Intelligent storage RFID robot reading perception and trajectory optimization method and system

Abstract

The embodiment of the invention relates to the technical field of ultrahigh frequency passive RFID, in particular to an intelligent storage RFID robot reading perception and trajectory optimization method and system. According to the method, the position coordinates of the goods shelf are obtained by building the model and utilizing the label characteristic values, the interaction state of the UHF RFID robot and the goods shelf can be sensed, the distribution state of the goods is further sensed, and the number and the position of the missed goods are estimated. Furthermore, the advancing track of the UHF RFID robot system in the warehouse management scene is designed and optimized according to the sensing result, so that the reading accuracy and the checking efficiency of the mobile RFID system are improved.

Description

Technical Field

[0001] This invention belongs to the field of ultra-high frequency passive RFID technology, specifically relating to a method and system for intelligent warehouse RFID robot reading, sensing and trajectory optimization. Background Technology

[0002] In existing warehouse management scenarios, mobile RFID (Radio Frequency Identification) systems typically rely on pre-defined fixed paths at warehouse rack locations during inventory checks. They can only collect information about the storage area of ​​tags, lacking the ability to autonomously perceive the environment and understand the actual distribution, stacking density, and potential obstruction of goods. This limitation necessitates manual recalibration when warehouse rack layout changes, and in situations with densely packed or heavily obstructed tags, significant missed reads can occur, affecting the completeness and accuracy of goods information.

[0003] In the prior art, Chinese invention patent application CN202410570538.X proposes a three-dimensional warehouse UHF RFID robot positioning system and method. The system consists of a warehouse UHF RFID robot, shelf tracks, a reader / writer, tags, and an information processing unit. The positioning method involves an RFID reader / writer mounted on the warehouse UHF RFID robot, and RFID tags installed on the shelf tracks to store location information. The reader / writer transmits the read location information to the information processing unit, which then controls the movement of the warehouse UHF RFID robot based on the received location information. This method requires pre-storing location information in the tags. If the shelf layout is adjusted or the RFID tag position changes, the location information stored in the tags must be rewritten, increasing the complexity and workload of the maintenance system.

[0004] Chinese invention patent application CN202310114230.X proposes a three-dimensional target perception method based on a mobile RFID reader and dual tags. By collecting tag feature values, a rough position coordinate of the tag relative to the shelf is calculated, and the MRRDT algorithm is designed to optimize and obtain the precise three-dimensional position coordinate of the tag. However, this method only focuses on the positioning of a single tag and lacks the overall perception capability of the distribution of tags in a group. It is difficult to comprehensively reflect the actual status of multiple targets on the shelf, and if the tag density is high and tags are missed, the tag position cannot be accurately located.

[0005] Chinese invention patent application CN202311265445.8 proposes a method for estimating the number of tags using the collision waveform of the initial frame. This method constructs a mapping relationship between the waveform and the number of tags, employs a random forest model, uses the waveform as the feature set and the number of tags as the label space for training; for a dense number of unknown tags, the number of tags can be predicted simply by inputting their waveform. However, it only uses the waveform as a feature set to predict the number of tags, resulting in a relatively singular input feature.

[0006] Chinese invention patent application number CN201620946485.8 proposes a warehouse goods identification system based on mobile robot control. The system comprises a host computer, a mobile robot, a wireless transceiver, a PLC controller, a servo motor system, a walking wheel transmission system, a speed sensor, a GPS locator, an RFID reader, and indicator lights. The system relies on manual operation to collect tag information from goods, making it unable to perceive the distribution of goods on shelves, resulting in low intelligence and high labor costs.

[0007] In summary, existing mobile RFID robot systems lack the ability to perceive the distribution, stacking density, and potential obstructions of goods on shelves in warehouse management scenarios, making it difficult to achieve autonomous trajectory optimization for mobile RFID robot systems. The system's tag recognition rate still needs to be improved. Summary of the Invention

[0008] This invention provides a method and system for intelligent warehouse RFID robot reading, sensing, and trajectory optimization, to solve the technical problems in existing mobile UHF RFID robot systems in warehouse management scenarios, such as the lack of ability to sense the distribution, stacking density, and potential obstruction of goods on shelves, the difficulty in achieving autonomous trajectory optimization of mobile UHF RFID robot systems, and the need to improve the tag recognition rate of the system.

[0009] To achieve the above objectives, the present invention adopts the following technical solution: A method for intelligent warehousing RFID robot reading, sensing, and trajectory optimization includes the following steps: The UHF RFID robot moves along a preset trajectory and speed to obtain shelf tag data, cargo tag data, and MAC layer data of the cargo tags. Based on shelf label data and the trajectory and speed of the UHF RFID robot, a position perception model for the interaction between the UHF RFID robot and the shelf is constructed to obtain the position coordinates of the shelf positioning label and the UHF RFID robot. Based on the position coordinates of the shelf positioning label and the UHF RFID robot, the time when the UHF RFID robot moves to the center of the optimal reading area of ​​the label to be read is calculated. When the time is reached, the UHF RFID robot is controlled to stop moving. Based on shelf label data and goods label data, a goods label density distribution perception model is constructed to obtain the density distribution of goods labels. Based on the density distribution of goods labels, the shelf area to be read is divided into different levels of goods omission risk areas. Based on the different levels of goods omission risk areas and the MAC layer data of goods labels, the number and location distribution of omission goods labels are calculated. Based on different levels of cargo misread risk areas and the number and location distribution of misread cargo tags, the UHF RFID robot's trajectory within the optimal reading area of ​​the cargo tags to be read is adjusted.

[0010] The process of enabling the UHF RFID robot to move along a preset trajectory and speed to obtain shelf tag data, cargo tag data, and MAC layer data of the cargo tags involves the following steps: The UHF RFID robot moves along a preset trajectory and speed, and uses a reader mounted on the UHF RFID robot to transmit electromagnetic waves of a set frequency to the shelf tags at a fixed frequency through dual antennas. This obtains EPC, RSSI, phase, timestamp, and antenna ID data returned by the shelf positioning tag, shelf reference tag, and cargo tag. A spectrum analyzer is then used to capture the transmission waveform of the command interaction process between the reader and the cargo tag to obtain the MAC layer data of the cargo tag per unit time.

[0011] The calculation of the moment when the UHF RFID robot reaches the center of the optimal reading area for the tag on the goods to be read is specifically as follows: A three-dimensional Cartesian coordinate system is established with the initial position of the UHF RFID robot as the origin, where the X-axis represents the direction of travel of the UHF RFID robot, the Y-axis represents the direction of the antenna reading the tag, and the Z-axis represents the direction perpendicular to the ground and upwards; combined with the preset trajectory and speed of the UHF RFID robot, the position coordinates of the UHF RFID robot in the three-dimensional Cartesian coordinate system are obtained; the phase difference between the reflected signal from the shelf positioning tag and the transmitted signal from the reader is analyzed, and the change characteristics of the phase difference are calculated to determine the position coordinates of the shelf positioning tag; based on the position coordinates of the UHF RFID robot and the shelf positioning tag, an estimated value is calculated for the moment when the UHF RFID robot enters the optimal reading area for the tag on the goods to be read. By combining shelf positioning tag and cargo tag data, the RSSI data variation characteristics of cargo tag data are analyzed and processed to calculate the estimated time when the UHF RFID robot enters the optimal reading area of ​​the cargo tag to be read. ,right and By performing combined weighting, the accurate value of the moment when the UHF RFID robot enters the optimal reading area of ​​the tag on the goods to be read can be obtained. .

[0012] The sampling frequency of the UHF RFID robot running along a preset trajectory and speed is set to N. , express The location of the UHF RFID robot at all times. for UHF RFID robot in a three-dimensional Cartesian coordinate system Position coordinates on the axis for UHF RFID robot in a three-dimensional Cartesian coordinate system Position coordinates on the axis For UHF RFID robots in a three-dimensional Cartesian coordinate system The height on the axis; the operation of the UHF RFID robot is divided into three stages: the stage before entering the optimal reading area of ​​the tag on the goods to be read. The stage where the goods label to be read is in the optimal reading area. And the stage after leaving the optimal reading area of ​​the cargo label to be read. Location of UHF RFID robot Represented as:

[0013] in, The moment when the UHF RFID robot enters the optimal reading area for the tags on the goods to be read. The initial moment for the UHF RFID robot inventory operation. This indicates the position coordinates at the initial moment of the UHF RFID robot's inventory operation. The coordinates of the position at the end of the UHF RFID robot inventory sampling. This indicates the coordinates of the UHF RFID robot when it enters the optimal reading area of ​​the tag on the goods to be read. This indicates the moment when the UHF RFID robot leaves the optimal reading area of ​​the tag on the goods to be read. This indicates the coordinates of the UHF RFID robot's position when it leaves the optimal reading area of ​​the tag on the goods to be read. This indicates the end time of UHF RFID robot inventory sampling. as well as They are respectively , as well as UHF RFID robot Axis coordinates The speed of the UHF RFID robot.

[0014] Set shelf location tags as The coordinates of the shelf positioning tag are The antenna is ,in, Index the shelf location tags. For antenna index, Indicates that the shelf location tag is in The height coordinates of the axis are derived from the geometric relationships in a three-dimensional rectangular coordinate system:

[0015] In the formula Indicates antenna To shelf positioning label The distance between them Indicates the use of antenna At any moment The measured data is used to calculate the shelf positioning label. Horizontal distance to the UHF RFID robot Shelf location label exist The coordinates along the axis are used to perform phase ranging based on the shelf positioning tag. The phase data change characteristics are analyzed to calculate the distance from the dual antennas to the shelf positioning tag. The calculation formula for the phase ranging is as follows:

[0016] In the formula Indicates in Time Antenna Read the shelf positioning tag The theoretical phase value, This indicates the wavelength of the electromagnetic wave emitted by the reader. The distance from the dual antennas to the shelf positioning tag is used to calculate the shelf positioning tag's position. The position coordinates are:

[0017] Shelf positioning tags The position coordinates are:

[0018] In the formula, express Time-of-use shelf location tags exist Coordinates along the axial direction, express Time-of-use shelf location tags exist Coordinates along the axial direction, express Time-of-use shelf location tags exist Coordinates along the axial direction, express Time-of-use shelf location tags exist Coordinates along the axis.

[0019] Based on shelf reference labels and cargo label data, this study analyzes the data variation characteristics of RSSI, phase, and number of label reads per unit time for shelf reference labels in different areas under different label densities, and establishes a cargo label density distribution perception model. According to the cargo label density distribution perception model, the K-means clustering algorithm is used to determine the density level of the area to be read where the shelf reference labels are located, and obtains the density distribution of cargo labels in the shelf. Based on the density distribution of cargo labels in the shelf, the area to be read on the shelf is divided into different levels of cargo missed reading risk areas.

[0020] Based on different levels of cargo misread risk areas and the MAC layer data of cargo tags, the command interaction process between the reader and cargo tags is analyzed. The frame length, number of time slots in each frame, and time slot type data of the communication data generated by the command interaction per unit time are statistically analyzed. Based on the obtained frame length, number of time slots in each frame, and time slot type data of each frame per unit time, the number of misread cargo tags is calculated, and the location distribution of the misread cargo tags is statistically analyzed. The MAC layer data of the cargo tags is as follows:

[0021] In the formula, These are the number of frame length, idle time slots, success time slots, and collision time slots, respectively. These represent the proportions of idle time slots, successful time slots, and collision time slots in the total time slots, respectively.

[0022] Based on the MAC layer data of cargo tags, a random forest algorithm for estimating the number of missed tags is used to construct and combine multiple random decision trees for regression prediction. When constructing each decision tree, the random forest randomly selects a portion of features during the splitting process at each node. For each decision tree... Perform Bootstrap sampling, where, For the first in a random forest An independent decision tree, For the index of the decision tree, Let be the total number of decision trees in the random forest; at each node split, m candidate features are randomly selected from all features, and the optimal splitting feature and splitting threshold are selected according to the criterion of minimizing the mean squared error. The decision trees are continuously grown until the stopping condition is met; the trained decision trees are... Adding to the random forest model ensemble, the final result is obtained from... A trained decision tree The resulting set constitutes the trained random forest model, where... For the first A fully trained decision tree, The total number of decision trees trained; from the original MAC layer data of the goods tags. Extracting new feature vectors Input the trained random forest model, calculate the average output of all decision trees, and obtain the number of missed cargo tags:

[0023] In the formula, For the first The predicted values ​​of each decision tree. For the first in a random forest Each decision tree provides a feature vector for the input MAC layer. After processing, the output is a predicted value for the total number of goods tags in the current storage area. This represents the actual number of tags read.

[0024] Based on different levels of cargo misread risk areas and the number and location distribution of misread cargo tags, the trajectory of the UHF RFID robot within the optimal reading area of ​​the cargo tags to be read is adjusted. Specifically, the optimal reading area of ​​each cargo tag to be read is taken as a node to be visited. The access order of each area is encoded, and a genetic algorithm is used to optimize the path combination. The genetic algorithm includes initializing the chromosome population, calculating fitness based on the total path length and area priority, performing parent selection, crossover, and mutation operations, and obtaining the optimal access order after multiple generations of iteration. Based on the optimal access order, the local trajectory and the inter-area connection path are integrated to generate the trajectory of the UHF RFID robot within the optimal reading area of ​​the cargo tags to be read, starting from the center position of the optimal reading area of ​​the cargo tags to be read.

[0025] An intelligent warehousing RFID robot reading, sensing and trajectory optimization system includes a data processing module, a location sensing module, a cargo miss detection module and a trajectory optimization module; The data processing module is used to enable the UHF RFID robot to move along a preset trajectory and speed, and to obtain shelf tag data, cargo tag data, and MAC layer data of the cargo tags. The position sensing module is used to construct a position sensing model of the interaction between the UHF RFID robot and the shelf based on the shelf label data and the trajectory and speed of the UHF RFID robot, obtain the position coordinates of the shelf positioning label and the UHF RFID robot, calculate the time when the UHF RFID robot runs to the center of the optimal reading area of ​​the label to be read based on the shelf positioning label and the UHF RFID robot, and control the UHF RFID robot to stop moving when the time is reached. The cargo omission detection module is used to construct a cargo label density distribution detection model based on shelf label data and cargo label data, obtain the density distribution of cargo labels, divide the shelf area to be read into different levels of cargo omission risk areas based on the density distribution of cargo labels, and calculate the number and location distribution of omission cargo labels based on the different levels of cargo omission risk areas and the MAC layer data of cargo labels. The trajectory optimization module is used to adjust the UHF RFID robot's trajectory within the optimal reading area for the cargo tags to be read, based on different levels of cargo miss reading risk areas and the number and location distribution of miss reading cargo tags.

[0026] Compared with the prior art, the present invention has the following beneficial effects: In this invention, data collected by dual antennas is cross-checked, and the precise distance information reflected by phase and signal strength characteristics are combined to reduce the impact of electromagnetic interference and cargo obstruction on location calculations in the warehouse environment. This makes the shelf location coordinates more realistic, avoids omissions due to positioning deviations, and improves shelf positioning accuracy. Instead of relying on a single parameter to determine the optimal entry time for the UHF RFID robot, it integrates the changing patterns of phase and RSSI data to determine the accurate entry time, ensuring that the UHF RFID robot performs readings during the period of optimal tag signal strength, improving the success rate of single readings and accurately capturing the best reading opportunity. Furthermore, based on tag density distribution modeling, it identifies areas with dense or sparse tags on the shelves, and then uses K-means clustering to classify the risk level of missed readings, allowing high-risk areas to be given priority attention. This proactive approach to avoiding missed reading risks at the regional classification level ensures that the risk of missed readings is controllable. Further analysis of interactive data such as frame length and time slot type in the MAC layer allows for a clearer understanding of the number and location distribution of missed tags, rather than simply counting the number of read tags. This avoids the problem of discrepancies between inventory records and physical inventory after warehouse checks, where the source of the missed tags cannot be traced, and reduces the need for manual secondary verification.

[0027] Furthermore, in this invention, areas with different levels of missed read risk are designated as nodes to be visited. The access order is adjusted according to priority, with high-risk areas being processed first and their local paths optimized, while low-risk areas have simplified paths, avoiding redundant time consumption under fixed trajectories and shortening the overall inventory cycle. A genetic algorithm is used to optimize the connection paths between areas, reducing the UHF RFID robot's circuitous movement between different areas. Simultaneously, the trajectory is dynamically adjusted based on the missed read situation, eliminating the need for repeated travel to low-risk areas without missed reads, thus reducing the UHF RFID robot's energy consumption and equipment wear, improving inventory efficiency, and reducing resource waste. Attached Figure Description

[0028] Figure 1 This is a schematic diagram of a method for intelligent warehousing RFID robot reading, sensing, and trajectory optimization in an embodiment of the present invention; Figure 2 This is a schematic diagram of the UHF RFID robot reading perception and trajectory optimization method and system scenario in an embodiment of the present invention; Figure 3 This is a three-dimensional physical model of the position perception model of the UHF RFID robot interacting with the shelf in this embodiment of the invention; Figure 4 This is a diagram showing the error results of the shelf position coordinates in an embodiment of the present invention; Figure 5 This is a graph showing the time error results of the UHF RFID robot entering the optimal reading area of ​​the tag on the goods to be read in an embodiment of the present invention; Figure 6 This is a diagram showing the label density classification method based on the K-means clustering algorithm in an embodiment of the present invention. Figure 7 This is a diagram showing the classification of cargo misread risk areas at different levels in an embodiment of the present invention; Figure 8 This is a schematic diagram of the complete trajectory generation process of the intelligent warehouse RFID robot reading, sensing and trajectory optimization system in an embodiment of the present invention. Detailed Implementation

[0029] To further understand the content of this invention, the invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments are merely illustrative and not limiting of the invention.

[0030] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0031] Example 1 This embodiment proposes a method for intelligent warehousing RFID robot reading, sensing, and trajectory optimization, including the following steps: The UHF RFID robot moves along a preset trajectory and speed to obtain shelf tag data, cargo tag data, and MAC layer data of the cargo tags. Based on shelf tags and the trajectory and speed of the UHF RFID robot, a position perception model for the interaction between the UHF RFID robot and the shelf is constructed. Obtain the location coordinates of the shelf positioning tag and the UHF RFID robot. Calculate the time when the UHF RFID robot will reach the center of the optimal reading area for the tag to be read based on the shelf positioning tag and the UHF RFID robot location coordinates. When that time is reached, control the UHF RFID robot to stop moving. Based on shelf label data and goods label data, a goods label density distribution perception model is constructed to obtain the density distribution of goods labels. Based on the density distribution of goods labels, the shelf area to be read is divided into different levels of goods omission risk areas. Based on the different levels of goods omission risk areas and the MAC layer data of goods labels, the number and location distribution of omission goods labels are calculated. Based on different levels of cargo misread risk areas and the number and location distribution of misread cargo tags, the UHF RFID robot's trajectory is adjusted so that the cargo tags to be read are within the optimal reading area.

[0032] Based on the above-mentioned steps of an intelligent warehousing RFID robot reading, sensing, and trajectory optimization method, combined with Figure 1 A flowchart illustrating a UHF RFID robot reading, sensing, and trajectory optimization method for an intelligent warehouse management scenario is presented. The method combines a computing server, shelves and goods, a reader / writer, an antenna, tags, and a spectrum analyzer to complete the following steps. Further, the data acquisition of multiple feature values ​​of the tags is completed by the reader / writer mounted on the UHF RFID robot, the MAC layer data acquisition of the goods tags is completed by the spectrum analyzer mounted on the UHF RFID robot, and the relevant calculations are performed by the computing server. Specifically, steps S100 to S106 are as follows: In step S100, the UHF RFID robot moves along a preset trajectory and speed, and uses the reader mounted on the UHF RFID robot to emit electromagnetic waves of a set frequency to the shelf, thereby obtaining shelf tag data, goods tag data, and MAC layer data of the goods tags. The shelf tag data includes shelf positioning tags and shelf reference tags, and the goods tag data includes EPC (Electronic Product Code), RSSI (Received Signal Strength Indicator), phase, timestamp, and antenna ID data returned by the goods tags.

[0033] Step S101: Use the spectrum analyzer mounted on the UHF RFID robot to obtain the MAC layer data of the cargo tag per unit time.

[0034] Step S102: Based on the shelf positioning tag obtained in step S100 and the preset trajectory and speed of the UHF RFID robot, construct a position perception model of the interaction between the UHF RFID robot and the shelf, and obtain the position coordinates of the shelf positioning tag and the position coordinates of the UHF RFID robot.

[0035] Step S103: Combine the UHF RFID robot position coordinates and the shelf positioning tag position coordinates to calculate the time when the UHF RFID robot enters the optimal reading area of ​​the tag to be read.

[0036] Step S104: Based on shelf reference labels and cargo label data, construct a cargo label density distribution perception model, analyze the data change characteristics of RSSI, phase, and number of label reads per unit time of shelf reference labels in different areas under different label densities, obtain the distribution of cargo labels in the shelf, and use the cargo label density distribution perception model that associates the location of shelf reference labels with cargo label density to divide the shelf to be read area into different levels of cargo misread risk areas.

[0037] Step S105: Based on different levels of cargo misread risk areas, and by statistically analyzing the time slot information set in the MAC layer data of cargo tags, the number and location distribution of misread cargo tags are obtained.

[0038] Step S106: Based on different levels of cargo misread risk areas and the number and location distribution of misread cargo tags, design and optimize the movement trajectory of the UHF RFID robot in the warehouse management scenario when the cargo tags to be read are in the optimal reading area.

[0039] Specifically, according to steps S100 to S101, the scenario diagram of the UHF RFID robot reading perception and trajectory optimization method and system in the intelligent warehouse management scenario is as follows: Figure 2 As shown, a UHF RFID robot moves along a preset trajectory and speed in a warehouse management scenario. Using a reader mounted on the UHF RFID robot, it emits electromagnetic waves of a set frequency towards the shelves, obtaining data such as EPC, RSSI, phase, timestamp, and antenna ID returned by the shelf positioning tags, shelf reference tags, and product tags. The shelf has 5 layers, is 1.25 meters long, 2 meters high, and has a layer height of 0.4 meters. The shelf positioning tags (… , ) Set at a height of 1m on the left and right sides of the shelf support, shelf reference label ( , , , , , Six zones are set up on the shelf, with product tags affixed to the product surfaces. Multiple products are randomly placed on any shelf level, forming a shelf area to be read. A mobile UHF RFID robot equipped with dual antennas moves from the right side to the left side of the shelf at a speed of 0.2 m / s, with the dual antennas continuously reading tag data throughout the process. Simultaneously, a spectrum analyzer captures the interaction between the reader and the tags, demodulating the interaction signals to analyze time slot information.

[0040] In step S102, based on the shelf positioning tag obtained in step S100 and the preset trajectory and speed of the UHF RFID robot, a position perception model of the interaction between the UHF RFID robot and the shelf is constructed to obtain the shelf tag position coordinates. Specifically, an example diagram of the three-dimensional physical model of the position perception model of the UHF RFID robot interacting with the shelf is shown below. Figure 3 As shown, a three-dimensional Cartesian coordinate system is established with the initial position of the UHF RFID robot as the origin. The X-axis represents the movement direction of the UHF RFID robot, the Y-axis represents the direction the antenna reads tags, and the Z-axis is perpendicular to the ground and upwards. The sampling frequency during the UHF RFID robot's inventory operation is set to N. express The location of the UHF RFID robot at all times, among which, for UHF RFID robot in a three-dimensional Cartesian coordinate system Position coordinates on the axis for UHF RFID robot in a three-dimensional Cartesian coordinate system Position coordinates on the axis For UHF RFID robots in a three-dimensional Cartesian coordinate system Height on the axis; This indicates the moment when the RFID UHF RFID robot enters the optimal reading area for the tag on the goods to be read. This indicates the moment the UHF RFID robot leaves the optimal reading area. The UHF RFID robot inventory process consists of three stages: the stage before entering the optimal reading area for the tags on the goods to be read. The stage where the goods label to be read is in the optimal reading area. The stage after leaving the optimal reading area of ​​the cargo label to be read Then the UHF RFID robot position coordinates The matrix can be represented as:

[0041] in, This indicates the coordinates of the UHF RFID robot when it enters the optimal reading area of ​​the tag on the goods to be read. This indicates the moment when the UHF RFID robot leaves the optimal reading area of ​​the tag on the goods to be read. This indicates the coordinates of the UHF RFID robot's position when it leaves the optimal reading area of ​​the tag on the goods to be read. This indicates the end time of the UHF RFID robot inventory sampling. The coordinates of the position at the end of the UHF RFID robot inventory sampling. Indicates the height coordinates of the UHF RFID robot. as well as They are respectively , as well as UHF RFID robot Axis coordinates The moment when the UHF RFID robot enters the optimal reading area for the tags on the goods to be read. The initial moment of the UHF RFID robot inventory operation. This indicates the position coordinates at the initial moment of the UHF RFID robot's inventory operation. The speed of the UHF RFID robot is determined. Phase ranging is performed based on the shelf positioning tag obtained in step S100. The phase data change characteristics are analyzed, and after phase data processing, the distance from the dual antennas to the shelf positioning tag is calculated. The position coordinates of the shelf positioning tag are calculated in a three-dimensional Cartesian coordinate system. Specifically, the shelf positioning tag is set as... The coordinates of the shelf positioning tag are The antenna is ,in, Index the shelf location tags. For antenna index, Indicates that the shelf location tag is in The height coordinates of the axis can be derived from the geometric relationships in a three-dimensional rectangular coordinate system:

[0042] In the formula Indicates antenna To shelf positioning label The distance between them Indicates the use of antenna At any moment The shelf positioning label is obtained after calculating the measured data. Horizontal distance between the robot and the UHF RFID robot Shelf location label exist The coordinates along the axial direction. The phase ranging formula is as follows:

[0043] In the formula Indicates in Time Antenna Read the shelf positioning tag The theoretical phase value, This indicates the wavelength of the electromagnetic wave emitted by the reader. The distance from the dual antennas to the shelf positioning tag is calculated to determine the shelf positioning tag's position. The position coordinates are:

[0044] Shelf positioning tags The position coordinates are:

[0045] In the formula, express Time-of-use shelf location tags exist Coordinates along the axial direction, express Time-of-use shelf location tags exist Coordinates along the axial direction, express Time-of-use shelf location tags exist Coordinates along the axial direction, express Time-of-use shelf location tags exist Coordinates along the axial direction, Indicates the label is in Axis coordinates.

[0046] According to shelf location labels and Get the coordinates of the shelf label location The error results of the shelf label position coordinates are shown in the figure below. Figure 4 As shown.

[0047]

[0048] In step S103, the time when the UHF RFID robot enters the optimal reading area of ​​the tag on the goods to be read is calculated based on the UHF RFID robot's position coordinates. Based on the shelf positioning tags collected in step S100, the RSSI data change characteristics are analyzed and data processing is performed to calculate an estimated value of the time when the UHF RFID robot enters the optimal reading area of ​​the tag on the goods to be read. Taking the optimal reading area of ​​a UHF RFID robot entering a cargo tag to be read as an example, based on the system's physical characteristics, in Within the range , and Within the range ,in, The straight-line distance from the UHF RFID robot to the first shelf positioning tag. This is the straight-line distance for the UHF RFID robot to reach the second shelf positioning tag. Therefore, when the UHF RFID robot is located... Location, i.e., when the UHF RFID robot is at the boundary of the optimal reading area for the tag on the goods to be read, has The UHF RFID robot acquires data during its movement. and The dataset is as follows:

[0049] Calculate the optimal reading time for the UHF RFID robot to enter the optimal reading area for the tag on the goods to be read. This is based on the data acquired by the UHF RFID robot during its movement. and The equation uses a zero-point cross-estimation algorithm based on linear interpolation to obtain an estimate of the optimal reading time when the UHF RFID robot enters the tag reading area of ​​the goods to be read. Define the antenna. In time Collected tags The signal strength is The RSSI values ​​collected during the RFID UHF RFID robot inventory operation are represented in a matrix as follows:

[0050] This can be further expressed as:

[0051] After data processing, RSSI selects its local extreme points as effective features, and obtains...

[0052] The estimated time when the UHF RFID robot enters the optimal reading area of ​​the tag on the goods to be read is calculated. A time estimation algorithm combining Multiple Linear Regression (MLR) and Ordinary Least Squares (OLS) is used to estimate time based on two independent time estimates obtained from different physical feature values. and Establish the following MLR (Multiple Linear Regression Model):

[0053] In the formula and These are the two input features of the MLR regression model, representing the preliminary estimates obtained by the two methods mentioned above, i.e., the independent variables; The intercept term characterizes the inherent bias of the system. and These are regression coefficients, reflecting the effect of each independent variable on... Contribution weight; The residual term represents the variance that the MLR regression model failed to explain. The weights of the MLR regression model were estimated using OLS to determine the time when the UHF RFID robot entered the optimal reading area of ​​the tag on the goods to be read. The time error result of the UHF RFID robot entering the optimal reading area of ​​the tag to be read is shown in the figure below. Figure 5 As shown.

[0054] In step S104, a cargo label density distribution perception model is constructed to analyze the distribution of cargo labels on the shelves and obtain the density distribution of cargo labels. Based on the density distribution of cargo labels, the location of shelf reference labels is used to associate the cargo label density distribution perception model, and the shelf area to be read is divided into different levels of cargo misread risk areas. Based on the shelf reference label and cargo label data obtained in step S100, the data change characteristics of RSSI, phase, and number of label reads per unit time of shelf reference labels in different areas under different label densities are analyzed. A cargo label density perception model is established, and a cargo label density classification method based on the K-means clustering algorithm is adopted. Using the RSSI and number of label reads per unit time of shelf reference labels and cargo labels under different density scenarios as input data, the cargo density of the area to which the shelf reference label belongs is obtained. The cargo density clustering result is shown in the figure below. Figure 6 As shown in the figure. The shelf area to be read is further divided into different levels of goods omission risk areas, as shown in the figure below. Figure 7 As shown.

[0055] In step S105, the time slot information set in the MAC layer data of the cargo tags is statistically analyzed. Based on different levels of cargo miss reading risk areas, the number and location distribution of miss-read cargo tags are calculated. Using the MAC layer data of the cargo tags obtained in step S101 within a unit of time, the command interaction process between the reader and the tags is analyzed. The number of time slots and the time slot types in each frame of the communication data generated by the command interaction within a unit of time are statistically analyzed. The time slot types include success time slots, collision time slots, and idle time slots. The MAC layer data of the cargo tags is as follows:

[0056] in These are the number of frame length, idle time slots, success time slots, and collision time slots, respectively. These represent the proportions of idle time slots, successful time slots, and collision time slots in the total time slots. A missed tag estimation algorithm based on MAC layer states and random forests is used to estimate the number and location distribution of missed cargo tags. Regression prediction is performed by constructing and combining multiple random decision trees. When constructing each decision tree, the random forest randomly selects a subset of features during the splitting process at each node. For each decision tree... Perform Bootstrap sampling, where, For the first in a random forest An independent decision tree, For the index of the decision tree, Let be the total number of decision trees in the random forest. At each node split, m candidate features are randomly selected from all features. The optimal splitting feature and splitting threshold are chosen based on the Minimum Mean Squared Error (MSE) criterion. The decision tree continues to grow until the stopping condition is met. The trained decision tree... Adding to the random forest model ensemble, the final result is obtained from... A trained decision tree The set This refers to the trained random forest model, where... For the first A fully trained decision tree, This represents the total number of completed decision trees. It is derived from the original cargo label MAC layer feature information. Extracting new feature vectors Input the trained random forest model, calculate the average output of all decision trees, and obtain the number of missed cargo tags:

[0057] In the formula, For the first The predicted values ​​of each decision tree. For the first in a random forest Each decision tree provides a feature vector for the input MAC layer. After processing, the output is a predicted value for the total number of goods tags in the current storage area. This represents the actual number of tags read.

[0058] Step S106 involves designing and optimizing the complete travel trajectory of the UHF RFID robot system in a warehouse management scenario. Based on different levels of cargo omission risk areas and the number and location distribution of omission cargo tags, priority is given to identifying key scanning locations within high-risk areas. Combining the UHF RFID robot's travel speed and turning radius, an ant colony algorithm is used to plan the optimal travel path that covers all high-omission risk areas within the shelf area to be read, ensuring complete cargo tag reading. The optimal reading area for each cargo tag is designated as a node to be visited. By encoding the access order of each area, a genetic algorithm is used to optimize the path combination. The genetic algorithm includes initializing the chromosome population, calculating fitness based on the total path length and area priority, performing parent selection, crossover, and mutation operations. After multiple generations of iteration, the optimal access order is obtained. Based on this order, the local travel trajectory and inter-area connection paths are integrated to generate the optimal travel trajectory of the UHF RFID robot system in the warehouse management scenario. A schematic diagram of the complete travel trajectory generation process of the UHF RFID robot system in the warehouse management scenario is shown below. Figure 8 As shown.

[0059] In summary, this invention proposes an intelligent warehousing RFID robot reading perception and trajectory optimization method. It constructs two core models: a UHF RFID robot interaction with shelf positioning and a cargo tag distribution perception model. By fully utilizing the features of the tag's physical layer and MAC layer, it achieves dynamic perception of shelf positioning, cargo distribution, and the risk of missed readings. Based on the state perception results, ant colony optimization and genetic algorithms are used to design and optimize the optimal trajectory of the UHF RFID robot system in intelligent warehousing management scenarios. This invention achieves multi-level state perception and enhanced autonomous interaction capabilities for a mobile RFID system from the physical layer to the MAC layer, significantly improving the reading accuracy and inventory efficiency of RFID UHF RFID robot systems in intelligent retail, smart warehousing, and other application scenarios. It possesses a high level of intelligence and broad application potential.

[0060] Example 2 Based on the intelligent warehousing RFID robot reading perception and trajectory optimization method proposed in Embodiment 1, this embodiment proposes an intelligent warehousing RFID robot reading perception and trajectory optimization system, including a data processing module, a location perception module, a cargo omission perception module, and a trajectory optimization module. The data processing module equips the UHF RFID robot with an ultra-high frequency reader and dual antennas, and simultaneously provides a spectrum analyzer. It presets the initial trajectory and speed of the UHF RFID robot along the shelf aisle. When the UHF RFID robot moves along the preset trajectory and speed, the reader emits electromagnetic waves at a fixed frequency to the shelf tags, collecting shelf tag data and goods tag data. The spectrum analyzer captures the command interaction transmission waveform between the reader and the goods tags, extracts the MAC layer data of the goods tags per unit time, and transmits all data to the system database in real time.

[0061] The position sensing module establishes a three-dimensional rectangular coordinate system with the initial starting position of the UHF RFID robot as the origin. Combining the preset trajectory and speed, it calculates the position of the UHF RFID robot in the coordinate system in real time. Based on the collected shelf positioning tag data, it analyzes and processes the phase change characteristics to calculate the distance from the dual antennas to the positioning tags. Then, combined with the position of the UHF RFID robot, it calculates the position coordinates of the two shelf positioning tags to determine the spatial position of the entire shelf group. It calculates the estimated time when the UHF RFID robot enters the optimal reading area using phase data and RSSI data, respectively. It then assigns weights to the two estimated values ​​to obtain the accurate entry time and simultaneously records the center position of the optimal reading area of ​​the shelf corresponding to the UHF RFID robot at that time.

[0062] The cargo omission detection module constructs a cargo label density distribution detection model based on cargo label data and shelf location coordinates, analyzes the RSSI, phase, and number of reads per unit time of labels in different shelf areas, and uses the K-means clustering algorithm to divide the shelf area to be identified into three omission risk levels: high, medium, and low. Combining the features of different risk level areas with MAC layer data, it calls a pre-trained random forest model, inputs the MAC layer data of cargo labels, and outputs the number of omission labels. At the same time, based on the risk area division results, it matches the shelf area corresponding to the omission labels to determine the distribution of omission locations.

[0063] The trajectory optimization module uses the optimal reading area of ​​each shelf as the node to be visited, and encodes the node access order according to the priority of high-risk areas first, medium-risk areas second, and low-risk areas last. A genetic algorithm is used to initialize the chromosome population, calculate the fitness of each path, and obtain the optimal access order after multiple generations of selection, crossover, and mutation operations. Based on the optimal access order, the local trajectories of each area and the connecting paths between areas are integrated to generate the final travel trajectory of the UHF RFID robot, which is then sent to the UHF RFID robot control system to guide the UHF RFID robot to complete the warehouse inventory count according to the optimized trajectory.

[0064] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can be appropriately combined to form other embodiments that can be understood by those skilled in the art. The above content is only for illustrating the technical concept of the present invention and should not be construed as limiting the scope of protection of the present invention. Any modifications made based on the technical concept proposed in this invention shall fall within the scope of protection of the claims of this invention.

Claims

1. A method for intelligent warehousing RFID robot reading, sensing, and trajectory optimization, characterized in that, Includes the following steps: The UHF RFID robot moves along a preset trajectory and speed to obtain shelf tag data, cargo tag data, and MAC layer data of the cargo tags. Based on shelf label data and the trajectory and speed of the UHF RFID robot, a position perception model for the interaction between the UHF RFID robot and the shelf is constructed to obtain the position coordinates of the shelf positioning label and the UHF RFID robot. Based on the position coordinates of the shelf positioning label and the UHF RFID robot, the time when the UHF RFID robot moves to the center of the optimal reading area of ​​the label to be read is calculated. When the time is reached, the UHF RFID robot is controlled to stop moving. Based on shelf label data and goods label data, a goods label density distribution perception model is constructed to obtain the density distribution of goods labels. Based on the density distribution of goods labels, the shelf area to be read is divided into different levels of goods omission risk areas. Based on different levels of cargo misread risk areas and MAC layer data of cargo tags, the number and location distribution of misread cargo tags are calculated. Based on different levels of cargo misread risk areas and the number and location distribution of misread cargo tags, the UHF RFID robot's trajectory within the optimal reading area of ​​the cargo tags to be read is adjusted.

2. The intelligent warehousing RFID robot reading, sensing, and trajectory optimization method according to claim 1, characterized in that, The process of enabling the UHF RFID robot to move along a preset trajectory and speed to obtain shelf tag data, cargo tag data, and MAC layer data of the cargo tags involves the following steps: The UHF RFID robot moves along a preset trajectory and speed, and uses a reader mounted on the UHF RFID robot to transmit electromagnetic waves of a set frequency to the shelf tags at a fixed frequency through dual antennas. This obtains EPC, RSSI, phase, timestamp, and antenna ID data returned by the shelf positioning tag, shelf reference tag, and cargo tag. A spectrum analyzer is then used to capture the transmission waveform of the command interaction process between the reader and the cargo tag to obtain the MAC layer data of the cargo tag per unit time.

3. The intelligent warehousing RFID robot reading, sensing, and trajectory optimization method according to claim 1, characterized in that, The calculation of the moment when the UHF RFID robot reaches the center of the optimal reading area for the tag on the goods to be read is specifically as follows: A three-dimensional Cartesian coordinate system is established with the initial position of the UHF RFID robot as the origin, where the X-axis represents the direction of travel of the UHF RFID robot, the Y-axis represents the direction of the antenna reading the tag, and the Z-axis represents the direction perpendicular to the ground and upwards; combined with the preset trajectory and speed of the UHF RFID robot, the position coordinates of the UHF RFID robot in the three-dimensional Cartesian coordinate system are obtained; the phase difference between the reflected signal from the shelf positioning tag and the transmitted signal from the reader is analyzed, and the change characteristics of the phase difference are calculated to determine the position coordinates of the shelf positioning tag; based on the position coordinates of the UHF RFID robot and the shelf positioning tag, an estimated value is calculated for the moment when the UHF RFID robot enters the optimal reading area for the tag on the goods to be read. By combining shelf positioning tag and cargo tag data, the RSSI data variation characteristics of cargo tag data are analyzed and processed to calculate the estimated time when the UHF RFID robot enters the optimal reading area of ​​the cargo tag to be read. ,right and By performing combined weighting, the accurate value of the moment when the UHF RFID robot enters the optimal reading area of ​​the tag on the goods to be read can be obtained. .

4. The intelligent warehousing RFID robot reading, sensing, and trajectory optimization method according to claim 3, characterized in that, The sampling frequency of the UHF RFID robot running along a preset trajectory and speed is set to N. , express The location of the UHF RFID robot at all times. for UHF RFID robot in a three-dimensional Cartesian coordinate system Position coordinates on the axis for UHF RFID robot in a three-dimensional Cartesian coordinate system Position coordinates on the axis For UHF RFID robots in a three-dimensional Cartesian coordinate system The height on the axis; the operation of the UHF RFID robot is divided into three stages: the stage before entering the optimal reading area of ​​the tag on the goods to be read. The stage where the goods label to be read is in the optimal reading area. And the stage after leaving the optimal reading area of ​​the cargo label to be read. Location of UHF RFID robot Represented as: in, The moment when the UHF RFID robot enters the optimal reading area for the tags on the goods to be read. The initial moment of the UHF RFID robot inventory operation. This indicates the position coordinates at the initial moment of the UHF RFID robot's inventory operation. The coordinates of the position at the end of the UHF RFID robot inventory sampling. This indicates the coordinates of the UHF RFID robot when it enters the optimal reading area of ​​the tag on the goods to be read. This indicates the moment when the UHF RFID robot leaves the optimal reading area of ​​the tag on the goods to be read. This indicates the coordinates of the UHF RFID robot's position when it leaves the optimal reading area of ​​the tag on the goods to be read. This indicates the end time of UHF RFID robot inventory sampling. as well as They are respectively , as well as UHF RFID robot Axis coordinates The speed of the UHF RFID robot.

5. The intelligent warehousing RFID robot reading, sensing, and trajectory optimization method according to claim 4, characterized in that, Set shelf location tags as The coordinates of the shelf positioning tag are The antenna is ,in, Index the shelf location tags. For antenna index, Indicates that the shelf location tag is in The height coordinates of the axis are derived from the geometric relationships in a three-dimensional rectangular coordinate system: In the formula Indicates antenna To shelf positioning label The distance between them Indicates the use of antenna At any moment The measured data is used to calculate the shelf positioning label. Horizontal distance to the UHF RFID robot Shelf location label exist The coordinates along the axis are used to perform phase ranging based on the shelf positioning tag. The phase data change characteristics are analyzed to calculate the distance from the dual antennas to the shelf positioning tag. The calculation formula for the phase ranging is as follows: In the formula Indicates in Time Antenna Read the shelf positioning tag The theoretical phase value, This indicates the wavelength of the electromagnetic wave emitted by the reader. The distance from the dual antennas to the shelf positioning tag is used to calculate the shelf positioning tag's position. The position coordinates are: Shelf positioning tags The position coordinates are: In the formula, express Time-of-use shelf location tags exist Coordinates along the axial direction, express Time-of-use shelf location tags exist Coordinates along the axial direction, express Time-of-use shelf location tags exist Coordinates along the axial direction, express Time-of-use shelf location tags exist Coordinates along the axis.

6. The intelligent warehousing RFID robot reading, sensing, and trajectory optimization method according to claim 2, characterized in that, Based on shelf reference labels and cargo label data, this study analyzes the data variation characteristics of RSSI, phase, and number of label reads per unit time for shelf reference labels in different areas under different label densities, and establishes a cargo label density distribution perception model. According to the cargo label density distribution perception model, the K-means clustering algorithm is used to determine the density level of the area to be read where the shelf reference labels are located, and obtains the density distribution of cargo labels in the shelf. Based on the density distribution of cargo labels in the shelf, the area to be read on the shelf is divided into different levels of cargo missed reading risk areas.

7. The intelligent warehousing RFID robot reading, sensing, and trajectory optimization method according to claim 6, characterized in that, Based on different levels of cargo misread risk areas and the MAC layer data of cargo tags, the command interaction process between the reader and cargo tags is analyzed. The frame length, number of time slots in each frame, and time slot type data of the communication data generated by the command interaction per unit time are statistically analyzed. Based on the obtained frame length, number of time slots in each frame, and time slot type data of each frame per unit time, the number of misread cargo tags is calculated, and the location distribution of the misread cargo tags is statistically analyzed. The MAC layer data of the cargo tags is as follows: In the formula, These are the number of frame length, idle time slots, success time slots, and collision time slots, respectively. These represent the proportions of idle time slots, successful time slots, and collision time slots in the total time slots, respectively.

8. The intelligent warehousing RFID robot reading, sensing, and trajectory optimization method according to claim 7, characterized in that, Based on the MAC layer data of cargo tags, a random forest algorithm for estimating the number of missed tags is used to construct and combine multiple random decision trees for regression prediction. When constructing each decision tree, the random forest randomly selects a portion of features during the splitting process at each node. For each decision tree... Perform Bootstrap sampling, where, For the first in a random forest An independent decision tree, For the index of the decision tree, Let be the total number of decision trees in the random forest; at each node split, m candidate features are randomly selected from all features, and the optimal splitting feature and splitting threshold are selected according to the criterion of minimizing the mean squared error. The decision trees are continuously grown until the stopping condition is met; the trained decision trees are... Adding to the random forest model ensemble, the final result is obtained from... A trained decision tree The resulting set constitutes the trained random forest model, where... For the first A fully trained decision tree, The total number of decision trees trained; from the original MAC layer data of the goods tags. Extracting new feature vectors Input the trained random forest model, calculate the average output of all decision trees, and obtain the number of missed cargo tags: In the formula, For the first The predicted values ​​of each decision tree. For the first in a random forest Each decision tree provides a feature vector for the input MAC layer. After processing, the output is a predicted value for the total number of goods tags in the current storage area. This represents the actual number of tags read.

9. The intelligent warehousing RFID robot reading, sensing, and trajectory optimization method according to claim 1, characterized in that, Based on different levels of cargo misread risk areas and the number and location distribution of misread cargo tags, the UHF RFID robot's trajectory within the optimal reading area of ​​the cargo tags to be read is adjusted. Specifically, the optimal reading area of ​​each cargo tag to be read is taken as a node to be visited. The access order of each area is encoded, and a genetic algorithm is used to optimize the path combination. The genetic algorithm includes initializing the chromosome population, calculating fitness based on the total path length and area priority, performing parent selection, crossover, and mutation operations, and obtaining the optimal access order after multiple generations of iteration. Based on the optimal access order, the local trajectory and the inter-area connection path are integrated to generate the UHF RFID robot's trajectory within the optimal reading area of ​​the cargo tags to be read, starting from the center position of the optimal reading area of ​​the cargo tags to be read.

10. An intelligent warehousing RFID robot reading, sensing, and trajectory optimization system, based on the intelligent warehousing RFID robot reading, sensing, and trajectory optimization method according to any one of claims 1 to 9, characterized in that, It includes a data processing module, a location awareness module, a cargo omission awareness module, and a trajectory optimization module; The data processing module is used to enable the UHF RFID robot to move along a preset trajectory and speed, and to obtain shelf tag data, cargo tag data, and MAC layer data of the cargo tags. The position sensing module is used to construct a position sensing model of the interaction between the UHF RFID robot and the shelf based on the shelf label data and the trajectory and speed of the UHF RFID robot, obtain the position coordinates of the shelf positioning label and the UHF RFID robot, calculate the time when the UHF RFID robot runs to the center of the optimal reading area of ​​the label to be read based on the shelf positioning label and the UHF RFID robot, and control the UHF RFID robot to stop moving when the time is reached. The cargo omission detection module is used to construct a cargo label density distribution detection model based on shelf label data and cargo label data, obtain the density distribution of cargo labels, and divide the shelf area to be read into different levels of cargo omission risk areas based on the density distribution of cargo labels. Based on different levels of cargo misread risk areas and MAC layer data of cargo tags, the number and location distribution of misread cargo tags are calculated. The trajectory optimization module is used to adjust the UHF RFID robot's trajectory within the optimal reading area for the cargo tags to be read, based on different levels of cargo miss reading risk areas and the number and location distribution of miss reading cargo tags.

Images