A method of material handling robot map availability assessment

Abstract

This invention provides a method for assessing the map availability of a material handling robot. Applied to the field of intelligent robot technology, the method includes acquiring multi-source operational data, dividing it into mapping and operational phase indicator data, and generating a standardized data stream by aligning it with a unified time benchmark; identifying the current map version and determining the mapping assessment window and operational monitoring window; extracting a first set of associated features from the mapping data in the mapping assessment window, calculating a mapping quality evaluation coefficient to determine the initial mapping state; identifying positioning anomalies in the operational monitoring window, extracting a second set of associated features to calculate an operational stability evaluation coefficient; calculating a map availability fusion score based on weighted ratios; if the score does not reach a preset threshold, the map is determined to be in a risky state, and an abnormal fragment index set is generated and output to guide map updates. This improves the accuracy of map availability assessment for material handling robots.

Description

Technical Field

[0001] This invention relates to the field of intelligent robot technology, and in particular to a method for evaluating the map availability of a material handling robot. Background Technology

[0002] With the rapid development of intelligent manufacturing and warehousing logistics automation industries, material handling robots have become core intelligent equipment for achieving efficient material flow in industrial production and intelligent warehousing scenarios. High-precision and highly available environmental maps are the core foundation for material handling robots to achieve autonomous positioning, path planning and continuous and stable operation. The accurate assessment of map availability directly determines the robot's operating efficiency, operational safety and full life cycle maintenance costs.

[0003] Currently, existing technologies for map availability assessment of material handling robots suffer from several technical shortcomings: most assessment schemes only collect data such as point cloud matching accuracy and loop closure detection performance during the mapping phase to conduct static mapping quality evaluation, or only perform post-event statistical analysis on positioning deviation data during the operation phase, failing to achieve a linkage analysis between mapping quality and operational stability; at the same time, existing schemes do not perform unified time benchmark alignment processing for multi-source heterogeneous operational data from the robot mapping and operation phases, resulting in severe data fragmentation and making it impossible to accurately trace the causal relationship between defects in the mapping phase and abnormal positioning events during the operation phase; furthermore, existing assessment methods are mostly qualitative judgments, lacking a quantitative evaluation system for the entire lifecycle of map versions, and cannot accurately locate the spatiotemporal position of abnormal events, making it difficult to provide accurate guidance for map iteration and updates, resulting in low accuracy in the map availability assessment of material handling robots. Summary of the Invention

[0004] To address the problems existing in the prior art, this invention provides a method for evaluating the map availability of a material handling robot. The method includes: Acquire multi-source operational data of the material handling robot, divide the multi-source operational data into mapping stage indicator data and operational stage indicator data, and align the mapping stage indicator data and the operational stage indicator data based on a unified time benchmark to generate a standardized data stream. The current map version is identified based on the standardized data stream, and a mapping evaluation window and an operation monitoring window are determined for the current map version. Within the mapping evaluation window, a first associated feature set is extracted based on the mapping stage indicator data in the standardized data stream, and a mapping quality evaluation coefficient is determined based on the first associated feature set to determine the initial mapping status of the current map version. When a location anomaly event is identified within the operation monitoring window, a second associated feature set is extracted based on the attributes of the location anomaly event, and an operation stability evaluation coefficient is determined based on the second associated feature set. The map availability fusion score is calculated based on the weighting ratio of the mapping quality evaluation coefficient and the operational stability evaluation coefficient, in order to determine whether the map availability fusion score has reached the preset availability threshold. If the map availability fusion score does not reach the availability threshold, the current map version is determined to be in a risky state. An abnormal fragment index set is generated based on the timing and spatial location of the location anomaly event, and the abnormal fragment index set is output as a guidance suggestion for map updates.

[0005] Furthermore, the generation of the standardized data stream includes: Determine the first timestamp source of the mapping phase indicator data and the second timestamp source of the operation phase indicator data; The synchronization deviation between the first timestamp source and the second timestamp source is calculated based on a preset time synchronization model. Determine whether the synchronization deviation value exceeds the preset allowable deviation range; If the synchronization deviation value exceeds the allowable deviation range, then time interpolation compensation is performed on the mapping stage indicator data based on the second timestamp source to eliminate the impact of time asynchrony; The mapping phase indicator data after eliminating the impact of time asynchrony are formatted to be the same as the operation phase indicator data, and the dataset after filtering out abnormal values ​​is used as the standardized data stream.

[0006] Furthermore, the determination of the mapping evaluation window and the operation monitoring window for the current map version includes: Extract the map version identifier from the standardized data stream, and set the map corresponding to the map version identifier as the current map version; Set the complete construction time period of the current map version as the map evaluation window; Obtain the launch time of the current map version, and determine the operation monitoring window based on the launch time and the preset number of task executions or runtime. When a change in the map version identifier in the standardized data stream is detected, the map corresponding to the newly identified map version identifier is automatically used as the updated current map version, and the corresponding mapping evaluation window and operation monitoring window are redefined.

[0007] Further, determining the initial mapping state of the current map version includes: Based on the graph construction stage index data, the loop closure error statistics, graph optimization residual statistics, and the number of peaks in the matching residual are extracted, and the loop closure error statistics, graph optimization residual statistics, and the number of peaks in the matching residual are aggregated into the first association feature set. Each feature item in the first associated feature set is normalized and assigned a corresponding first weight value. The normalized feature terms are multiplied by their corresponding first weight values ​​to obtain the mapping quality evaluation coefficients. Determine whether the mapping quality evaluation coefficient is greater than a preset mapping quality threshold; If the mapping quality evaluation coefficient is greater than the mapping quality threshold, then the initial mapping status of the current map version is determined to be qualified. If the mapping quality evaluation coefficient is less than or equal to the mapping quality threshold, then the initial mapping state of the current map version is determined to be in a state to be optimized.

[0008] Further, the extraction of the second associated feature set based on the attributes of the located anomaly event includes: Real-time acquisition of pose change, positional reliability, and repositioning frequency from the operational phase indicator data; Determine whether the change in pose exceeds the jump threshold within a preset first time interval. If so, determine that the attribute of the positioning abnormal event is a positioning jump event. Determine whether the location confidence level is lower than the lower confidence threshold within a preset second time interval. If so, determine that the location anomaly event is a confidence decline event. Determine whether the relocation frequency exceeds the frequency threshold within a unit time. If so, determine that the attribute of the abnormal relocation event is an abnormal relocation event. Extract the duration, severity, and frequency of all location anomalies, and combine the duration, severity, and frequency of each event into the second associated feature set.

[0009] Further, determining the operational stability evaluation coefficient based on the second associated feature set includes: For each type of location anomaly event in the second set of associated features, a penalty factor is calculated based on the corresponding duration, severity, and frequency of occurrence. Obtain the pre-defined penalty weights corresponding to each location-specific abnormal event; The sum of the products of the penalty factors and corresponding penalty weights for all located abnormal events is calculated to obtain the comprehensive penalty value; The difference between the preset baseline stability full score and the comprehensive penalty value is used as the operational stability evaluation coefficient.

[0010] Further, the calculation of the map usability fusion score based on the weighted ratio of the mapping quality evaluation coefficient and the operational stability evaluation coefficient includes: Obtain the initially set basic weights for map building and basic weights for operation; Determine whether the runtime of the current map version exceeds a preset maturity threshold; If the runtime exceeds the maturity threshold, the basic mapping weight is reduced and the basic runtime weight is increased to obtain the updated mapping weight and runtime weight. If the runtime does not exceed the maturity threshold, then the basic mapping weight and the basic runtime weight will be used as the mapping weight and the runtime weight, respectively. Calculate the product of the mapping quality evaluation coefficient and the mapping weight, and the product of the operational stability evaluation coefficient and the operational weight. Add the two products together to obtain the map availability fusion score.

[0011] Further, determining whether the map usability fusion score reaches a preset usability threshold includes: The availability threshold is divided into a first warning threshold and a second warning threshold, wherein the first warning threshold is greater than the second warning threshold; When the map availability fusion score is greater than or equal to the first warning threshold, the current map version is determined to be in an available state; When the map availability fusion score is less than the first warning threshold and greater than or equal to the second warning threshold, the current map version is determined to be in the risk state. When the map availability fusion score is less than the second warning threshold, the current map version is determined to be unavailable.

[0012] Furthermore, the step of generating an abnormal fragment index set based on the timing and spatial location of the location anomaly event includes: Extract the start and end times of the location anomaly event to generate a time segment index; Obtain the area identifier or path segment identifier where the material handling robot is located when the positioning anomaly event occurs, and generate a spatial segment index; Map the spatial coordinates of the location anomaly event to the underlying raster region of the current map version to generate a map fragment index; The time segment index, spatial segment index, and map segment index are associated and bound together to form the abnormal segment index set.

[0013] Furthermore, the step of outputting the set of abnormal fragment indices as guiding suggestions for map updates includes: Obtain the cluster density of localized abnormal events under the same spatial segment index in the abnormal segment index set; determine whether the cluster density exceeds a preset density threshold; if the cluster density exceeds the density threshold, determine the local feature unstable region based on the map segment index, and generate a local rescan suggestion for the local feature unstable region; If the map availability fusion score indicates that the current map version is unavailable, then a global reconstruction suggestion is generated based on the abnormal fragment index set; The local patching suggestions or the global reconstruction suggestions are combined with the abnormal fragment index set to generate a structured evaluation report for output.

[0014] This invention acquires multi-source operational data from material handling robots, divides the data into mapping and operational phases, and aligns them based on a unified time benchmark to generate a standardized data stream. This achieves temporal consistency, dimensional completeness, and standardization of the evaluation data, eliminating evaluation errors caused by temporal misalignment and missing dimensions in multi-source data, and improving the accuracy of the basic data for map usability assessment. By determining a dedicated mapping evaluation window and operational monitoring window for the current map version, and extracting corresponding related feature sets in stages, and quantifying and calculating the mapping quality evaluation coefficient and operational stability evaluation coefficient respectively, this invention achieves a two-dimensional refined evaluation of the map from initial mapping quality to full-cycle operational stability, avoiding the result bias caused by single-sided evaluation and improving the comprehensiveness and accuracy of map usability assessment indicators. By calculating the map usability fusion score through the weighted ratio of mapping quality and operational stability evaluation coefficients, and generating anomaly fragment index sets for maps that do not reach the usability threshold by combining the timing and spatial location of location anomaly events, this invention achieves a comprehensive quantitative judgment of map usability and source tracing of anomalies, improving the credibility of map usability assessment results and the accuracy of risk identification.

[0015] It should be understood that the description in the Summary of the Invention is not intended to limit the key or essential features of the embodiments of the present invention, nor is it intended to restrict the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0016] The above and other features, advantages, and aspects of the various embodiments of the present invention will become more apparent from the accompanying drawings and the following detailed description. The drawings are provided for a better understanding of the invention and are not intended to limit the scope of the invention. In the drawings, the same or similar reference numerals denote the same or similar elements, wherein: Figure 1A flowchart illustrating a method for assessing the map availability of a material handling robot according to an embodiment of the present invention is shown. Figure 2 A detailed flowchart illustrating the determination of the initial mapping status of the current map version in the material handling robot map availability assessment method according to an embodiment of the present invention is shown. Figure 3 A detailed flowchart illustrating the second associated feature set extracted based on the attributes of the location anomaly event in the material handling robot map availability assessment method according to an embodiment of the present invention is shown. Figure 4 A detailed flowchart illustrating the method for evaluating the map availability of a material handling robot according to an embodiment of the present invention is shown, illustrating the calculation of a map availability fusion score based on the weighted ratio of the mapping quality evaluation coefficient and the operational stability evaluation coefficient. Figure 5 A detailed flowchart illustrating the process of determining whether the map availability fusion score of a material handling robot reaches a preset availability threshold is shown in the material handling robot map availability assessment method according to an embodiment of the present invention. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Furthermore, the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0019] Figure 1 A flowchart of a material handling robot map availability assessment method according to an embodiment of the present invention is shown, the method comprising: S101, acquire multi-source operation data of the material handling robot, divide the multi-source operation data into mapping stage indicator data and operation stage indicator data, and align the mapping stage indicator data and the operation stage indicator data based on a unified time benchmark to generate a standardized data stream. S102, Identify the current map version based on the standardized data stream, and determine the mapping evaluation window and the operation monitoring window for the current map version; S103, within the mapping evaluation window, a first associated feature set is extracted based on the mapping stage indicator data in the standardized data stream, and a mapping quality evaluation coefficient is determined based on the first associated feature set to determine the initial mapping status of the current map version; S104, when a location anomaly event is identified in the operation monitoring window, a second associated feature set is extracted based on the attributes of the location anomaly event, and an operation stability evaluation coefficient is determined based on the second associated feature set; S105, calculate the map availability fusion score based on the weight ratio of the mapping quality evaluation coefficient and the operation stability evaluation coefficient, so as to determine whether the map availability fusion score reaches the preset availability threshold. S106, if the map availability fusion score does not reach the availability threshold, the current map version is determined to be in a risky state. An abnormal fragment index set is generated based on the timing and spatial location of the location anomaly event, and the abnormal fragment index set is output as a guidance suggestion for map updates.

[0020] In some embodiments, generating a standardized data stream includes: determining a first timestamp source for the mapping phase indicator data and a second timestamp source for the operation phase indicator data; wherein the first timestamp source is the hardware clock timestamp of each mapping sensor synchronously collected during the mapping phase, and the second timestamp source is the system global clock timestamp of the main controller of the material handling robot during the operation phase; calculating the synchronization deviation value between the first timestamp source and the second timestamp source based on a preset time synchronization model; wherein the preset time synchronization model is a clock deviation estimation model based on linear fitting, which obtains the clock frequency deviation and a fixed offset by collecting the corresponding time series of two timestamp sources under the same triggering event, and then calculates the synchronization deviation value; determining whether the synchronization deviation value exceeds a preset allowable deviation range; if the synchronization deviation value exceeds the allowable deviation range, then Using the second timestamp source as a reference, time interpolation compensation is performed on the mapping stage indicator data to eliminate the impact of time asynchrony. The time interpolation compensation employs a sliding window-based spline interpolation method, using the time reference of the second timestamp source as the interpolation node to resample the sampling sequence of the mapping stage indicator data, ensuring that the timestamps of the resampled data are completely aligned with the time reference of the operational stage indicator data. The mapping stage indicator data and the operational stage indicator data, after eliminating the impact of time asynchrony, are formatted uniformly, and the dataset after filtering out outliers is used as the standardized data stream. The format unification involves converting both types of data into a unified key-value pair structured format, where the key name includes a data type identifier, a timestamp identifier, and a value identifier. The outlier filtering is based on the Laida criterion to remove outliers exceeding three standard deviations from the data sequence, while also removing invalid data with missing data frames or overflowing values. According to embodiments of the present invention, by determining the dual timestamp sources and calculating the synchronization deviation, the time reference alignment of multi-source data is achieved, eliminating the evaluation error caused by time asynchrony; by time interpolation compensation, format unification and abnormal value filtering, the standardized processing of multi-source data is achieved, improving the consistency and reliability of subsequent evaluation data.

[0021] For example, for a material handling robot with a rated speed of 1.2 m / s, a LiDAR sampling frequency of 10 Hz, and a main controller system clock frequency of 100 Hz, the first timestamp source is the hardware clock timestamp of the mapping LiDAR and IMU sensors, and the second timestamp source is the global clock timestamp of the AGV main controller's Linux system. 100 sets of dual timestamp sequences are collected under the same event triggered by the material handling robot's startup synchronization pulse. A clock frequency deviation of 12 ppm and a fixed offset of 25 ms are obtained through linear fitting, resulting in a calculated synchronization deviation of 32 ms. The preset allowable deviation range is ±20 ms. Since 32 ms exceeds this range, the global clock of the main controller is used as the reference. A cubic spline interpolation method with a window size of 10 sampling points is employed, using the main controller's 100 ms interval time nodes as interpolation nodes, to resample the mapping sensor data. After resampling, the time reference deviation between the data timestamp and the data during the operation phase is ≤2 ms. Both types of data are then uniformly converted to {"data_type": "xxx", "timestamp": 1714521600000, The key-value pair format "value": xxx} is used for IMU acceleration data sequences. The standard deviation is calculated to be 0.2 m / s². Outliers exceeding ±0.6 m / s² are removed based on the Raida criterion. Invalid data with a data frame missing rate >10% and numerical overflow are also removed, and finally a standardized data stream is generated.

[0022] In some embodiments, determining the mapping evaluation window and the operation monitoring window for the current map version includes: extracting a map version identifier from the standardized data stream, and setting the map corresponding to the map version identifier as the current map version; wherein, the map version identifier is a unique version code built into the map file, generated synchronously with the completion of mapping and embedded in the corresponding map file and the pre-running data stream, and includes mapping timestamp, mapping device number and version iteration sequence information; setting the complete construction time period of the current map version as the mapping evaluation window; wherein, the complete construction time period is a continuous time interval from the start timestamp of the start of mapping for this version of the map to the end timestamp of the completion of mapping and generation of the final map file; obtaining the online time point of the current map version, and determining the operation monitoring window based on the online time point and a preset number of task executions or runtime; wherein, the online time point is the timestamp when this version of the map is first loaded by the material handling robot for formal operation; The operation monitoring window starts at the online time and ends at the time when the preset number of consecutive task executions is completed or the preset continuous running time ends. When the preset termination condition is not met, the operation monitoring window extends dynamically as the robot operates in real time. When the material handling robot encounters abnormal situations such as positioning deviation, path conflict, or environmental changes during formal operation, the system triggers a new round of map building process, generates a new map version, and updates the map version identifier, causing a change in the map version identifier in the data stream. When a change in the map version identifier in the standardized data stream is detected, the map corresponding to the newly identified map version identifier is automatically used as the updated current map version, and the corresponding mapping evaluation window and operation monitoring window are redefined. The monitoring of map version identifier changes involves real-time parsing of the version identifier field in the standardized data stream. When a jump in the encoding of the version identifier field is detected and a new encoding is continuously and stably output, it is determined that the map version identifier has changed. According to embodiments of the present invention, the evaluation object is locked by extracting and identifying the map version identifier, thus avoiding confusion in the evaluation of multiple map versions; the evaluation dimensions are divided by setting the map building and running windows in stages, thereby improving the pertinence of the map lifecycle evaluation; and the evaluation process is dynamically adapted by automatically responding to version changes, ensuring the real-time performance and accuracy of the evaluation.

[0023] For example, the map version identifier MAP-20260320-AGV087-V2.1 is extracted from the standardized data stream. This code contains the mapping timestamp 2026-03-20 09:30:00, the mapping equipment number AGV087, and the version iteration number V2.1. This version of the map is set as the current map version. A continuous 75-minute interval from the start timestamp of mapping for this version of the map (2026-03-20 08:15:00) to the end timestamp of the final map file being generated (2026-03-20 09:30:00) is set as the mapping evaluation window. The first time this version of the map was loaded for formal operations is obtained as 2026-03-20. At 10:00:00, the preset termination condition for the operation monitoring window is 72 consecutive hours of operation or the completion time of 200 consecutive material handling tasks. If the termination condition is not met, the operation monitoring window will dynamically extend backward as the material handling robot operates in real time. The version identifier field in the standardized data stream is parsed in real time. When the version code is detected to jump to MAP-20260325-AGV087-V2.2 and is stably output for 50 consecutive data frames, it is determined that the map version identifier has changed. The new code is automatically used as the updated current map version, and the corresponding mapping evaluation window and operation monitoring window are redefined.

[0024] In some embodiments, Figure 2This document illustrates a detailed flowchart of a method for assessing the availability of a material handling robot map according to an embodiment of the present invention, specifically for determining the initial mapping state of the current map version. The method includes: extracting loop closure error statistics, graph optimization residual statistics, and the number of peaks in the matching residual based on the mapping stage index data; aggregating the loop closure error statistics, graph optimization residual statistics, and the number of peaks in the matching residual into a first associated feature set; wherein the loop closure error statistics are the root mean square value of the pose deviation between loop closure frames after all loop closure detections are successfully matched throughout the mapping process; the graph optimization residual statistics are the average residual of all pose nodes and edge constraints after backend graph optimization during the mapping process; and the number of peaks in the matching residual is the statistical value of the number of times the inter-frame matching residual exceeds a preset residual upper limit during the mapping process; normalizing each feature item in the first associated feature set and assigning a corresponding first weight value; wherein the normalization process uses min-m... The ax normalization method maps the values ​​of each feature item to the [0,1] interval; the first weight value is predetermined based on the influence of each feature item on the mapping quality using the analytic hierarchy process, and the sum of all weight values ​​is 1; the normalized feature items are multiplied by their corresponding first weight values ​​to obtain the mapping quality evaluation coefficient; the value range of the mapping quality evaluation coefficient is [0,1], and a higher value indicates better initial mapping quality; it is determined whether the mapping quality evaluation coefficient is greater than a preset mapping quality threshold; the mapping quality threshold is a critical value determined based on the statistical distribution of qualified mapping samples, used to distinguish whether the mapping quality meets the basic usability requirements; if the mapping quality evaluation coefficient is greater than the mapping quality threshold, the initial mapping state of the current map version is determined to be qualified; if the mapping quality evaluation coefficient is less than or equal to the mapping quality threshold, the initial mapping state of the current map version is determined to be in a state to be optimized. According to embodiments of the present invention, by extracting and aggregating multi-dimensional mapping features, a full-dimensional representation of mapping quality is achieved, avoiding the evaluation bias of a single indicator; by feature normalization and weighted summation calculation, a quantitative evaluation of mapping quality is achieved, improving the accuracy of mapping status determination; and by threshold comparison for hierarchical determination, a clear division of mapping status is achieved, providing a reliable basis for subsequent evaluation.

[0025] For example, based on the index data of the mapping stage, the root mean square value of the pose deviation of 1.8 cm from 126 successful loop closures in the entire mapping process was extracted as the loop closure error statistics. The average residual of all pose nodes and edge constraints after the back-end graph optimization solution was 0.9 cm as the graph optimization residual statistics. The number of times the inter-frame matching residual exceeded the preset residual upper limit of 5 cm was 12 times as the number of matching residual peaks. The above three data items were aggregated into the first associated feature set. The min-max normalization method was used to map each feature item to the [0,1] interval, where the loop closure error statistics were... The normalized result of the calculation is (5-1.8) / (5-0.5)=0.711, the normalized result of the graph optimization residual statistics is (3-0.9) / (3-0.3)=0.778, and the normalized result of the number of matching residual peaks is (50-12) / 50=0.76. The first weight values ​​are predetermined by the analytic hierarchy process: loop closure error statistics 0.5, graph optimization residual statistics 0.35, and number of matching residual peaks 0.15. The sum of all weight values ​​is 1. The mapping quality evaluation coefficient is calculated as 0.711×0.5 + 0.778×0.35 + 0.76×0.15=0.7418. The preset mapping quality threshold is 0.7. Since 0.7418>0.7, the initial mapping status of the current map version is determined to be qualified.

[0026] In some embodiments, Figure 3This document illustrates a detailed flowchart of a method for evaluating the map availability of a material handling robot according to an embodiment of the present invention, specifically for extracting a second associated feature set based on the attributes of a positioning anomaly event. The flowchart includes: real-time acquisition of pose change, positional confidence, and repositioning frequency from the operational phase index data; wherein the pose change is a weighted sum of the Euclidean distance and angular deviation between the pose estimates of the material handling robot at adjacent sampling times; the positional confidence is the probability confidence value of the pose estimation result output by the positioning algorithm; the repositioning frequency is the number of times the robot triggers a repositioning process and performs a pose reset per unit time; determining whether the pose change exceeds a jump threshold within a preset first time interval, and if so, classifying the attribute of the positioning anomaly event as a positioning jump event; wherein the preset first time interval is the single-frame sampling period of the robot positioning algorithm, and the jump threshold is an upper limit value of the pose change determined based on the robot's rated motion speed and maximum acceleration; and determining whether the positional confidence is lower than a confidence lower limit threshold within a preset second time interval, and if so... The location anomaly event is then determined to be a confidence-decreased event. The preset second time interval is N consecutive location sampling periods, where N is a positive integer, and the confidence threshold is the lowest confidence threshold value for the location algorithm to stably output a valid pose. It is then determined whether the relocation frequency exceeds a frequency threshold within a unit time. If so, the location anomaly event is determined to be an abnormal relocation event. The unit time is a preset sliding statistical time window, and the frequency threshold is the upper statistical limit of the relocation frequency under normal robot operation. The duration, severity, and frequency of all location anomalies are extracted, and each duration, severity, and frequency is used to form the second associated feature set. The duration is the time span from the triggering time to the time of recovery for a single location anomaly event. The severity is determined by classifying the pose deviation amplitude and the impact level on the task based on the location anomaly event. The frequency is the number of times a location anomaly event with the same attribute is triggered within the operation monitoring window. According to embodiments of the present invention, real-time acquisition of multi-dimensional operational indicators enables full-dimensional monitoring of positioning status, covering various positioning anomaly scenarios; classification and judgment of anomaly events through multi-threshold comparison enables identification of anomaly attributes and refines the evaluation dimensions of positioning anomalies; extraction and aggregation of multi-dimensional features of anomaly events enables quantitative characterization of operational anomalies, providing reliable data support for subsequent stability evaluation.

[0027] For example, real-time acquisition of operational phase indicator data: pose change is the weighted sum of the Euclidean distance and angle deviation of pose at adjacent sampling times, with weights of 0.6 and 0.4 respectively; positioning confidence is the probability value of the 0-1 interval output by the AMCL positioning algorithm; repositioning frequency is the number of repositioning triggers within a 1-minute sliding window; the single-frame sampling period of the positioning algorithm is 100ms, and the jump threshold is determined to be 0.2m based on the rated speed of the material handling robot (1.2m / s) and the maximum acceleration (0.8m / s²). When a pose change of 0.28m or more is detected in adjacent sampling periods, it is judged as a positioning jump event; a preset judgment interval of 5 consecutive positioning sampling periods is used, and the lower confidence threshold is 0.6. When a pose change of 0.28m or more is detected in adjacent sampling periods, it is judged as a positioning jump event. When the confidence level of the location is below 0.6, it is judged as a confidence decline event. A 1-minute sliding statistical time window is preset, and the frequency threshold is 3 times / minute. When the relocation frequency is detected to be 4 times within 1 minute, which is more than 3 times / minute, it is judged as an abnormal relocation event. The corresponding data of all abnormal location events in the operation monitoring window are extracted: 8 location jump events occurred, with an average duration of 0.8s per event and a severity level of 2 (out of 5); 12 confidence decline events occurred, with an average duration of 2.5s per event and a severity level of 3; 3 abnormal relocation events occurred, with an average duration of 4s per event and a severity level of 4. The duration, severity, and frequency of the above three types of abnormal location events together constitute the second association feature set.

[0028] In some embodiments, determining the operational stability evaluation coefficient based on the second associated feature set includes: for each type of positioning anomaly event in the second associated feature set, calculating a penalty factor for the corresponding event based on its duration, severity, and frequency; wherein the penalty factor is a weighted product of the duration, severity level coefficient, and frequency of occurrence of a single type of positioning anomaly event, used to quantify the negative impact of a single type of anomaly event on positioning stability; obtaining preset penalty weights corresponding to each positioning anomaly event; wherein the penalty weights are predetermined based on the priority of the impact of various types of positioning anomalies on the operational safety and task execution of the material handling robot. The weight values ​​are positively correlated with the risk level of the event. The sum of the products of the penalty factors and corresponding penalty weights for all positioning anomalies is calculated to obtain the comprehensive penalty value. This comprehensive penalty value is the sum of the weighted penalty values ​​of all types of positioning anomalies, used to quantify the overall negative impact of all positioning anomalies on operational stability. The comprehensive penalty value is subtracted from the preset baseline stability full score, and the difference is used as the operational stability evaluation coefficient. The baseline stability full score is the stability benchmark value when there are no positioning anomalies, and the lower limit of the operational stability evaluation coefficient is 0; a higher value indicates better positioning stability during robot operation. According to this embodiment, by calculating the penalty factors of multi-dimensional parameters for a single type of anomaly, differentiated quantification of the impact of different anomalies is achieved; by matching penalty weights and calculating the comprehensive penalty value, global quantification of the impact of operational anomalies is achieved; and by calculating the stability coefficient through full score deduction, an intuitive quantitative evaluation of operational stability is achieved, improving the readability and rationality of the evaluation results.

[0029] For example, for the three types of abnormal location events in the second set of associated features, corresponding penalty factors are calculated. The severity levels 1-5 correspond to a grade coefficient of 0.1-0.5, and the penalty factor = cumulative duration of a single event × severity grade coefficient × frequency. The cumulative duration of a location jump event is 6.4s, with a grade coefficient of 0.2 and a frequency of 8, so the penalty factor = 6.4 × 0.2 × 8 = 10.24. The cumulative duration of a confidence decrease event is 30s, with a grade coefficient of 0.3 and a frequency of 12, so the penalty factor = 30 × 0.3 × 12 = 108. The cumulative duration of an abnormal relocation event is 12s, with a grade coefficient of 0.4 and a frequency of 3, so the penalty factor = 12 × 0.4 × 3 = 14.4. Based on the priority of event impact, penalty weights are preset: 0.2 for location jump events, 0.3 for confidence decrease events, and 0.5 for abnormal relocation events. These weights are positively correlated with the risk level. The calculated comprehensive penalty value = 10.24 × 0.2 + 108×0.3 + 14.4×0.5=41.648; The preset basic stability full score is 100, and the theoretical maximum comprehensive penalty value within the operation monitoring window is 100. Therefore, the operation stability evaluation coefficient = 100-41.648=58.352, which is 0.5835 after normalization to the [0,1] interval.

[0030] In some embodiments, Figure 4This document illustrates a detailed flowchart of a material handling robot map availability assessment method according to an embodiment of the present invention, illustrating the calculation of a map availability fusion score based on a weighted ratio of the mapping quality evaluation coefficient and the operational stability evaluation coefficient. The process includes: obtaining initially set mapping base weights and operational base weights; wherein the initial sum of the initial values ​​of the mapping base weights and operational base weights is 1, and the initial weight ratio is pre-set based on the degree of dominant influence of mapping quality on availability in the early stages of map deployment; determining whether the runtime of the current map version exceeds a preset maturity threshold; wherein the runtime is the cumulative operation runtime of the current map version from the deployment time to the current statistical time, and the maturity threshold is a pre-determined time threshold based on the map feature decay cycle and environmental change patterns of the robot's operational scenario. If the runtime exceeds the maturity threshold, the basic mapping weight is reduced and the basic operation weight is increased to obtain updated mapping weight and operation weight. If the runtime does not exceed the maturity threshold, the basic mapping weight and the basic operation weight are used as mapping weight and operation weight, respectively. When the maturity threshold is not exceeded, the initial ratio of mapping weight and operation weight remains unchanged to prioritize the impact of initial mapping quality on map usability. The product of the mapping quality evaluation coefficient and the mapping weight, and the product of the operation stability evaluation coefficient and the operation weight are calculated, and the two products are added together to obtain the map usability fusion score. The numerical range of the map usability fusion score is [0,1], and the higher the value, the better the overall usability of the current map version. According to embodiments of the present invention, by comparing map runtime with maturity threshold, dynamic adaptation of weight allocation is achieved, which conforms to the evaluation characteristics of the entire map lifecycle; by dynamically and differentially allocating weights, a reasonable balance between mapping quality and operational stability is achieved, thereby improving the rationality of fusion scoring; and by calculating the weighted sum of two coefficients, a comprehensive quantification of map availability is achieved, providing a core basis for availability determination.

[0031] For example, the initial mapping base weight is set to 0.6 and the operational base weight to 0.4, with the sum of the initial values ​​being 1. This initial ratio is pre-set based on the dominant influence of mapping quality in the early stages of map launch. The current map version launch time is 10:00 on March 20, 2026, the current statistical time is 10:00 on March 23, 2026, and the cumulative operation time is 72 hours. Based on the decay cycle of map features in the warehousing scenario and the pre-determined maturity threshold of 360 hours, the 72 hours have not exceeded the 360-hour maturity threshold. Therefore, the initial ratio of mapping weight 0.6 and operational weight 0.4 remains unchanged. The calculated map usability fusion score is 0.7418 × 0.6 + 0.5835 × 0.4 = 0.44508 + 0.2334 = 0.6785.

[0032] In some embodiments, Figure 5 This document illustrates a detailed flowchart of a material handling robot map availability assessment method according to an embodiment of the present invention, illustrating the process of determining whether the map availability fusion score has reached a preset availability threshold. The method includes: dividing the availability threshold into a first warning threshold and a second warning threshold, wherein the first warning threshold is greater than the second warning threshold; wherein the first warning threshold is the minimum score threshold at which the map can be stably used for all scenarios, and the second warning threshold is the minimum score threshold at which the map can meet basic operational requirements; both thresholds are preset based on the statistical distribution of availability scores and fault correspondences across a large number of robot operation scenarios; when the map availability fusion score is greater than or equal to the first warning threshold, the current map version is determined. The map is in an available state; wherein, the available state is a state in which the map can support stable operation of the robot in all scenarios, and no map update or optimization is required; when the map availability fusion score is less than the first warning threshold and greater than or equal to the second warning threshold, the current map version is determined to be in the risk state; wherein, the risk state is a state in which the map can still support basic robot operations, but there is a risk of positioning anomalies, and local optimization is required; when the map availability fusion score is less than the second warning threshold, the current map version is determined to be in an unavailable state; wherein, the unavailable state is a state in which the map can no longer support stable robot operations, there is a high safety risk, and it is necessary to immediately stop using it and rebuild the map. According to the embodiments of the present invention, by dividing the map availability into two levels of warning thresholds, a multi-gradient determination of map availability is realized, and the evaluation level of availability is refined; by comparing the fusion score with the threshold in segments, the map status is classified, providing a clear basis for subsequent differentiated handling; by clearly defining multiple states, the guidance and practicality of the map availability evaluation results are improved.

[0033] For example, based on the statistical distribution of availability scores and fault correspondence of 1200 sets of warehouse material handling robot operation scenarios, the first warning threshold is set to 0.7 and the second warning threshold is set to 0.6. The current map availability fusion score is 0.6785, which is less than the first warning threshold of 0.7 and greater than the second warning threshold of 0.6. Therefore, the current map version is determined to be in a risky state, that is, the map can still support the basic handling operations of material handling robots, but there is a risk of positioning anomalies, and local optimization is required.

[0034] In some embodiments, generating an abnormal segment index set based on the occurrence timing and spatial location of the positioning anomaly includes: extracting the start time and end time of the positioning anomaly to generate a time segment index; wherein, the time segment index includes a unique event ID, start timestamp, end timestamp, and duration field of the anomaly, used to uniquely identify the time dimension information of the anomaly; obtaining the area identifier or path segment identifier where the material handling robot is located when the positioning anomaly occurs, and generating a spatial segment index; wherein, the area identifier and path segment identifier are unique codes for pre-divided functional areas and fixed travel paths in the material handling robot's operation map, and the spatial segment index includes the area / path segment code corresponding to the anomaly, the event occurrence time, and the time segment index. The system generates a global coordinate range field; maps the spatial coordinates of the location anomaly event to the underlying grid area of ​​the current map version, generating a map fragment index; wherein, the underlying grid area is the smallest grid unit of the map, the mapping is to convert the robot pose coordinates at the time of the anomaly event into the row and column numbers of the map grid, and the map fragment index includes the grid row and column number range covered by the anomaly event and grid feature attribute fields; the time fragment index, spatial fragment index, and map fragment index are associated and bound to form the anomaly fragment index set; wherein, the association and binding is to use the unique event ID of the anomaly event as the primary key to associate and store the three types of indexes corresponding to the same event, and the anomaly fragment index set is a structured set of associated indexes of all location anomalies. According to the embodiments of the present invention, by generating multi-dimensional indexes of time, space, and map grids, full-dimensional positioning of location anomalies is achieved, covering the spatiotemporal full-dimensional information of the anomaly occurrence; by associating and binding multi-dimensional indexes, the structured integration of anomaly information is achieved, improving the traceability and readability of anomaly fragments, and providing accurate positioning guidance for map optimization.

[0035] In some embodiments, outputting the abnormal fragment index set as a guidance suggestion for map updates includes: obtaining the cluster density of location anomalies under the same spatial fragment index in the abnormal fragment index set; determining whether the cluster density exceeds a preset density threshold; if the cluster density exceeds the density threshold, determining a local feature unstable region based on the map fragment index, and generating a local rescan suggestion for the local feature unstable region; wherein, the cluster density is the number of location anomalies occurring per unit area or per unit length within a path segment corresponding to the same spatial fragment index; the density threshold is a critical value determined based on the background occurrence density of anomalies in a normal operating scenario; the local rescan suggestion includes the area range for local rescanning, and the rescanning... The system includes path planning and rescanning operation parameter requirements. If the map availability fusion score indicates that the current map version is unavailable, a global reconstruction suggestion is generated based on the abnormal fragment index set. This global reconstruction suggestion includes the global mapping operation scope, mapping path planning, mapping sensor parameter configuration, and mapping quality control requirements. The system combines the local rescanning suggestion or the global reconstruction suggestion with the abnormal fragment index set to generate a structured evaluation report for output. This structured evaluation report includes basic map version information, mapping quality evaluation results, operational stability evaluation results, comprehensive map availability judgment conclusions, abnormal fragment index set, and map update guidance suggestions, outputting synchronously in a parsable structured file format and a visual report format. According to this embodiment, by comparing the threshold of abnormal event aggregation density, local unstable areas are identified, generating targeted local rescanning suggestions and reducing map optimization costs. Global reconstruction suggestions are generated by determining unavailability, enabling tiered handling of map problems. The packaged output of the structured evaluation report enhances the practicality and executability of the map optimization guidance suggestions.

[0036] It should be understood that the various processes described above can be used to rearrange, add, or delete steps. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0037] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A material handling robot map availability assessment method, characterized by, include: Acquire multi-source operational data of the material handling robot, divide the multi-source operational data into mapping stage indicator data and operational stage indicator data, and align the mapping stage indicator data and the operational stage indicator data based on a unified time benchmark to generate a standardized data stream. The current map version is identified based on the standardized data stream, and a mapping evaluation window and an operation monitoring window are determined for the current map version. Within the mapping evaluation window, a first associated feature set is extracted based on the mapping stage indicator data in the standardized data stream, and a mapping quality evaluation coefficient is determined based on the first associated feature set to determine the initial mapping status of the current map version. When a location anomaly event is identified within the operation monitoring window, a second associated feature set is extracted based on the attributes of the location anomaly event, and an operation stability evaluation coefficient is determined based on the second associated feature set. The map availability fusion score is calculated based on the weighting ratio of the mapping quality evaluation coefficient and the operational stability evaluation coefficient, in order to determine whether the map availability fusion score has reached the preset availability threshold. If the map availability fusion score does not reach the availability threshold, the current map version is determined to be in a risky state. An abnormal fragment index set is generated based on the timing and spatial location of the location anomaly event, and the abnormal fragment index set is output as a guidance suggestion for map updates.

2. The material handling robot map availability assessment method according to claim 1, characterized in that, The generation of the standardized data stream includes: Determine the first timestamp source of the mapping phase indicator data and the second timestamp source of the operation phase indicator data; The synchronization deviation between the first timestamp source and the second timestamp source is calculated based on a preset time synchronization model. Determine whether the synchronization deviation value exceeds the preset allowable deviation range; If the synchronization deviation value exceeds the allowable deviation range, then time interpolation compensation is performed on the mapping stage indicator data based on the second timestamp source to eliminate the impact of time asynchrony; The mapping phase indicator data after eliminating the impact of time asynchrony are formatted to be the same as the operation phase indicator data, and the dataset after filtering out abnormal values ​​is used as the standardized data stream.

3. The material handling robot map availability assessment method according to claim 2, characterized in that, The process of determining the mapping evaluation window and the operation monitoring window for the current map version includes: Extract the map version identifier from the standardized data stream, and set the map corresponding to the map version identifier as the current map version; Set the complete construction time period of the current map version as the map evaluation window; Obtain the launch time of the current map version, and determine the operation monitoring window based on the launch time and the preset number of task executions or runtime. When a change in the map version identifier in the standardized data stream is detected, the map corresponding to the newly identified map version identifier is automatically used as the updated current map version, and the corresponding mapping evaluation window and operation monitoring window are redefined.

4. The material handling robot map availability assessment method according to claim 3, characterized in that, The determination of the initial mapping status of the current map version includes: Based on the graph construction stage index data, the loop closure error statistics, graph optimization residual statistics, and the number of peaks in the matching residual are extracted, and the loop closure error statistics, graph optimization residual statistics, and the number of peaks in the matching residual are aggregated into the first association feature set. Each feature item in the first associated feature set is normalized and assigned a corresponding first weight value. The normalized feature terms are multiplied by their corresponding first weight values ​​to obtain the mapping quality evaluation coefficients. Determine whether the mapping quality evaluation coefficient is greater than a preset mapping quality threshold; If the mapping quality evaluation coefficient is greater than the mapping quality threshold, then the initial mapping status of the current map version is determined to be qualified. If the mapping quality evaluation coefficient is less than or equal to the mapping quality threshold, then the initial mapping state of the current map version is determined to be in a state to be optimized.

5. The material handling robot map availability assessment method according to claim 4, characterized in that, The extraction of the second associated feature set based on the attributes of the located anomaly event includes: Real-time acquisition of pose change, positional reliability, and repositioning frequency from the operational phase indicator data; Determine whether the change in pose exceeds the jump threshold within a preset first time interval. If so, determine that the attribute of the positioning abnormal event is a positioning jump event. Determine whether the location confidence level is lower than the lower confidence threshold within a preset second time interval. If so, determine that the location anomaly event is a confidence decline event. Determine whether the relocation frequency exceeds the frequency threshold within a unit time. If so, determine that the attribute of the abnormal relocation event is an abnormal relocation event. Extract the duration, severity, and frequency of all location anomalies, and combine the duration, severity, and frequency of each event into the second associated feature set.

6. The material handling robot map availability assessment method according to claim 5, characterized in that, The step of determining the operational stability evaluation coefficient based on the second associated feature set includes: For each type of location anomaly event in the second set of associated features, a penalty factor is calculated based on the corresponding duration, severity, and frequency of occurrence. Obtain the pre-defined penalty weights corresponding to each location-specific abnormal event; The sum of the products of the penalty factors and corresponding penalty weights for all located abnormal events is calculated to obtain the comprehensive penalty value; The difference between the preset baseline stability full score and the comprehensive penalty value is used as the operational stability evaluation coefficient.

7. The material handling robot map availability assessment method according to claim 6, characterized in that, The calculation of the map usability fusion score based on the weighted ratio of the mapping quality evaluation coefficient and the operational stability evaluation coefficient includes: Obtain the initially set basic weights for map building and basic weights for operation; Determine whether the runtime of the current map version exceeds a preset maturity threshold; If the runtime exceeds the maturity threshold, the basic mapping weight is reduced and the basic runtime weight is increased to obtain the updated mapping weight and runtime weight. If the runtime does not exceed the maturity threshold, then the basic mapping weight and the basic runtime weight will be used as the mapping weight and the runtime weight, respectively. Calculate the product of the mapping quality evaluation coefficient and the mapping weight, and the product of the operational stability evaluation coefficient and the operational weight. Add the two products together to obtain the map availability fusion score.

8. The material handling robot map availability assessment method according to claim 7, characterized in that, The step of determining whether the map usability fusion score reaches a preset usability threshold includes: The availability threshold is divided into a first warning threshold and a second warning threshold, wherein the first warning threshold is greater than the second warning threshold; When the map availability fusion score is greater than or equal to the first warning threshold, the current map version is determined to be in an available state; When the map availability fusion score is less than the first warning threshold and greater than or equal to the second warning threshold, the current map version is determined to be in the risk state. When the map availability fusion score is less than the second warning threshold, the current map version is determined to be unavailable.

9. The material handling robot map availability assessment method according to claim 8, characterized in that, The generation of an abnormal fragment index set based on the timing and spatial location of the location anomaly event includes: Extract the start and end times of the location anomaly event to generate a time segment index; Obtain the area identifier or path segment identifier where the material handling robot is located when the positioning anomaly event occurs, and generate a spatial segment index; Map the spatial coordinates of the location anomaly event to the underlying raster area of ​​the current map version to generate a map fragment index; The time segment index, spatial segment index, and map segment index are associated and bound together to form the abnormal segment index set.

10. The material handling robot map availability assessment method according to claim 9, characterized in that, The step of outputting the set of anomalous fragment indices as guidance suggestions for map updates includes: Obtain the cluster density of localized abnormal events under the same spatial segment index in the abnormal segment index set; determine whether the cluster density exceeds a preset density threshold; if the cluster density exceeds the density threshold, determine the local feature unstable region based on the map segment index, and generate a local rescan suggestion for the local feature unstable region; If the map availability fusion score indicates that the current map version is unavailable, then a global reconstruction suggestion is generated based on the abnormal fragment index set; The local rescanning suggestions or the global reconstruction suggestions are combined with the abnormal fragment index set to generate a structured evaluation report for output.

Images