are used to annotate professional images using software tools such as LabelImg and Roboflow to generate YOLO-format label files;

[0249] (3) Divide the training set, validation set, and test set according to the ratio of 8:1:1 and balance the number of samples in each category;

[0250] (4) Perform parameter fine-tuning on the YOLO-v11 model, and the thawing and freezing strategies are:

[0251] ①In the stage of freezing the shallow CSP cross-stage partial network of the Backbone model, since the shallow layer of the Backbone is responsible for extracting basic features (such as edges and textures), and the low-level features of industrial instruments are highly similar to those of natural images, there is no need for retraining;

[0252] ②In the stage of unfreezing the deep C2PSA (including multiple PSA Blocks, pyramid squeeze attention blocks, and each PSA contains 1 Attention attention block, as Figure 8 shown) of the Backbone model, the PSA Block is a core module in the Swin Transformer. Its main function is to improve the computational efficiency and feature extraction ability of the model through windowed multi-head self-attention (W-MSA) and shifted window multi-head self-attention (SW-MSA). The PSA Block reduces the computational amount by restricting the scope of attention calculation within the window and enhances the interaction between windows through the window shift mechanism, thus improving the representation ability of the model. The working principle of the PSA Block includes: Patch Partition (dividing the input image into non-overlapping Patch image patches), Linear Embedding (mapping each Patch to a high-dimensional vector space), and Swin Transformer Block (including two main steps: W-MSA and SW-MSA. W-MSA calculates multi-head self-attention within each window, while SW-MSA recalculates attention through window shifting to enhance the interaction between windows. Both steps include Layer Normalization (LN) and Multi-Layer Perceptron (MLP) modules); in addition, the PSA Block can be combined with the Patch Merging Layer, which merges adjacent Patch features, reduces the resolution while increasing the number of channels, and further improves the feature extraction ability. The Swin Transformer module is good at capturing global context relationships and needs to be adjusted to identify the structural associations of mechanical components (such as the topological relationship between a hook and a sling). The ultimate goal of this step is to retain the general feature extraction ability (edges / textures) and adjust the high-level semantic features to adapt to the form of industrial instruments;

[0253] ③Unfreeze all BiFPN-XL layers of the Neck model, aiming to enhance the multi-scale feature fusion ability and optimize the collaborative detection of small targets (such as slings relying on high-resolution feature maps) and large targets (such as boxes requiring semantic information of low-resolution feature maps). The dynamic weight mechanism needs to relearn the feature importance in the industrial scenario;

[0254] ④Unfreeze the classification branch parameters (Conv2d and CLS Loss) of the Detect detector in the Head model, as Figure 9As shown, since the state of the vehicle's rear outriggers (extended / retracted) belongs to a newly defined binary classification task, a classifier needs to be trained from scratch; the initial layers of the regression branch (the first two DWConv and Conv) are frozen because the pre-trained model already has the general object localization ability, and only the high-level parameters need to be adjusted during fine-tuning to adapt to the precise offset of mechanical components. The ultimate goal of this step is that the classification task needs to adapt to new categories (such as the state of the vehicle's rear outriggers), and the regression task retains the pre-trained localization ability and fine-tunes local parameters;

[0255] (5) Training optimization strategy

[0256] 1) Optimize the data augmentation configuration for the industrial scenario of unmanned lifting operations, including but not limited to: enhancing hue changes (simulating different lighting), increasing saturation (enhancing the visibility of rust / stains), controlling brightness (adapting to night operations), increasing the rotation range (variable vehicle entry angles), enabling MixUp augmentation (improving the robustness of small targets), etc.;

[0257] 2) For the scenario where the position of the sling is sensitive and the hook is aligned with and hooked to the sling, the loss function considers the localization loss CIoULoss + increasing the weight coefficient of small targets to enhance the loss weight:

[0258]

[0259] Among them, ￡ CLoU is a loss function that calculates the distance between two bounding boxes in object monitoring. It is improved based on IoU and can more accurately reflect the similarity between two bounding boxes. ￡ CLoU also considers factors such as the position, size, and shape of the bounding boxes, and can effectively improve the accuracy of object detection.

[0260] IoU is the intersection over union ratio between two bounding boxes, and ρ 2 (b pred ,b gt ) represents the square of the center distance between the predicted box and the GT (GroundTruth) box, c is the length of the diagonal of the smallest enclosing box of the two boxes, α is the adjustment coefficient, v represents the aspect ratio of the bounding box, w is the width of the image, h is the height of the image, img_size is the size of the image, and ω small is the weight coefficient of small targets:

[0261] Dynamically calculate: ω small = 1.0 + 2 * (1 - (wh) / img_size)

[0262] Fixed setting: For targets with an area less than 32 * 32, ω small is given a weight 2.0 - 3.0 times.

[0263] 3) For the case of unbalanced vehicle rear outrigger state samples, the loss function considers the classification focal loss FocalLoss + influence factor to parametrically improve the focal loss:

[0264]

[0265] Among them, ￡ CLS represents the binary cross-entropy loss function, α dynamic represents the weight coefficient, p t represents the predicted probability, γ adaptive is the adaptive adjustment factor, epoch represents the number of cycles of model training, and total_epoch represents the total number of cycles of model training;

[0266] (6) Optimize the acceleration strategy, deploy using Tensor RT, convert the fine-tuned YOLO-v11 model to FP16 precision, and optimize the computation graph. For edge device adaptation, chips such as Nvidia Jetson AGX Orin can be used, and the real-time inference speed ≥ 45 FPS;

[0267] (7) Model verification and debugging, considering key indicators such as overall accuracy, strict threshold, and detection recall rate in the lifting operation scenario (to avoid operation failures caused by missed detections); the visualization tool TensorBoard can be used to monitor the feature map response to ensure significant activation of small targets such as the sling area;

[0268] Freeze or unfreeze the specified layers and modules of the YOLO-v11 model before fine-tuning;

[0269] Perform fine-tuning training based on the YOLO-v11 model;

[0270] Evaluate based on the fine-tuned YOLO-v11 model;

[0271] Perform real-time inference based on the fine-tuned YOLO-v11 model.

[0272] In summary, by specifically freezing the shallow layers of the Backbone, unfreezing the deep layers and Neck, and optimizing the Head classification branch, YOLO-v11 can efficiently adapt to the detection requirements of the unmanned lifting operation of hook-arm garbage trucks. In actual deployment, the model is respectively deployed to the MEC edge computing system and the vehicle-end LODC lifting operation domain controller, and the fine-tuned YOLO-v11 with parameter adjustment will feedback the real-time detection results (such as coordinate offsets) to the control unit in real time to achieve unmanned full-automatic lifting operations.

[0273] Example 4:

[0274] Based on Example 1, an automatic lifting operation method for a hook-arm garbage truck includes:

[0275] Step S1, Automatic parking process, including:

[0276] A1. The driver drives the hook-arm garbage truck to the urban fixed garbage collection point and stops about 5 meters in front of the special line parking space, then starts the automatic parking.

[0277] A2. The Automatic Parking Domain Controller (APDC) performs line parking space detection and recognition, surrounding environment perception of the operation area, and obstacle recognition and positioning based on the video data input by the image detection unit.

[0278] A3. The Automatic Parking Domain Controller (APDC) generates a parking path, starts parking, and performs collision detection, trajectory tracking control, and exception handling.

[0279] A4. The hook-arm garbage truck stops at the central position of the line parking space, and the automatic parking is completed.

[0280] Step S2, Status check before starting the lifting operation, including:

[0281] (1) The Lifting Operation Domain Controller (LODC) determines whether the status of the hydraulic oil quantity, hydraulic pipeline, and quick connector reported by the detection module is normal.

[0282] (2) The Lifting Operation Domain Controller (LODC) determines whether the status reported by the electric control module is normal.

[0283] (3) The Body Domain Controller (BDC) determines whether the tire pressure data is normal.

[0284] (4) The Automatic Parking Domain Controller (APDC) confirms whether the status of the Electric Power Steering System (EPS) and the Electro-Mechanical Brake (EMB) of the chassis system is normal.

[0285] (5) The Automatic Parking Domain Controller (APDC) confirms that there are no people or obstacles in the lifting operation area.

[0286] If any of (1)-(4) is abnormal or there are people or obstacles in the lifting operation area, an alarm is issued; otherwise, the status check before the lifting operation is completed.

[0287] Step S3, Lifting operation process, including:

[0288] B1. Start the unmanned lifting operation.

[0289] B2. The Vehicle Control Unit (VCU) controls the Electronic Parking Brake (EPB) of the chassis system to brake the vehicle's handbrake; the Vehicle Control Unit (VCU) controls the Rear Transmission Control Unit (TCUR) of the power system to put the vehicle in neutral gear.

[0290] B3. The Lifting Operation Domain Controller (LODC) controls the rear outrigger electric control module in the electric control module to lower the vehicle's rear outriggers.

[0291] B4. The vehicle control unit (VCU) controls the power supply and power domain module group of the built-in lifting operation hydraulic system to start the vehicle engine, turn on the power supply of the hydraulic system, and the VCU controls the MCRT to provide power.

[0292] B5. The vehicle control unit (VCU) controls the rear transmission control unit (TCUR) to drive the synchronizer, and the lifting operation domain controller (LODC) controls the power take-off electronic control module in the electronic control module. The two cooperate to complete the power take-off action of the hydraulic pump.

[0293] B6. The lifting operation domain controller (LODC) controls the main boom lock hook electronic control module in the electronic control module to control the main boom lock hook to be in a fully released state.

[0294] B7. The lifting operation domain controller (LODC) controls the boom telescoping electronic control module in the electronic control module to continuously extend the main boom cylinder until the sub-boom hook flips to a position slightly lower than the hook center of the garbage compactor's lifting ring.

[0295] B8. The vehicle control unit (VCU) controls the rear transmission control unit (TCUR) to drive the synchronizer, and the lifting operation domain controller (LODC) controls the power take-off electronic control module in the electronic control module. The two cooperate to close the power take-off of the hydraulic pump.

[0296] B9. The vehicle control unit (VCU) controls the rear transmission control unit (TCUR) to drive the synchronizer to shift the vehicle into reverse gear. The VCU controls the rear motor controller of the power system to integrally rotate the MCRT to make the vehicle reverse backward to the position where the hook is at the hook center of the garbage compactor's lifting ring. The connection state between the hook and the lifting ring is monitored in real time. After the LODC fully confirms that it is firmly hooked, the VCU controls the EMB to brake the vehicle.

[0297] B10. Perform hydraulic power take-off again: The vehicle control unit (VCU) controls the rear transmission control unit (TCUR) to drive the synchronizer, and the lifting operation domain controller (LODC) controls the power take-off electronic control module in the electronic control module. The two cooperate to complete the power take-off action of the hydraulic pump.

[0298] B11. The lifting operation domain controller (LODC) controls the boom telescoping electronic control module to continuously retract the main boom cylinder, tow the garbage compactor to the slide rail on the rear frame, and continuously lift it until the box is completely suspended and parallel to the frame until the box is completely pulled onto the rear frame.

[0299] B12. Monitor the centering of the middle position between the box and the two guide wheels on the rear frame. The lifting operation domain controller (LODC) performs real-time analysis and judgment. If it is determined that the box deviates from the centering position, an early warning message will be immediately sent to notify the vehicle to adjust the position and angle.

[0300] B13. During the process of loading the box body onto the vehicle, monitor whether the bottom of the garbage compression box is fully attached to the slide rail of the rear frame;

[0301] B14. Receive the message of the fitting state between the box body and the slide rail of the rear frame. If it is fully attached, the main boom locking hook electronic control module is controlled by the lifting operation domain controller LODC to make the main boom locking hook in a fully locked state, and proceed to the next step; if it is prompted that they cannot be attached, the vehicle needs to adjust its position and angle. After receiving the message that they cannot be fully attached multiple times, the vehicle will suspend the lifting operation;

[0302] B15. After the vehicle's lifting operation is normally completed, the vehicle control unit VCU controls the rear transmission control unit TCUR to drive the synchronizer, and the lifting operation domain controller LODC controls the power take-off electronic control module. The two cooperate to close the hydraulic pump power take-off and release the hydraulic pressure;

[0303] B16. The vehicle control unit VCU controls the power domain module group to turn off the power supply of the lifting operation hydraulic system; the lifting operation domain controller LODC controls the rear outrigger electronic control module to retract the vehicle's rear outriggers, and the lifting operation is completed.

[0304] Embodiment 5:

[0305] Based on Embodiment 2 or 3, an automatic lifting operation method for a hook-arm type garbage truck includes:

[0306] The automatic parking process of the hook-arm type garbage truck is as Figure 3 shown, and specifically includes:

[0307] (C1) The driver drives the hook-arm type garbage truck to a fixed urban garbage collection point and stops about 5 meters in front of the dedicated line parking space. The driver gets out of the vehicle;

[0308] (C2) The driver uses the mobile phone APP to establish a wireless connection such as Bluetooth / WIFI with the vehicle and starts automatic parking with one key;

[0309] (C3) The AVM surround-view camera corrects the distortion using a polynomial model through calibration parameters (intrinsic parameters, extrinsic parameters), and then uses inverse perspective mapping IPM to project the 4-channel camera images onto a unified bird's-eye view coordinate system;

[0310] (C4) Detection and recognition of the line parking space: Use the HSV color space to segment the marking area (such as white / yellow), and use Canny to detect the edges. Use the Hough transform to detect straight line segments and fit the parking space boundary. Use a semantic segmentation model (such as UNet) to directly predict the parking space mask;

[0311] (C5) Obstacle Recognition and Localization: Use YOLO-v11 or EfficientDet to detect pedestrians, dynamic obstacles, etc. within the hoisting operation area, estimate the relative distance between the vehicle and the obstacles, and perform real-time tracking;

[0312] (C6) According to the parking space coordinates, the obstacle list, and the current pose of the vehicle, use the hybrid A* algorithm, considering the vehicle kinematic constraints, to generate a parking path. Expand the obstacle boundary by the vehicle outline, take points at equal intervals along the path, and determine whether they intersect with the inflated obstacles;

[0313] (C7) Use the Model Predictive Control algorithm MPC to optimize the control input, minimize the path tracking error and the change rate of the control quantity. Send control commands to the EPS to control the steering wheel angle. Send control commands to the EMB to achieve vehicle speed control and maintain a constant low speed of <5 km / h with a wire-controlled throttle / brake. Predict the parking space position based on historical data. If an exception is triggered due to the loss of parking space information, trigger emergency braking;

[0314] (C8) The vehicle accurately stops at the center position of the on-street parking space, the automatic parking action is completed, and the intelligent driving mode is exited;

[0315] The status check process before unmanned hoisting operations is as Figure 4 shown, specifically including:

[0316] (D1) The maintenance personnel at the urban garbage collection point confirm whether the status of the garbage compactor meets the conditions for unmanned hoisting operations, including whether it is placed in a garbage house on flat ground, whether the box is deformed, whether the position of the lifting ring is damaged, whether the garbage weight is overloaded (a weight sensor can be installed in the garbage house to weigh the garbage compactor, or observe whether the metal garbage cover can be tightly closed without garbage overflow), etc.;

[0317] (D2) The maintenance personnel at the urban garbage collection point confirm that the ground in the hoisting operation area needs to be solid and flat to avoid vehicle tilt caused by softness or slopes;

[0318] (D3) After the status of the garbage compactor and the ground status of the operation environment are confirmed, all the maintenance personnel at the garbage point leave the site, and then notify the driver that unmanned hoisting operations can be carried out. The driver uses the mobile phone APP outside the vehicle to start the status check function before hoisting operations with one key;

[0319] (D4) The TBox 5G-V2X module integrated in the vehicle ICG gateway and the field-side MEC establish a V2I short-range wireless connection to enable the field-side MEC to periodically send the environmental perception results to the vehicle side for decision-making;

[0320] (D5) The vehicle-end LODC receives the status check instruction before the hoisting operation from the ICG gateway, and requests the hydraulic oil quantity detection module, the hydraulic pipeline detection module, and the quick connector detection module to report their detection results to itself respectively;

[0321] (D6) The LODC determines whether the hydraulic lubricating oil quantity is normal, whether the hydraulic pipeline is intact, whether the quick connector is damaged, and whether there is any lubricating oil leakage;

[0322] (D7) In addition, the LODC requests the power take-off electronic control module, the rear outrigger electronic control module, the main boom hook electronic control module, and the hook arm telescopic electronic control module to report their statuses to itself respectively;

[0323] (D8) The LODC determines whether the statuses of the power take-off, rear outriggers, main boom hook, and hook arm telescopic electronic control modules are normal;

[0324] (D9) The vehicle-end body domain controller BDC receives the status check instruction before the hoisting operation from the ICG gateway, requests the TPMM and TPMS to report the tire pressure detection data to itself, and analyzes and determines whether the main tire pressure and the slave tire pressure of the vehicle are normal;

[0325] (D10) The chassis EPS and EMB check whether their working statuses are normal and report their statuses;

[0326] (D11) The vehicle-end AVM surround view camera and APDC confirm that there are no personnel or dangerous objects within a radius of 5 meters of the hoisting operation area;

[0327] (D12) The field-end camera and MEC assist in confirming that there are no personnel or dangerous objects within a radius of 5 meters of the hoisting operation area, and send the confirmation result message to the vehicle through V2I;

[0328] (D13) All the above status confirmation information before the hoisting operation is reported by the LODC, BDC, APDC, EPS, and EMB to the vehicle ICG gateway respectively, and is sent to the driver's mobile APP through the TBox Bluetooth wireless protocol integrated in the ICG. The mobile APP will visually list the detailed results of each status check in detail, and abnormal statuses need to be warned on the mobile phone side;

[0329] The workflow of unmanned hoisting operation is as Figure 5 shown, and specifically includes:

[0330] (E1) The driver enables the unmanned hoisting operation function with one key through the mobile APP;

[0331] (E2) The operation instruction is sent to the TBox integrated in the vehicle ICG through the Bluetooth protocol, and the ICG forwards the operation instruction to the VCU. The VCU controls the EPB to brake the vehicle's handbrake. The VCU controls the TCUR to put the vehicle in neutral;

[0332] (E3) The ICG forwards the operation instruction to the LODC, and the LODC controls the rear outrigger electronic control module to lower the rear outriggers of the vehicle.

[0333] (E4) The VCU controls the power domain module group (the power supply for the hoisting operation hydraulic system is already built-in), starts the vehicle engine, and turns on the power supply of the hydraulic system. In addition, the VCU controls the MCRT to maintain the speed at 1500 - 2000 rpm to provide sufficient power.

[0334] (E5) The VCU controls the TCUR to drive the synchronizer, and the LODC controls the power take-off electronic control module. The two cooperate to complete the power take-off operation of the hydraulic pump.

[0335] (E6) The LODC controls the main boom hook electronic control module to set the hook to the "loosen" position and hold it for 5 - 8 seconds until the main boom hook is completely loosened.

[0336] (E7) The LODC controls the hook arm telescoping electronic control module to continuously extend the main boom cylinder until the sub-boom hook flips to a position slightly lower than the hook center of the garbage compactor's lifting ring.

[0337] (E8) The VCU controls the TCUR to drive the synchronizer, and the LODC controls the power take-off electronic control module. The two cooperate to close the power take-off of the hydraulic pump.

[0338] (E9) The VCU controls the TCUR to drive the synchronizer to put the vehicle in reverse gear. The VCU controls the MCRT to reverse the vehicle backward to the position where the hook is at the hook center of the garbage compactor's lifting ring. The camera at the vehicle hook monitors the connection status between the hook and the lifting ring in real time. After the LODC fully confirms that it is firmly hooked, the VCU controls the EMB to brake the vehicle.

[0339] (E10) Perform hydraulic power take-off again: The VCU controls the TCUR to drive the synchronizer, and the LODC controls the power take-off electronic control module. The two cooperate to complete the power take-off operation of the hydraulic pump.

[0340] (E11) The LODC controls the hook arm telescoping electronic control module to continuously retract the main boom cylinder, tow the garbage compactor to the slide rail on the rear frame, and continuously lift it until the box is completely suspended and parallel to the frame until the box is completely pulled onto the rear frame.

[0341] (E12) The camera at the top frame of the garbage house door monitors the middle position between the box and the two guide wheels on the rear frame for centering monitoring, and the MEC performs real-time analysis and judgment. If it is determined that the box deviates from the centering position, an early warning message will be immediately sent through V2I to notify the vehicle to adjust the position and angle.

[0342] (E13) During the process of loading the box body onto the vehicle, cameras on the left and right sides in the middle of the field-end line parking space monitor whether the bottom of the garbage compression box is fully attached to the sliding rail of the rear vehicle frame, and the MEC performs real-time analysis and judgment. The MEC sends the attachment status message to the vehicle through V2I;

[0343] (E14) When the vehicle receives the attachment status message of the box body and the sliding rail of the rear vehicle frame through V2I, if it is fully attached, the LODC controls the main arm hook electronic control module to set the hook to the "locked" position and hold it for 5 to 8 seconds until the main arm hook is in a fully locked state. If the V2I message indicates that it cannot be attached, the vehicle needs to adjust its position and angle. After receiving the message that it cannot be fully attached multiple times, the vehicle will suspend the lifting operation and send a warning to the driver's mobile phone APP via wireless Bluetooth for manual intervention;

[0344] (E15) After the vehicle's lifting operation is completed normally, the VCU controls the TCUR to drive the synchronizer, and the LODC controls the power take-off electronic control module. The two cooperate to close the hydraulic pump power take-off and release the hydraulic pressure;

[0345] (E16) The VCU controls the power domain module group to turn off the power supply of the lifting operation hydraulic system; the LODC controls the rear outrigger electronic control module to retract the vehicle's rear outriggers; the vehicle's TBox notifies the driver's mobile phone APP via wireless Bluetooth that the lifting operation is completed;

[0346] (E17) After the driver gets into the vehicle and manually lowers the EPB handbrake, after the vehicle's ICG receives this signal, it disconnects the V2I connection with the field-end MEC, and the driver drives away.

[0347] Although the present invention has been described herein with reference to its illustrative embodiments, the above embodiments are only the preferred embodiments of the present invention. The embodiments of the present invention are not limited by the above embodiments. It should be understood that those skilled in the art can design many other modifications and embodiments, and these modifications and embodiments will fall within the scope and spirit of the principles disclosed in this application.

Claims

1. An automatic lifting operation system for a hook-arm garbage truck, characterized in that, It includes an image detection module, an intelligent central computing gateway ICG, and an automatic parking domain controller APDC, a chassis system, a vehicle control unit VCU, a power system, a body domain controller BDC, an intelligent cockpit domain controller ICDC, and a lifting operation domain controller connected to the intelligent central computing gateway ICG. Among them: The image detection module is used to obtain the video data of the line parking space of the hook-arm garbage truck and the video data of the surrounding environment of the lifting operation, and input them into the automatic parking domain controller APDC; it is also used to collect the video data of the hook of the hook-arm garbage truck and the lifting ring of the garbage compression box, and input them into the lifting operation domain controller LODC; The intelligent central computing gateway ICG is used to implement protocol stack encapsulation and data routing and forwarding among the automatic parking domain controller APDC, the chassis system, the vehicle control unit VCU, the power system, the body domain controller BDC, the intelligent cockpit domain controller ICDC, and the lifting operation domain controller LODC; The automatic parking domain controller APDC is used to identify the images input by the image detection module, communicate with the vehicle control unit VCU, and the chassis system of the vehicle control unit VCU realizes automatic parking, or the automatic parking domain controller APDC directly issues control instructions to the chassis system to realize automatic parking; The power system is used to provide transmission control, speed change control, and provide power; The vehicle control unit VCU is used to control the chassis system and the power system; The body domain controller BDC is used to check and give early warnings about the tire pressure status of the vehicle; The lifting operation domain controller LODC is communicatively connected to the image detection module and is used to judge whether the hook is aligned with and hooked to the lifting ring and whether the hook is firm on the lifting ring according to the input video data; it also connects the detection module and the electronic control module through a private CAN. The detection module is used to detect the hydraulic oil volume, hydraulic pipeline, and quick connector of the hook-arm garbage truck, and the electronic control module is used to control the power take-off, rear outriggers, main arm lock hook, and hook arm telescoping of the hook-arm garbage truck to realize the status detection before the lifting operation and start the lifting operation.

2. The automatic lifting operation system of a hook-arm garbage truck according to claim 1, characterized in that, The image detection module includes a first wide-angle camera arranged at the hook of the hook-arm garbage truck and AVM surround-view cameras respectively arranged at the front, rear, left, and right sides of the hook-arm garbage truck; it also includes a second wide-angle camera arranged at the top frame of the garbage house door, a third wide-angle camera on the left side in the middle of the line parking space, and a fourth wide-angle camera on the right side in the middle of the line parking space. The AVM surround-view cameras are used to obtain the video data of the surrounding environment of the lifting operation, the first wide-angle camera is used to collect the video data of the hook and the lifting ring of the garbage compression box, the line parking space is a rectangular frame, the side of the rectangular frame close to the garbage house is N meters away from the garbage house, the width of the rectangular frame = the inner width of the garbage house exit, and the length of the rectangular frame ≥ the length of the hook-arm garbage truck.

3. The automatic lifting operation system of a hook-arm garbage truck according to claim 2, characterized in that, It also includes a field-edge multi-access edge computing (MEC) system and intelligent terminals. The field-edge multi-access edge computing (MEC) system is communicatively connected to the second wide-angle camera, the third wide-angle camera, and the fourth wide-angle camera, and is used to process and judge the input video data of the hook-arm garbage truck's line parking space, and communicate with the automatic parking domain controller (APDC) and the hoisting operation domain controller (LODC) through the intelligent central computing gateway (ICG). The intelligent terminal communicates with the intelligent central computing gateway (ICG) via Bluetooth or WiFi and is used to start the hoisting operation of the hook-arm garbage truck with one key.

4. The automatic lifting operation system of a hook-arm garbage truck according to claim 3, characterized in that, The field-edge multi-access edge computing (MEC) system includes: An image acquisition and preprocessing module, which is used to process the video data collected by the second wide-angle camera, the third wide-angle camera, and the fourth wide-angle camera in parallel, and obtain high-quality image data frames through one or more algorithms among noise suppression, data standardization, color space transformation, and geometric distortion correction; An environmental perception and implementation inference module, which is used to perform object recognition and tracking on the input high-quality image data frames; A logic calculation and judgment warning module, which is used to judge whether the hoisting operation area is safe based on the object recognition and tracking results, and during the hoisting process, judge: Whether the garbage compactor remains balanced and stable and whether there is a center deviation; When the garbage compactor is completely hoisted onto the rear frame, whether its bottom is completely fitted with the rear frame slide rail; Whether the vehicle's rear outriggers are deployed when the hoisting operation is started and whether the vehicle's rear outriggers are restored and retracted when the hoisting operation is completed; A V2I communication module, which is used to actively broadcast the above judgment results to the intelligent central computing gateway (ICG).

5. The automatic hoisting operation system of a hook-arm garbage truck according to claim 1, characterized in that, The chassis system includes an electronic parking brake (EPB), an electro-mechanical brake (EMB), and an electric power steering system (EPS). The automatic parking domain controller (APDC) is communicatively connected to the chassis system through the chassis domain CAN, and the automatic parking domain controller (APDC) is also communicatively connected to the electro-mechanical brake (EMB) and the electric power steering system (EPS) through a private CAN.

6. The automatic hoisting operation system of a hook-arm type garbage truck according to claim 5, characterized in that, The electro-mechanical brake (EMB) adopts a wire-controlled brake dual redundancy setting.

7. The automatic hoisting operation system of a hook-arm garbage truck according to claim 1, characterized in that, The intelligent cockpit domain controller (ICDC) is connected to a central control screen (CSD). The automatic parking domain controller (APDC) forwards the spliced image of the video data input by the image detection module to the intelligent cockpit domain controller (ICDC) through in-vehicle Ethernet and the intelligent central computing gateway (ICG), and the intelligent cockpit domain controller (ICDC) displays it in real time through the central control screen (CSD).

8. The automatic lifting operation system of a hook-arm garbage truck according to claim 1, wherein The power system includes a power domain module group, a rear motor controller integrated rotation (MCRT), and a rear transmission control unit (TCUR). The rear motor controller integrated rotation (MCRT) and the rear transmission control unit (TCUR) are communicatively connected to the vehicle control unit (VCU) and the intelligent central computing gateway (ICG) through the power domain CAN-1; the power domain module group is communicatively connected to the vehicle control unit (VCU) and the intelligent central computing gateway (ICG) through the power domain CAN-2.

9. An automatic lifting operation method for a hook-arm garbage truck implemented by an automatic lifting operation system for a hook-arm garbage truck as described in claim 1, characterized in that, It includes: Step S1, the automatic parking process, includes: A1. Start automatic parking; A2. The Automatic Parking Domain Controller (APDC) performs line parking space detection and recognition, surrounding environment perception of the operation area, and obstacle recognition and positioning based on the video data input by the image detection unit. A3. The Automatic Parking Domain Controller (APDC) generates a parking path, starts parking, and performs collision detection, trajectory tracking control, and exception handling. A4. The hook-arm garbage truck stops at the center position of the line parking space, and the automatic parking is completed. Step S2. Conduct a status check before starting the lifting operation, including: (1) The Lifting Operation Domain Controller (LODC) determines whether the status of the hydraulic oil volume, hydraulic pipeline, and quick connector reported by the detection module is normal. (2) The Lifting Operation Domain Controller (LODC) determines whether the status reported by the electronic control module is normal. (3) The Body Domain Controller (BDC) determines whether the tire pressure data is normal. (4) The Automatic Parking Domain Controller (APDC) confirms whether the status of the Electric Power Steering System (EPS) and the Electro-Mechanical Brake (EMB) of the chassis system is normal. (5) The Automatic Parking Domain Controller (APDC) confirms that there are no personnel or obstacles in the lifting operation area. If any of (1)-(4) is abnormal or there are personnel or obstacles in the lifting operation area, an alarm is issued; otherwise, the status check before the lifting operation is completed. Step S3. The lifting operation process, including: B1. Start the unmanned lifting operation. B2. The Vehicle Control Unit (VCU) controls the Electronic Parking Brake (EPB) of the chassis system to brake the vehicle's handbrake; the Vehicle Control Unit (VCU) controls the Rear Transmission Control Unit (TCUR) of the power system to put the vehicle in neutral. B3. The Lifting Operation Domain Controller (LODC) controls the rear outrigger electronic control module in the electronic control module to lower the vehicle's rear outriggers. B4. The Vehicle Control Unit (VCU) controls the Power Supply and Power Domain Module Group of the built-in lifting operation hydraulic system to start the vehicle engine and turn on the hydraulic system power supply. The Vehicle Control Unit (VCU) controls the MCRT to provide power. B5. The Vehicle Control Unit (VCU) controls the rear transmission control unit (TCUR) to drive the synchronizer, and the Lifting Operation Domain Controller (LODC) controls the power take-off electronic control module in the electronic control module. The two cooperate to complete the power take-off action of the hydraulic pump. B6. The Lifting Operation Domain Controller (LODC) controls the main boom lock hook electronic control module in the electronic control module to control the main boom lock hook to be in a fully released state. B7. The Lifting Operation Domain Controller (LODC) controls the hook arm telescopic electronic control module in the electronic control module to continuously extend the main boom cylinder until the sub-boom hook flips to a position slightly lower than the hook center of the garbage compactor lifting ring. B8. The Vehicle Control Unit (VCU) controls the rear transmission control unit (TCUR) to drive the synchronizer, and the Lifting Operation Domain Controller (LODC) controls the power take-off electronic control module in the electronic control module. The two cooperate to close the power take-off of the hydraulic pump. B9. The vehicle control unit (VCU) controls the rear transmission control unit (TCUR) to drive the synchronizer, shift the vehicle into reverse gear. The VCU of the vehicle control system controls the integrated rotation of the rear motor controller (MCRT) of the power system to reverse the vehicle backward to the position where the hook aligns with the shackle center of the garbage compactor. It monitors the connection status between the hook and the shackle in real time. After the lifting operation domain controller (LODC) fully confirms a secure hook-up, the VCU of the vehicle control system controls the EMB to brake the vehicle. B10. Perform hydraulic power take-off again: The VCU of the vehicle control system controls the TCUR to drive the synchronizer, and the LODC of the lifting operation domain controller controls the power take-off electronic control module in the electronic control module. The two cooperate to complete the hydraulic pump power take-off operation. B11. The LODC of the lifting operation domain controller controls the hook arm telescopic electronic control module to continuously contract the main boom cylinder, tow the garbage compactor to the slide rail of the rear frame, and continuously lift it until the box is completely suspended and parallel to the frame until the box is completely pulled onto the rear frame. B12. Monitor the centering of the middle position between the box and the two guide wheels on the rear frame. The LODC of the lifting operation domain controller performs real-time analysis and judgment. If it determines that the box deviates from the centering position, it immediately sends a warning message to notify the vehicle to adjust its position and angle. B13. During the process of loading the box onto the vehicle, monitor whether the bottom of the garbage compactor is completely fitted with the slide rail of the rear frame. B14. Upon receiving the message of the fitting status between the box and the slide rail of the rear frame, if it is completely fitted, the LODC of the lifting operation domain controller controls the main boom locking hook electronic control module to keep the main boom locking hook in a fully locked state and proceed to the next step. If it indicates non-fitting, the vehicle needs to adjust its position and angle. After receiving the non-fitting message multiple times, the vehicle will suspend the lifting operation. B15. After the vehicle's lifting operation is successfully completed, the VCU of the vehicle control system controls the TCUR to drive the synchronizer, and the LODC of the lifting operation domain controller controls the power take-off electronic control module. The two cooperate to close the hydraulic pump power take-off and release the hydraulic pressure. B16. The VCU of the vehicle control system controls the power domain module group to turn off the power supply of the lifting operation hydraulic system. The LODC of the lifting operation domain controller controls the rear outrigger electronic control module to retract the vehicle's rear outriggers. The lifting operation is completed.

10. A method for automatically hoisting operation of a hook-arm garbage truck according to claim 9, characterized in that, The system further includes a field multi-access edge computing (MEC) system and an intelligent terminal. The field MEC system is communicatively connected to the image detection module, used to process and judge the input video data of the hook-arm garbage truck's line parking space, and communicate with the automatic parking domain controller (APDC) and the lifting operation domain controller (LODC) through the intelligent central computing gateway (ICG). The intelligent terminal communicates with the ICG through Bluetooth or WiFi, used to start the lifting operation of the hook-arm garbage truck with one key. In the method: The method of starting automatic parking in A1 is: The intelligent terminal establishes a communication connection with the hook-arm garbage truck, and starts automatic parking with one key on the intelligent terminal. The starting method of the status check before the hoisting operation in step S2 is as follows: The intelligent terminal starts the status check function before the hoisting operation with one key, and the hook-arm garbage truck establishes a V2I short-range wireless connection with the field multi-access edge computing MEC; Step S2 also includes (6) The field multi-access edge computing MEC confirms that there are no personnel or obstacles in the hoisting operation area; If any one of (1)-(4) is abnormal or there are personnel or obstacles in the hoisting operation area in step S2, an alarm is also sent to the intelligent terminal; Replace B1 with: The intelligent terminal starts the unmanned hoisting operation with one key, and the operation instruction is forwarded to the vehicle control unit VCU and the hoisting operation domain controller LODC through the ICG of the hook-arm garbage truck; Replace B12 with: Monitor the center position between the two guide wheels on the box body and the rear frame, and the field multi-access edge computing MEC performs real-time analysis and judgment. If it is judged that the box body deviates from the center position, immediately send a warning message to notify the vehicle to adjust the position and angle; Replace B13 with: During the process of loading the box body, monitor whether the bottom of the garbage compression box is completely attached to the slide rail of the rear frame, and the field multi-access edge computing MEC performs real-time analysis and judgment, and sends the attachment message to the hoisting operation domain controller LODC; Replace B16 with: The vehicle control unit VCU controls the power domain module group to turn off the power supply of the hoisting operation hydraulic system; The hoisting operation domain controller LODC controls the rear outrigger electronic control module to restore and retract the vehicle's rear outriggers, and sends a message to the intelligent terminal that the hoisting operation is completed.

Images