Airport baggage automatic sorting system and method based on adaptive degree of freedom robot arm

Abstract

The application discloses an airport luggage automatic sorting system and method based on an adaptive degree-of-freedom mechanical arm, and the system comprises a mechanical arm body, a sensing module, a control module and an interface module. The mechanical arm body comprises six adjustable stiffness joints, adopts a pneumatic-electromagnetic hybrid driving design, and can dynamically adjust the degree of freedom; the sensing module collects luggage states and environmental information in real time; the control module runs a reinforcement learning algorithm based on a Transform framework; and the interface module exchanges data with an airport central control system through a ROS 2 middleware. The system solves the problems of low efficiency and insufficient adaptability of traditional mechanical arms, improves luggage processing efficiency and accuracy, and reduces deployment costs.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of airport baggage sorting, and particularly relates to an airport baggage automatic sorting system and method based on an adaptive degree-of-freedom mechanical arm. BACKGROUND

[0002] With the increasing busy operation of airports, the efficiency and accuracy of baggage handling systems have become key factors. Traditional baggage handling systems mainly rely on fixed degree-of-freedom mechanical arms or manual operation, which can complete tasks such as baggage sorting, transfer and loading to a certain extent, but have obvious efficiency bottlenecks and insufficient adaptability. On the one hand, although the fixed degree-of-freedom mechanical arm can perform basic grasping tasks, it is not flexible enough to handle complex environmental motion requirements, such as obstacle avoidance or high-precision baggage stacking operations, which affects the overall processing efficiency. On the other hand, manual operation depends on high skills and long-term repetitive labor, which is not only costly, but also prone to errors or injuries due to fatigue.

[0003] In order to solve the above problems, the prior art attempts to improve the control strategy to improve the flexibility and adaptability of the system. For example, a rule-based algorithm is used to cope with the diversity of baggage and the dynamic changes of the processing environment. However, this strategy is relatively single when faced with complex and variable actual situations, and lacks flexibility and the ability to respond to emergencies. Therefore, the existing system still has low integration between mechanical arm operation, sensor monitoring and baggage management system, which leads to system response delay or error accumulation, and requires manual intervention to handle sorting errors and other problems.

[0004] In summary, the existing baggage handling system has multiple limitations such as fixed degree of freedom, single control strategy and low system integration, and a new technical solution that can balance flexibility and accuracy and adapt to various environments is needed to improve the overall efficiency and safety of the airport baggage handling system. In view of this, the present application is conceived in depth and actively researched and improved to develop the present application in view of the many deficiencies and inconveniences caused by the existing baggage handling system. SUMMARY

[0005] The present application aims to overcome the shortcomings of the prior art and provide an airport baggage automatic sorting system and method based on an adaptive degree-of-freedom mechanical arm that can balance flexibility and accuracy and adapt to various environments, thereby improving the overall efficiency and safety of airport baggage handling.

[0006] In order to achieve the above purpose, the solution of the present application is:

[0007] An airport baggage automatic sorting system based on an adaptive degree-of-freedom mechanical arm, comprising: a mechanical arm body, a sensing module, a control module and an interface module;

[0008] The mechanical arm body adopts a modular design and comprises a plurality of joints with adjustable rigidity and a plurality of connecting rods, and each joint is connected with a control module;

[0009] The sensing module comprises an FBG sensor, a visual camera and an RFID reader, the FBG sensor is embeddedly installed in the mechanical arm body and connected with the control module, the visual camera is used to acquire the position information of the luggage, including the position information and the attitude information of the luggage, and the RFID reader is used to identify the luggage tag; the FBG sensor, the visual camera and the RFID reader are connected with the control module respectively;

[0010] The control module runs a reinforcement learning algorithm based on a Transformer architecture, dynamically adjusts the processing order according to the luggage priority and the system load, and outputs the joint rigidity instruction and the corresponding driving instruction to the mechanical arm body;

[0011] The interface module is integrated on the mechanical arm body, the joint module realizes data exchange with the luggage management system of the airport through a ROS 2 middleware connected with the control module, including input information and output information, the input information comprises a luggage trajectory, a luggage priority and a system state, and the output information is a loading state feedback.

[0012] Further, the joint adopts a hybrid driving mode of pneumatic-electromagnetic, the rigidity of the joint is determined by the following formula: = Kp + Ke

[0013] = × + × ;

[0014] Wherein, Pj is the air pressure value of the jth joint, the value range is 0-1MPa;

[0015] Ij is the electromagnetic coil current of the jth joint, the value range is 0-5A;

[0016] Kp is the pneumatic rigidity coefficient, which depends on the diaphragm material and structure of the joint;

[0017] Ke is the electromagnetic rigidity coefficient, which is related to the performance of the permanent magnet and the number of turns of the coil;

[0018] Kc is the critical value of the rigidity of the joint when the joint is locked;

[0019] When ≥ When the joint enters the locking state, the degree of freedom of the robot arm is reduced.

[0020] Further, the policy function π of the reinforcement learning algorithm is defined as = π( ),

[0021] wherein is the state observation at time t, including the luggage pose, the conveyor belt speed v, the joint angles and strain data of the robot arm;

[0022] is the action output, including the set values of the joint and angles of the robot arm;

[0023] θ is the policy network parameter, which is obtained by minimizing the loss function ζ:

[0024] θ = argmin ζ( , ),

[0025] Argmin represents the set of independent variable values that make the function take the minimum value; represents the discount factor value 0.99;

[0026] The loss function ζ takes into account the task completion time, energy consumption and operation accuracy:

[0027] ζ = + ‖ ‖

[0028] wherein the weight coefficients , , are determined by domain knowledge, denotes the pose of the luggage, denotes the target pose of the luggage.

[0029] Further, the input information of the interface module includes the luggage trajectory =[x,y,z,α,β,η];x,y,z represents the three-dimensional coordinates of the luggage on the conveyor belt, α is the yaw angle of the luggage, β is the pitch angle of the luggage, and η is the roll angle of the luggage;

[0030] Priority flag f∈{0,1} ;f=0 means low-priority luggage, and f=1 means high-priority luggage;

[0031] System state code c∈{0,...,255} ; represents the binary numerical identifier of the overall running state of the airport luggage handling system, with a value range of 0 to 255;

[0032] The output information is a loading state feedback s belongs to {0, 1}, when s=0, a system level error correction process is triggered.

[0033] Further, the mechanical arm body includes 6 joints with adjustable stiffness, and the connecting rods are carbon fiber reinforced polymers.

[0034] Further, the FBG sensors in the sensing module are embedded in the carbon fiber reinforced polymer connecting rods of the mechanical arm body, are closely combined with the connecting rods, and synchronously generate strain with the deformation of the connecting rods, the FBG sensors monitor the state of the mechanical arm body in real time, the strain data of the connecting rods are detected, the force conditions, motion postures and potential vibration risks of each joint of the mechanical arm body are indirectly obtained, and closed-loop feedback is provided for a control algorithm; when the mechanical arm body takes luggage with different weights or shapes, the FBG sensors feed back the strain of the connecting rods in real time, the auxiliary control module judges whether it is necessary to adjust the stiffness of the joints, and dynamic adjustment of the degrees of freedom is realized.

[0035] Further, the luggage management system and the ROS 2 middleware realize data interaction through a collaborative optimization protocol, and the collaborative optimization protocol dynamically allocates sorting tasks based on task priority and system load.

[0036] Further, the mechanical arm body, the sensing module, the control module and the interface module are connected through a wired or wireless mode, real-time data transmission and execution of control instructions are realized.

[0037] The application can also adopt the following technical solutions:

[0038] An airport luggage automatic sorting method based on the adaptive degree of freedom mechanical arm-based airport luggage automatic sorting system, comprising the following steps:

[0039] S1, luggage conveying stage: when the luggage is conveyed to the collection range of the sensing module, the RFID reader obtains the luggage information, the vision camera collects the accurate position and posture of the luggage, and sends the collected data to the control module; the control module calculates the optimal grabbing path according to the received data information, and configures the degrees of freedom of the mechanical arm body by adjusting the joint stiffness;

[0040] S2, luggage grabbing stage, the mechanical arm keeps high degree of freedom configuration to ensure flexibility, and grabs the luggage according to the optimal grabbing path calculated by the control module;

[0041] S3, luggage conveying and placing stage, the mechanical arm switches to low degree of freedom configuration to improve precision and stability after grabbing the luggage, and conveys and places the luggage through the optimal path;

[0042] During the whole process, the FBG sensors continuously monitor the strain state of the mechanical arm body to provide real-time feedback for the control algorithm.

[0043] Further, in the face of different weights and shapes of luggage, the mechanical arm body automatically adjusts the air pressure value and current value of the joints according to the guidance of the strategy function.

[0044] After the above scheme, the airport luggage automatic sorting system and method based on the adaptive degree of freedom mechanical arm of the present application have the following advantages compared with the prior art:

[0045] 1、The multiple joints of the mechanical arm body of the present application adopt a hybrid drive joint design, and a pneumatic-electromagnetic composite stiffness adjustment mechanism is created. Each joint adopts a pneumatic-electromagnetic hybrid drive mode, breaking through the response speed limit of the traditional single drive mode and realizing rapid stiffness adjustment of the joint. The joint can adjust its stiffness characteristics in real time according to different task requirements to solve the key limitations of the traditional mechanical arm degree of freedom and control strategy, and can improve the adaptability of the present application to complex luggage sorting scenarios. When grabbing irregularly shaped luggage, the joint stiffness can be reduced to make the mechanical arm more flexible; when carrying and placing luggage, the joint stiffness is increased to enhance stability. This real-time adjustment based on task requirements forms a programmable stiffness curve. The adjustment of joint stiffness can be completed in real time during the operation of the mechanical arm; the characteristics of the traditional mechanical arm joint are fixed or difficult to adjust, while the present application can change the joint characteristics in time according to the luggage state (position, shape, priority) and environmental conditions (conveyor speed, space limitation) to realize pre-reconfiguration. For example, when encountering luggage of different sizes and shapes, or when the conveyor speed changes, the mechanical arm can quickly adjust the joint stiffness and degree of freedom, optimize the motion strategy, and improve the efficiency and accuracy of luggage handling.

[0046] 2、The mechanical arm body of the present application adopts dynamic degree of freedom management, which can adjust the degree of freedom of the mechanical arm body in real time based on task requirements, uses FBG sensors to monitor the vibration state of the mechanical arm body in real time, feeds back to the control module, and can timely adjust the control strategy to improve the adaptability and stability of the mechanical arm body and ensure efficient and accurate luggage sorting.

[0047] 3、The control module of the application adopts an intelligent control architecture, combines a hybrid decision model of Transformer and reinforcement learning, and considers the design of a strategy function in view of the dynamic characteristics of the mechanical arm body. The joint stiffness and degrees of freedom of the mechanical arm body have an important influence on the dynamic characteristics thereof, and the strategy function is closely related to these factors. When the joint stiffness and degrees of freedom are adjusted according to the task requirements, the strategy function will change accordingly. In the grabbing stage, the mechanical arm needs to maintain a high degree of freedom configuration to ensure flexibility, at this time the strategy function will tend to give more rewards to flexible grabbing actions, such as allowing appropriate increase in energy consumption and time within a certain range, in order to successfully grab irregular-shaped or position-uncertain luggage. In the carrying and placing stage, the mechanical arm switches to a low degree of freedom configuration to improve precision and stability, and the strategy function pays more attention to position accuracy and reduces vibration, giving higher rewards to precise placing actions, and punishing behaviors that may cause accuracy to decrease, such as excessive joint movement amplitude or unstable posture.

[0048] The reinforcement learning algorithm based on the Transformer architecture optimizes the control strategy of the mechanical arm body through the strategy function. In the training process, the agent obtains reward feedback according to the action in different states, and continuously adjusts the strategy to maximize the cumulative reward. Since the strategy function takes into account the dynamic characteristics of the mechanical arm, the agent will gradually learn to select actions that conform to the laws of dynamics during the learning process. When facing luggage of different weights and shapes, the agent will automatically adjust the joint air pressure value and current value according to the guidance of the strategy function, to achieve the best grabbing and carrying effect, while meeting the energy consumption and accuracy requirements.

[0049] 4、The system-level innovation of the application includes the integration of FBG sensors and the mechanical arm body, which can continuously monitor the strain state of the mechanical arm body and provide real-time feedback for the control algorithm; through the interface module as a bridge for external communication of the application, the luggage management system and the mechanical arm body can be optimized cooperatively. The integration of adjustable stiffness joints (hybrid pneumatic-electromagnetic drive) and carbon fiber links, and the integration of carbon fiber links and FBG sensors provide a physical basis for dynamic degree of freedom adjustment, which is hardware; the distributed communication framework based on ROS 2 realizes real-time data interaction between the mechanical arm and the central system, ensuring seamless connection of sensing data, control instructions and task scheduling, which is a software architecture; the reinforcement learning model combines the dynamic characteristics of the mechanical arm and system-level constraints (luggage priority, conveyor speed, etc.), to generate a globally optimal strategy for algorithm optimization. Through the deep cooperation of hardware-software-algorithm, the efficiency and accuracy of luggage handling are significantly improved.

[0050] In summary, the present application proposes an innovative airport baggage automatic sorting system and method based on adaptive degree of freedom mechanical arm, which is specially designed for airport baggage automatic sorting and loading scene. The core innovation of the system lies in its dynamic degree of freedom adjustment capability and intelligent control strategy. By adjusting the joint stiffness and degree of freedom configuration of the mechanical arm in real time, seamless switching between flexible grabbing and accurate placing modes is realized. Through dynamic degree of freedom adjustment, the present application can quickly adapt to baggage of different sizes and shapes, solving the problem of grabbing failure caused by fixed structure of traditional mechanical arm. Through intelligent control strategy, the processing order can be dynamically adjusted according to the priority of baggage and system load, optimizing the overall throughput. Furthermore, the present application can be seamlessly integrated with existing baggage management infrastructure, greatly reducing the deployment cost while ensuring the reliability and maintainability of the system. Compared with traditional fixed degree of freedom mechanical arm, the present application can automatically optimize its motion strategy according to the real-time state (such as position, shape, priority) of baggage and environmental conditions (such as conveyor speed, space limitation), significantly improving the efficiency and accuracy of baggage handling. BRIEF DESCRIPTION OF DRAWINGS

[0051] Figure 1 Figure 1 is an integrated schematic diagram of the airport baggage handling system and adaptive degree of freedom mechanical arm of the present application.

[0052] Figure 2 Figure 2 is a structural schematic diagram of the adaptive degree of freedom mechanical arm system of the present application. DETAILED DESCRIPTION

[0053] In order to further explain the technical scheme of the present application, the present application will be described in detail through specific embodiments.

[0054] As shown in Figure 1 and Figure 2 , the present application discloses an airport baggage automatic sorting system based on adaptive degree of freedom mechanical arm, which is connected with a baggage handling system and a central management system. The baggage handling system is used to handle airport baggage processing, including check-in counter, baggage drop-off point, conveyor system and security equipment. The check-in counter is used for the initial link of passenger handling baggage consignment, generating baggage label information; the baggage drop-off point is a self-service baggage consignment device, which sends baggage into the conveyor system; the conveyor system is a core conveying equipment for transporting baggage to different processing links; the security equipment is used for security scanning of baggage to ensure compliance with aviation safety standards.

[0055] The central management system is the central control unit of airport baggage handling.

[0056] The airport baggage automatic sorting system based on adaptive degree of freedom mechanical arm of the present application includes a mechanical arm body, a sensing module, a control module and an interface module.

[0057] The mechanical arm body adopts a modular design and comprises six adjustable stiffness joint mechanisms and carbon fiber reinforced polymer connecting rods. The carbon fiber reinforced polymer connecting rod is a lightweight and high-strength mechanical arm structural component. The adjustable stiffness joint mechanism is a key executive component for realizing dynamic adjustment of the degrees of freedom of the mechanical arm. Each joint adopts a pneumatic-electromagnetic hybrid driving mode, breaks through the response speed limit of the traditional single driving mode, realizes rapid stiffness adjustment, and the pneumatic-electromagnetic hybrid driving of the application is an innovative driving mechanism. The stiffness of the joint is determined by the following formula: = Kp + Ke

[0058] = × + ×

[0059] Wherein, is the air pressure value (0-1MPa) of the jth joint;

[0060] is the electromagnetic coil current (0-5A) of the jth joint;

[0061] is the pneumatic stiffness coefficient, which depends on the diaphragm material and structure;

[0062] is the electromagnetic stiffness coefficient, which is related to the performance of the permanent magnet and the number of turns of the coil;

[0063] is the critical value of the stiffness of the joint when the joint is locked;

[0064] When ≥ , the joint enters the locked state, and the degrees of freedom of the mechanical arm are correspondingly reduced.

[0065] This design enables a single joint to complete the switching from complete flexibility to complete rigidity within 10ms, providing a hardware foundation for real-time adjustment of the degrees of freedom of the mechanical arm.

[0066] The sensing module is responsible for real-time acquisition of luggage state and environmental information, including FBG sensors, vision cameras and RFID readers.

[0067] The FBG sensor (fiber bragg grating sensor) is embedded in the carbon fiber reinforced polymer connecting rod of the mechanical arm body and is tightly combined with the connecting rod material and synchronously strained with the connecting rod deformation. The FBG sensor can monitor the state of the mechanical arm body in real time, indirectly obtain the force condition, motion posture and potential vibration risk of each joint of the mechanical arm through the strain data of the connecting rod, and provide closed-loop feedback for the control algorithm; when the mechanical arm grabs luggage of different weights or shapes, the FBG sensor feeds back the connecting rod strain in real time, helps the control module to judge whether it is necessary to adjust the joint stiffness (such as switching from a flexible state of high degrees of freedom to a rigid state of low degrees of freedom), ensures the stability of grabbing and the motion accuracy, and assists in realizing dynamic adjustment of degrees of freedom. Through long-term monitoring of abnormal fluctuations of strain data, fatigue or damage of the mechanical arm structure can be found in advance, the system failure risk is reduced, and the maintainability is improved.

[0068] The visual camera and the RFID reader can be deployed on the conveyor belt system or the periphery thereof, the visual camera is used to acquire the luggage pose information, including acquiring the position information and the attitude information of the luggage; and the RFID reader is used to identify the luggage tag.

[0069] The control module is the intelligent decision center of the application, the control module runs a reinforcement learning algorithm based on a Transformer architecture, dynamically adjusts the processing order according to the luggage priority and the system load, optimizes the overall throughput, and outputs joint stiffness instructions and pneumatic valve instructions to the mechanical arm body. The joint stiffness instruction is a control signal output to the mechanical arm body, and the pneumatic valve instruction is a precise control command for adjusting the air pressure.

[0070] The policy function π of the reinforcement learning algorithm (i.e. the agent) based on the Transformer architecture is defined as =π( ),

[0071] Wherein, is the state observation at time t, including the luggage pose, the conveyor belt speed v, the angle and strain data of each joint of the mechanical arm, etc.

[0072] is the action output, including the set values of the angles and strains of each joint and

[0073] θ is the policy network parameter, which is obtained by minimizing the loss function ζ:

[0074] θ= argmin ζ( , ),

[0075] Argmin represents the set of independent variable values that make the function take the minimum value; ​denotes the discount factor, set to 0.99;

[0076] The loss function ζ takes into account the task completion time, energy consumption and operation accuracy:

[0077] ζ = w1 * T + w2 * E + w3 * A + ‖ ‖

[0078] where the weight coefficients w1, w2 and w3 are determined by domain knowledge, , , which is determined by domain knowledge, denotes the pose of the luggage, denotes the target pose of the luggage.

[0079] Domain knowledge refers to professional experience, industry rules and actual needs in the airport luggage sorting scene. T denotes the task completion time, and the value logic is that the airport luggage processing needs to meet the flight schedule, and overtime may cause luggage missing or flight delay; E denotes the energy consumption (air pressure + current), and the value logic is that the energy consumption of the pneumatic system and electromagnetic coil directly affects the operating cost, especially in large-scale deployment (such as deploying hundreds of robotic arms in a single airport), and the performance and energy consumption need to be balanced; A denotes the operation accuracy, and the value logic is that insufficient luggage placement accuracy may cause the conveyor belt to be blocked or loaded incorrectly (such as luggage sliding out of the sorting grid), and w3 needs to be ensured to be less than the safety threshold.

[0080] , , The initial weights can be calibrated by expert experience, and the initial weights (such as w1:0.5, w2:0.3, w3:0.2) can be set by robotic arm control algorithm engineers and airport logistics experts to prioritize sorting efficiency and accuracy. , , The coefficients can also be optimized through field tests: during the trial operation stage in the airport, the coefficients are adjusted through measured data (such as sorting success rate under different weights, energy consumption curve), for example, if it is found that the robotic arm has a high risk of overload in a certain scene, w2 can be temporarily increased to reduce joint stiffness (reduce current / air pressure output).

[0081] The combination of weight coefficients and domain knowledge reflects the adaptation of the algorithm from “theoretical model” to “engineering landing”, ensuring that the control strategy meets the actual operation needs of the airport, and avoiding unfeasible solutions caused by pure data-driven, such as sacrificing the sorting efficiency of all luggage to pursue accuracy.

[0082] ​​​Discount Factor A coefficient for measuring the present value of future rewards, with a value range of [0, 1]. In this invention, the discount factor is valued at 0.99, mainly based on long-term revenue orientation, stability considerations, and industry practice, close to 1, indicating that the agent places more emphasis on long-term cumulative rewards, suitable for scenarios that require multi-step planning (such as robotic arm path optimization). In airport baggage sorting, the robotic arm needs to consider the coordination of current and subsequent tasks (such as avoiding excessive current energy consumption that prevents subsequent tasks from being executed), so it needs to encourage the algorithm to pursue long-term optimal strategies. A larger discount factor can reduce the sensitivity of the strategy to immediate noise (such as minor fluctuations in conveyor speed), making the robotic arm's movements smoother and more predictable, meeting the high reliability requirements of airport systems. The discount factor of 0.99 in this invention emphasizes the priority of long-term system stability.

[0083] The value of the discount factor strengthens the long-term stability and multi-task coordination capability of the system, and is a key parameter support for the dynamic degree of freedom optimization of the "intelligent control architecture". It forms a logical closed loop with the innovation point of "adjusting the strategy according to the priority of the luggage and the load of the system" in this invention.

[0084] Through the above design, the control algorithm of this invention can quickly respond to real-time states (such as obstacle avoidance in case of sudden obstacles), and can also optimize overall performance from a global perspective, significantly improving the intelligent level of the airport baggage sorting system.

[0085] The interface module is a bridge for the system of this invention to communicate with the outside world, and exchanges data with the baggage management system through the ROS 2 middleware, which is a communication protocol for integration with the airport management system.

[0086] The input information includes:

[0087] Luggage trajectory =[x, y, z, α, β, η]; x, y, z represent the three-dimensional coordinates of the luggage on the conveyor belt, α is the yaw angle of the luggage, β is the pitch angle of the luggage, and η is the roll angle of the luggage.

[0088] Priority flag f∈{0,1} ; f=0 indicates low-priority processed luggage, and f=1 indicates high-priority processed luggage;

[0089] System state code c∈{0,...,255} ; represents a binary numerical identifier of the overall operation state of the airport baggage handling system, with a value range of 0 to 255 (256 states in total).

[0090] The output information is a loading state feedback s e {0, 1}, when s = 0, the system-level error correction process is triggered. The loading state feedback and the system-level error correction process build a complete closed loop from "abnormal perception" to "recovery execution" through automatic fault detection, multi-level response and algorithm optimization, improve the efficiency and reliability of airport baggage handling.

[0091] Baggage trajectory parameters =[x, y, z, a, b, h] are the core inputs of the reinforcement learning algorithm, which judge the complexity of the baggage shape through the pose parameters, and drive the joint stiffness formula = × + × Calculate the optimal stiffness configuration; the central management system allocates the sorting task to the nearest robot arm according to the position parameters x, y and priority (f) through ROS 2. Through the comprehensive perception of the six parameters of the baggage trajectory, the precise modeling of the baggage state is realized, which provides a solid data foundation for the intelligent operation of the robot arm.

[0092] Through the interface module, the baggage management system and the present application can be optimized cooperatively to ensure the efficient operation of the airport baggage automatic sorting system. The cooperation of the two is realized through the ROS 2 middleware, including data interaction, task allocation and system feedback correction.

[0093] Data interaction: the ROS 2 middleware builds a data interaction bridge between the baggage management system and the robot arm, so that the control module receives information from the baggage management system, including baggage trajectory [x, y, z, a, b, h], priority flag f, and system state code c. These data provide key basis for the control module, such as accurate positioning of the grabbing position according to the baggage trajectory, determining the processing order according to the priority flag, and adjusting the working state flexibly according to the system state code. At the same time, the control module feeds back the loading state s to the baggage management system, if s = 0, the system-level error correction process is triggered, which ensures the accuracy of baggage handling.

[0094] Task allocation: the baggage management system allocates tasks reasonably according to the obtained baggage information and the real-time state (busy or idle) of the robot arm. For high-priority baggage, it is preferentially allocated to idle or high-efficiency robot arms to ensure fast processing. When there are multiple pieces of baggage and multiple robot arms, the system optimizes task scheduling by considering factors such as the load, position and destination of each robot arm, improving overall processing efficiency. For example, baggage going to the same flight is allocated to a specific robot arm, reducing the movement distance and time of the robot arm.

[0095] Dynamic adjustment of joint stiffness and freedom: the control module adjusts the joint stiffness and freedom according to the luggage information and system state provided by the luggage management system. When grabbing luggage with irregular shape or unstable position, high freedom and low joint stiffness are maintained to enhance the flexibility of the robot arm and adapt to complex grabbing requirements; during the carrying and placing of luggage, the robot arm is switched to a low freedom and high joint stiffness state to improve precision and stability and ensure accurate placement of the luggage.

[0096] System feedback and error correction mechanism: the robot arm feeds back the loading state to the luggage management system in real time, and if loading fails (s=0), the luggage management system quickly starts the error correction process. First, the robot arm is instructed to retry locally, and the grabbing path is re-planned and the grabbing parameters are adjusted; if the local retry fails, other robot arms or conveyors are coordinated to correct errors, such as scheduling adjacent robot arms to assist or adjusting the running state of the conveyor; if it still cannot be solved, manual intervention is triggered to notify the staff to handle it, ensuring the continuous and stable operation of the system.

[0097] The present application interacts with the luggage handling system and the intermediate management system through the interface module, and the distributed architecture of ROS2 allows the robot arm body to intervene in the existing luggage handling system and intermediate management system of the airport as an independent module, without the need for large-scale modification of infrastructure (such as conveyors and security equipment), reducing deployment costs and allowing seamless integration with existing luggage management infrastructure.

[0098] The present application also discloses an airport luggage automatic sorting method using the above-mentioned adaptive freedom-based robot arm airport luggage automatic sorting system, which comprises the following steps:

[0099] S1, luggage conveying stage: when the luggage is conveyed to the collection range of the sensing module, the RFID reader acquires the luggage information, the vision camera collects the accurate position and attitude of the luggage, and sends the collected data to the control module, which calculates the optimal grabbing path according to the received data information and configures the freedom of the robot arm body by adjusting the joint stiffness;

[0100] S2, luggage grabbing stage, the robot arm maintains a high freedom configuration to ensure flexibility and grabs the luggage according to the optimal grabbing path calculated by the control module;

[0101] S3, during the carrying and placing of the luggage, the robot arm switches to a low freedom configuration to improve precision and stability after grabbing the luggage, and carries and places the luggage through the optimal path;

[0102] Throughout the process, the FBG sensor continuously monitors the strain state of the robot arm body to provide real-time feedback for the control algorithm.

[0103] The value of the present application mainly embodies in three aspects: first, through dynamic degree of freedom adjustment, the system can quickly adapt to different sizes and shapes of luggage, solving the problem of grabbing failure caused by fixed structure of traditional mechanical arm; second, the intelligent control strategy can dynamically adjust the processing order according to the luggage priority and system load, optimizing the overall throughput; finally, the seamless integration of the system with the existing luggage management infrastructure greatly reduces the deployment cost, while ensuring the reliability and maintainability of the system.

[0104] The above embodiments and drawings are not limited to the product form and style of the present application, and any appropriate changes or modifications made by those skilled in the art shall be considered as not departing from the patent scope of the present application.

Claims

1. An airport luggage automatic sorting system based on adaptive degree of freedom robot arm, characterized in that, The application relates to a luggage handling robot, which comprises a mechanical arm body, a sensing module, a control module and an interface module. The mechanical arm body adopts a modular design and comprises a plurality of joints with adjustable rigidity and a plurality of connecting rods, and each joint is connected with the control module. The sensing module comprises FBG sensors, a visual camera and an RFID reader, the FBG sensors are embeddedly installed in the mechanical arm body and connected with the control module, the visual camera is used to acquire luggage position information, including position information and attitude information of the luggage, and the RFID reader is used to identify a luggage label; the FBG sensors, the visual camera and the RFID reader are connected with the control module. The control module runs a reinforcement learning algorithm based on a Transformer framework, dynamically adjusts a processing sequence according to luggage priority and system load, and outputs joint rigidity instructions and corresponding driving instructions to the mechanical arm body. The interface module is integrated on the mechanical arm body, and the joint module realizes data exchange with a luggage management system of an airport through a ROS 2 middleware connected with the control module, including input information and output information, the input information comprises luggage trajectory, luggage priority and system state, and the output information is a loading state feedback. Theta is a policy network parameter, which is obtained by minimizing a loss function zeta. The joint adopts a pneumatic-electromagnetic hybrid driving mode, and the stiffness of the joint is determined by the following formula: = × + × ； wherein, Pj is the air pressure value of the jth joint, with a value range of 0-1 MPa; The electromagnetic coil current of the jth joint is 0-5 A. Cp is the pneumatic stiffness coefficient, which depends on the diaphragm material and structure of the joint; For the electromagnetic stiffness coefficient, it is related to the permanent magnet performance and the number of turns of the coil. Joint stiffness threshold value at which the joint locks; When ≥ The joint enters the locking state, and the degree of freedom of the robot arm is reduced. The policy function p of the reinforcement learning algorithm is defined as = p( ), wherein, is the state observation at time t, including the luggage pose, the conveyor speed v, the angles of each joint of the robot arm, and the strain data. For action output, include each joint of and Set value; Argmin represents a set of independent variable values that make a function minimum. θ = argmin ζ( , ) The loss function zeta comprehensively considers task completion time, energy consumption and operation accuracy. represents a discount factor taking the value 0.99; The priority flag f is equal to 0 or 1; f=0 indicates low-priority luggage, and f=1 indicates high-priority luggage. ζ = + ‖ ‖ wherein the weight coefficients , , are determined by domain knowledge, denotes the pose of the luggage, denotes the target pose of the luggage.

2. The adaptive degree of freedom robot arm based airport baggage automatic sorting system according to claim 1, characterized in that: The input information of the interface module includes the luggage trajectory =[x, y, z, a, b, h]; x, y, z represent the three-dimensional coordinates of the luggage on the conveyor belt, a is the yaw angle of the luggage, b is the pitch angle of the luggage, and h is the roll angle of the luggage The system state code c is equal to 0, 1, 2 or 255; the system state code c is a binary numerical value indicating the overall operation state of the airport luggage handling system, and the value range is 0 to 255. The output information is a loading state feedback s, and when s=0, a system-level error correction process is triggered. The mechanical arm body comprises six joints with adjustable rigidity, and the connecting rod is a carbon fiber reinforced polymer.

3. The adaptive degree of freedom robot arm based airport baggage automatic sorting system according to claim 1, characterized in that: The FBG sensors in the sensing module are embeddedly installed in the carbon fiber reinforced polymer connecting rod of the mechanical arm body, are tightly combined with the connecting rod, synchronously generate strain along with deformation of the connecting rod, and monitor the state of the mechanical arm body in real time; through detection of strain data of the connecting rod, force conditions, motion attitudes and potential vibration risks of each joint of the mechanical arm body are indirectly acquired, and closed-loop feedback is provided for a control algorithm; when the mechanical arm body takes luggage with different weights or shapes, the FBG sensors feed back the strain of the connecting rod in real time, and the control module is assisted to judge whether the rigidity of the joint needs to be adjusted, so that dynamic adjustment of degrees of freedom is realized.

4. The adaptive degree of freedom based mechanical arm airport baggage automatic sorting system according to claim 3, characterized in that: The luggage management system and the ROS 2 middleware realize data interaction through a collaborative optimization protocol, and the collaborative optimization protocol dynamically allocates sorting tasks based on task priority and system load.

5. The adaptive degree of freedom based mechanical arm airport baggage automatic sorting system according to claim 3, characterized in that: The mechanical arm body, the sensing module, the control module and the interface module are connected through a wired or wireless mode, so that real-time data transmission and execution of control instructions are realized.

6. The adaptive degree of freedom robot based airport baggage automatic sorting system according to claim 1, wherein: The application further discloses a luggage handling method, which comprises the following steps:

7. The airport baggage automatic sorting method of the airport baggage automatic sorting system based on the adaptive degree-of-freedom robot arm according to any one of claims 1 to 6, characterized in that, ​ S1, luggage conveying stage: when the luggage is conveyed into the collection range of the sensing module, the RFID reader acquires the luggage information, the visual camera collects the accurate position and attitude of the luggage, and sends the collected data to the control module, the control module calculates the optimal grabbing path according to the received data information, and adjusts the joint stiffness to configure the degrees of freedom of the mechanical arm body; S2, luggage grabbing stage, the mechanical arm keeps high degree of freedom configuration to ensure flexibility, and grabs the luggage according to the optimal grabbing path calculated by the control module; S3, luggage conveying and placing stage, the mechanical arm switches to low degree of freedom configuration after grabbing the luggage to improve precision and stability, and conveys and places the luggage through the optimal path; During the whole process, the FBG sensor continuously monitors the strain state of the mechanical arm body, and provides real-time feedback for the control algorithm.

8. The airport baggage automatic sorting method according to claim 7, characterized in that: When facing luggage of different weights and shapes, the mechanical arm body automatically adjusts the air pressure value and current value of the joint according to the guidance of the strategy function.

Images