Long-period digital map updating method and system applied to unmanned forklift

Abstract

The invention provides a long-period digital map updating method and system applied to an unmanned forklift, and relates to the field of digital maps, and the method comprises the steps: loading a fixed map of a warehouse, and obtaining a real-time position of the unmanned forklift until the unmanned forklift moves to an initial position corresponding to a carriage loading task; determining the size parameters of the carriage through a mobile base station on the unmanned forklift and a plurality of positioning tags on the carriage; loading an initial temporary map corresponding to the carriage based on the size parameters of the carriage and the historical temporary map; based on the size parameters of the carriage and a historical temporary map, scanning auxiliary information corresponding to the carriage is determined, internal point cloud data of the carriage is collected through a point cloud collection device on the unmanned forklift, a temporary map corresponding to the carriage is constructed, and a loading task is executed; after the truck loading task is completed and the truck is returned to the initial position, the temporary map corresponding to the compartment is deleted, the fixed map of the warehouse is reloaded, and the method has the advantage that the positioning capacity and the operation efficiency of the unmanned forklift in the complex and changeable environment are improved.

Description

Technical Field

[0001] This invention relates to the field of digital maps, and in particular to a method and system for long-cycle digital map updates applied to unmanned forklifts. Background Technology

[0002] In the field of smart logistics, unmanned forklifts, as key equipment for automated operations, currently rely mainly on pre-drawn maps to achieve positioning and navigation in fixed scenarios to complete tasks such as material handling. However, with the rapid development of smart warehousing and logistics, higher demands are being placed on the operational capabilities of unmanned forklifts. They need to accurately locate themselves in dynamically changing and complex scenarios to adapt to the diverse and flexible operational needs of end-to-end warehousing and logistics. Among these, enabling unmanned forklifts to transport goods into the cargo compartment of trucks for automated loading is an extremely important and challenging operational scenario. Due to the significant differences in cargo compartment dimensions and internal structures among different trucks, the operating environment is constantly changing.

[0003] Existing positioning technologies can only perform positioning operations on pre-loaded, fixed maps. Once the work scenario changes, such as entering the interior of a truck with a different size and structure, the positioning algorithm cannot accurately determine the location of the unmanned forklift due to the lack of updated map information, making it difficult to guide it to complete tasks such as automated loading. This limitation severely restricts the application scope and operational efficiency of unmanned forklifts in changing scenarios, and cannot meet the high requirements for flexibility and adaptability of automated equipment in the rapid development of intelligent warehousing and logistics.

[0004] Therefore, there is a need to provide a long-term digital map update method and system for unmanned forklifts to improve their positioning capabilities and operational efficiency in complex and ever-changing environments. Summary of the Invention

[0005] This invention provides a long-cycle digital map update method for unmanned forklifts, comprising: receiving a loading task from warehouse outbound to truck bed; loading a fixed map of the warehouse and obtaining the real-time position of the unmanned forklift until the unmanned forklift moves to the initial position corresponding to the loading task in the truck bed; determining the size parameters of the truck bed using a mobile base station on the unmanned forklift and multiple positioning tags on the truck bed; loading an initial temporary map corresponding to the truck bed based on the size parameters of the truck bed and historical temporary maps; determining scanning auxiliary information corresponding to the truck bed based on the size parameters of the truck bed and historical temporary maps; collecting internal point cloud data of the truck bed using a point cloud acquisition device on the unmanned forklift based on the scanning auxiliary information corresponding to the truck bed; constructing a temporary map corresponding to the truck bed based on the initial temporary map and internal point cloud data, and executing the loading task; after the unmanned forklift completes the loading task and returns to the initial position, deleting the temporary map corresponding to the truck bed and reloading the fixed map of the warehouse.

[0006] Furthermore, the multiple positioning tags on the carriage include at least a first positioning tag, a second positioning tag, a third positioning tag, and a fourth positioning tag respectively set at the four corners of the carriage door, and also include a fifth positioning tag set along the depth direction of the carriage, wherein the straight-line distance between the fifth positioning tag and one of the first, second, third, and fourth positioning tags is the depth of the carriage.

[0007] Furthermore, the size parameters of the truck body are determined by using the mobile base station on the unmanned forklift and the positioning tags on the truck body, including: determining multiple key location points; for each key location point, acquiring the characteristics of the interaction signals between the mobile base station on the unmanned forklift and the multiple positioning tags on the truck body; and determining the size parameters of the truck body based on the characteristics of the interaction signals between the mobile base station on the unmanned forklift and the multiple positioning tags on the truck body corresponding to each key location point.

[0008] Furthermore, multiple key location points are identified, including: randomly sampling multiple location points; for each location point, acquiring the characteristics of the interaction signals between the mobile base station on the unmanned forklift and multiple positioning tags on multiple test compartments, wherein the size parameters of any two test compartments differ; for each location point, calculating the response difference value of the location point based on the characteristics of the interaction signals between the mobile base station on the unmanned forklift and multiple positioning tags on multiple test compartments; and using a genetic algorithm, determining multiple key location points based on the response difference value of each location point.

[0009] Furthermore, the characteristics of the interaction signals between the mobile base station on the unmanned forklift and the multiple positioning tags on the truck body include at least the strength of the interaction signals between the mobile base station on the unmanned forklift and the first, second, third, fourth, and fifth positioning tags respectively. Based on the characteristics of the interaction signals between the mobile base station on the unmanned forklift and the multiple positioning tags on the truck body corresponding to each key location point, the size parameters of the truck body are determined, including: constructing a size determination model based on the characteristics of the interaction signals between the mobile base station on the unmanned forklift and the multiple positioning tags on the multiple test truck bodies corresponding to each key location point; and determining the size parameters of the truck body based on the characteristics of the interaction signals between the mobile base station on the unmanned forklift and the multiple positioning tags on the truck body corresponding to each key location point using the size determination model.

[0010] Furthermore, based on the size parameters of the carriage and the historical temporary map, the initial temporary map corresponding to the carriage is loaded, including: filtering multiple matching historical temporary maps based on the size parameters of the carriage and the size parameters corresponding to the historical temporary map; and loading the initial temporary map corresponding to the carriage based on the time decay coefficient of the multiple matching historical temporary maps.

[0011] Furthermore, based on the time decay coefficients of multiple matched historical temporary maps, the initial temporary map corresponding to the carriage is loaded, including: for each matched historical temporary map, calculating the similarity between the matched historical temporary map and other matched historical temporary maps; based on the similarity between the matched historical temporary map and other matched historical temporary maps and the time decay coefficients of other matched historical temporary maps, calculating the comprehensive priority value of the matched historical temporary map; and based on the comprehensive priority value of each matched historical temporary map, loading the initial temporary map corresponding to the carriage.

[0012] Furthermore, based on the size parameters of the carriage and historical temporary maps, the scanning assistance information corresponding to the carriage is determined, including: dividing the carriage into multiple sub-regions based on multiple matched historical temporary maps, and determining the scanning weight of each sub-region; and determining the scanning assistance information corresponding to the carriage based on the scanning weight of each sub-region, wherein the scanning assistance information corresponding to the carriage includes the scanning parameters of each sub-region.

[0013] Furthermore, based on the initial temporary map and internal point cloud data corresponding to the carriage, a temporary map corresponding to the carriage is constructed, including: for each sub-region, extracting the first point cloud data of the sub-region from the initial temporary map corresponding to the carriage, extracting the second point cloud data of the sub-region from the internal point cloud data, calculating the matching degree between the first point cloud data and the second point cloud data of the sub-region, updating the initial temporary map corresponding to the carriage based on the matching degree between the first point cloud data and the second point cloud data of the sub-region, and constructing the temporary map corresponding to the carriage.

[0014] This invention provides a long-cycle digital map update system for unmanned forklifts, comprising: a task receiving module for receiving loading tasks from warehouse to truck; a map loading module for loading a fixed map of the warehouse and obtaining the real-time position of the unmanned forklift until it moves to the initial position corresponding to the loading task in the truck; a size prediction module for determining the size parameters of the truck using a mobile base station on the unmanned forklift and multiple positioning tags on the truck; the map loading module is also used to load an initial temporary map corresponding to the truck based on the truck's size parameters and historical temporary maps; a point cloud acquisition module for determining the scanning auxiliary information corresponding to the truck based on the truck's size parameters and historical temporary maps, and collecting internal point cloud data of the truck based on the scanning auxiliary information of the truck using a point cloud acquisition device on the unmanned forklift; the map loading module is also used to construct a temporary map corresponding to the truck based on the initial temporary map and internal point cloud data, and execute the loading task; the map loading module is also used to delete the temporary map corresponding to the truck and reload the fixed map of the warehouse after the unmanned forklift completes the loading task and returns to the initial position.

[0015] Compared with existing technologies, the long-cycle digital map update method and system for unmanned forklifts provided by this invention have at least the following beneficial effects: 1. By interacting with carefully placed positioning tags on the truck bed via a mobile base station on the unmanned forklift, the truck bed's dimensions can be accurately determined. Positioning tags are distributed at the corners of the doors and along the depth of the truck bed, ensuring data accuracy through multi-dimensional positioning. Simultaneously, based on the dimensions and historical temporary maps, a matching map is selected and scan auxiliary information is determined. This scan auxiliary information clarifies the scanning parameters for each sub-region, making subsequent point cloud data collection more targeted. This precise determination of truck bed information provides a solid foundation for the unmanned forklift's operations within the truck bed, avoiding operational errors caused by inaccurate information, improving operational reliability and stability, and ensuring that the unmanned forklift can move and operate accurately within the truck bed, laying a good foundation for efficiently completing loading tasks.

[0016] 2. A temporary map is constructed based on an initial temporary map and real-time collected internal point cloud data. The map is updated by calculating the matching degree of point cloud data in sub-regions, which can reflect changes inside the truck compartment in a timely manner. Historical temporary maps are filtered and loaded using a time decay coefficient to ensure that the most valuable map is selected. A comprehensive priority calculation considers similarity and timeliness. This intelligent update method enables the digital map to dynamically adapt to changes in cargo stacking and structure inside the truck compartment, allowing the unmanned forklift to always operate based on the latest and most accurate map. This improves operational flexibility and adaptability, effectively copes with complex and ever-changing loading environments, reduces operational interruptions or errors caused by environmental changes, and improves overall operational efficiency.

[0017] 3. After the unmanned forklift completes its loading task and returns to its initial position, the temporary map of the truck bed is deleted and the fixed warehouse map is reloaded. This map resource management method is both reasonable and efficient. During the loading task, the temporary map provides necessary support for the operation; it is deleted promptly after the task is completed to avoid consuming excessive storage resources and ensure smooth system operation. Reloading the fixed warehouse map allows the unmanned forklift to quickly switch to warehouse operation mode and continue performing other tasks. This method of managing map resources rationally according to task stages optimizes system performance, improves resource utilization, ensures efficient operation of the unmanned forklift in different operating scenarios, and enhances the overall operational efficiency of the logistics system. Attached Figure Description

[0018] This specification will be further described by way of exemplary embodiments, which will be described in detail with reference to the accompanying drawings. These embodiments are not limiting; in these embodiments, the same reference numerals denote the same structures, wherein: Figure 1 This is a flowchart illustrating a long-cycle digital map update method for unmanned forklifts, as shown in some embodiments of this specification. Figure 2 This is a schematic diagram of multiple positioning tags according to some embodiments of this specification; Figure 3 This is a schematic diagram of a long-cycle digital map update system for unmanned forklifts, as shown in some embodiments of this specification.

[0019] In the diagram, 110 is the first location tag; 120 is the second location tag; 130 is the third location tag; 140 is the fourth location tag; and 150 is the fifth location tag. Detailed Implementation

[0020] To more clearly illustrate the technical solutions of the embodiments in this specification, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are merely some examples or embodiments of this specification. For those skilled in the art, these drawings can be applied to other similar scenarios without creative effort. Unless obvious from the context or otherwise specified, the same reference numerals in the drawings represent the same structures or operations.

[0021] Figure 1 This is a flowchart illustrating a long-cycle digital map update method for unmanned forklifts, as shown in some embodiments of this specification. Figure 1 As shown, the long-cycle digital map update method applied to unmanned forklifts may include the following steps.

[0022] S1. Receive the loading task from the warehouse to the car.

[0023] S2. Load the fixed map of the warehouse and obtain the real-time position of the unmanned forklift until the unmanned forklift moves to the initial position corresponding to the loading task in the truck.

[0024] Specifically, in intelligent warehousing and logistics scenarios, to achieve efficient task scheduling of unmanned forklifts, a fixed map of the warehouse must first be loaded as the basic positioning framework. This map is typically generated through simultaneous localization and mapping (SLAM) using LiDAR or manual surveying, containing the 3D coordinates and topological relationships of key landmarks such as shelves, aisles, and loading / unloading areas, and stored as a high-precision raster map or point cloud model. After loading, the pose data of the unmanned forklift is acquired in real time through wireless tag positioning technology (e.g., UWB tag arrays or Bluetooth beacon networks): receivers installed on the bottom of the forklift or on the forks continuously scan the surrounding tag signals, and the precise position of the forklift in the map coordinate system is calculated (accuracy up to ±5cm) using TOA (Time of Arrival) or TDOA (Time Difference of Arrival) algorithms. At the same time, positioning drift caused by short-term occlusion or dynamic bumps is corrected through IMU data fusion.

[0025] The initial position for loading the truck is set to a designated point within the truck parking area. During the forklift's movement, its current position is continuously compared with the initial position coordinates. The optimal path is dynamically planned using the A* algorithm, while the obstacle avoidance module adjusts the trajectory in real time to avoid dynamic obstacles (such as other forklifts or personnel). When the forklift enters a 0.5-meter radius around the initial position, it is determined that the unmanned forklift has moved to the initial position corresponding to the truck loading task.

[0026] S3. Determine the dimensions of the truck bed using the mobile base station on the unmanned forklift and multiple positioning tags on the truck bed.

[0027] The multiple positioning tags on the carriage include at least a first positioning tag 110, a second positioning tag 120, a third positioning tag 130, and a fourth positioning tag 140 respectively set at the four corners of the carriage door, and a fifth positioning tag 150 set along the depth direction of the carriage. The straight-line distance between the fifth positioning tag 150 and one of the first positioning tag 110, the second positioning tag 120, the third positioning tag 130, and the fourth positioning tag 140 is the depth of the carriage.

[0028] Specifically, first positioning tags 110 to fourth positioning tags 140 are installed at the four corners of the carriage door. These tags serve as spatial reference points to determine the planar position and dimensions of the carriage door: the first positioning tag 110 and the second positioning tag 120 are located on the top sides of the door, and the third positioning tag 130 and the fourth positioning tag are located on the bottom sides. The rectangle formed by these four tags can accurately describe the geometric parameters of the carriage (such as width and height). Simultaneously, a fifth positioning tag 150 is set along the depth direction of the carriage (i.e., the direction extending from the door into the carriage interior). Its position must satisfy the condition that the straight-line distance from any corner tag (such as the first positioning tag 110) is equal to the depth value of the carriage. This design forms a three-dimensional positioning coordinate system: the four corner positioning tags define the planar coordinate system of the carriage door, while the fifth positioning tag 150, through its depth distance from the corner tags, extends the positioning dimension to three-dimensional space, simultaneously sensing the planar position and depth direction of the carriage.

[0029] The first, second, third, fourth, and fifth positioning tags can be RFID (Radio Frequency Identification) tags, UWB (Ultra-Wideband) tags, etc.

[0030] In some embodiments, the size parameters of the vehicle compartment are determined using a mobile base station on the unmanned forklift and a positioning tag on the compartment, including: Identify multiple key location points; For each key location point, the characteristics of the interaction signals between the mobile base station on the unmanned forklift and multiple positioning tags on the truck body are obtained. The characteristics of the interaction signals between the mobile base station on the unmanned forklift and multiple positioning tags on the truck body include at least the strength of the interaction signals between the mobile base station on the unmanned forklift and the first positioning tag, the second positioning tag, the third positioning tag, the fourth positioning tag and the fifth positioning tag respectively. Based on the characteristics of the interaction signals between the mobile base station on the unmanned forklift and multiple positioning tags on the truck body corresponding to each key location point, the size parameters of the truck body are determined.

[0031] In some embodiments, determining multiple key location points includes: Randomly sample multiple locations, for example, in the area in front of the carriage door, randomly sample multiple locations; For each location point, the characteristics of the interaction signals between the mobile base station on the unmanned forklift and multiple positioning tags on multiple test compartments are acquired. Among them, the dimensional parameters of any two test compartments are different, that is, at least one of the width, height and depth is different. For example, compartment A: width 2.5m, height 2.2m, depth 4m; compartment B: width 2.8m, height 2.5m, depth 4.5m; compartment C: width 2.3m, height 2.0m, depth 3.8m. For each location point, the response difference value of the location point is calculated based on the characteristics of the interaction signals between the mobile base station on the unmanned forklift and multiple positioning tags on multiple test compartments; Using a genetic algorithm, multiple key location points are determined based on the response difference value of each location point.

[0032] Specifically, the response difference value at the location point can be calculated according to the following process: S11. For each positioning tag, namely the first positioning tag, the second positioning tag, the third positioning tag, the fourth positioning tag and the fifth positioning tag, the variance of the intensity of the interaction signal of the positioning tag at the corresponding position point on the multiple test carriages can be calculated as the response difference value of the corresponding position point of the positioning tag. S12. Calculate the average of the response difference values ​​for each location tag corresponding to the location point, and use it as the response difference value of the location point.

[0033] The following process can be used to determine multiple key location points based on the response difference value of each location point using a genetic algorithm: S21. For each location point, construct the intensity vector corresponding to the location point. The intensity vector is composed of the intensity of the interaction signals of the first location tag, the second location tag, the third location tag, the fourth location tag, and the fifth location tag. For any two location points, calculate the cosine similarity of the intensity vectors corresponding to the two location points. S22. For each location point, calculate the sampling probability of the location point based on the response difference value of the location point. The larger the response difference value of the location point, the greater the sampling probability of the location point. S23. Based on the sampling probability of each location point, sample multiple location points to generate multiple individuals, where each individual is a set of multiple location points, for example, at least 5 location points. This is just an example. When generating individuals, generate a random number [0,1] for each location point independently. If the random number is less than the sampling probability of that location point, include it in the current individual; otherwise, skip it. S24. Construct a fitness function, where the fitness function is related to the mean response difference value, the mean cosine similarity value, and the number of locations included in the individual. The mean response difference value is the mean of the response difference values ​​of the multiple locations included in the individual, and the mean cosine similarity value is the mean of the cosine similarity between any two locations included in the individual. The larger the mean response difference value, the smaller the mean cosine similarity value, and the smaller the number, the larger the fitness function will be. S25. Calculate the fitness function value for each individual; S26. Based on the fitness function value of each individual, perform selection, crossover, and mutation operations to update the population until the maximum number of iterations is reached, the fitness function value converges, or the maximum fitness function value is greater than the fitness function value threshold.

[0034] Randomly sampling locations and introducing multiple test trucks with varying dimensional parameters comprehensively covers different truck conditions, making the acquired interaction signal characteristics more representative. This lays the foundation for accurately calculating response difference values, thereby ensuring the rationality of key location point determination. A genetic algorithm is employed to process the location points. This involves constructing intensity vectors and calculating cosine similarity, determining sampling probabilities to generate individuals based on response difference values, and constructing a fitness function related to the mean response difference value, mean cosine similarity, and quantity. The population is continuously optimized during the iterative process. This approach automatically selects locations with large response difference values, low cosine similarity, and an appropriate number as key location points, effectively avoiding the blindness and subjectivity of manual selection and improving the accuracy and efficiency of key location point determination. Ultimately, based on these key location points, the truck size parameters can be accurately determined, providing a reliable basis for unmanned forklift operations.

[0035] In some embodiments, the size parameters of the vehicle compartment are determined based on the characteristics of the interaction signals between the mobile base station on the unmanned forklift corresponding to each key location point and multiple positioning tags on the vehicle compartment, including: Based on the characteristics of the interaction signals between the mobile base station on the unmanned forklift corresponding to each key location point and multiple positioning tags on multiple test carriages, a size determination model is constructed. The size determination model determines the size parameters of the truck body based on the characteristics of the interaction signals between the mobile base station on the unmanned forklift and multiple positioning tags on the truck body at each key location point.

[0036] Specifically, the size determination model can be a convolutional neural network model. The input to the size determination model can include the intensity of the interaction signals between the mobile base station on the unmanned forklift corresponding to each key location point and the first, second, third, fourth, and fifth positioning tags. The output of the size determination model can include the dimensions of the vehicle body. In the size determination model, the convolutional layer is responsible for automatically extracting local features from the input data. Through the sliding operation of the convolutional kernel on the input data, it captures the potential correlations and patterns between different location points and between the interaction signal intensities of different positioning tags. The pooling layer downsamples the feature map output by the convolutional layer, reducing the data dimensionality while retaining the main features, enhancing the robustness and generalization ability of the model, and preventing overfitting. After multiple convolutional and pooling operations, the data is passed to the fully connected layer, which integrates and transforms the previously extracted features to prepare for the final output. The output layer is set to output the dimensions of the vehicle body, such as the width, height, and depth of the vehicle body.

[0037] Training, validation, and test sets can be constructed based on the characteristics of the interaction signals between the mobile base station on the unmanned forklift corresponding to each key location point and multiple positioning tags on multiple test carriages. The training set is used for parameter learning and optimization of the size determination model. By continuously adjusting parameters such as weights and biases in the model, the error between the prediction results of the size determination model on the training set and the true labels gradually decreases. The validation set is used to monitor the performance of the size determination model during training, preventing the model from overfitting on the training set and losing its generalization ability to unknown data. By observing the performance of the size determination model on the validation set, hyperparameters such as learning rate, convolution kernel size, and pooling method can be adjusted in a timely manner to obtain better model performance. The test set is used to independently evaluate the final performance of the size determination model after training is completed, ensuring that the size determination model can accurately and reliably determine the carriage size parameters in practical applications. Through training and optimization with a large number of training samples, the size determination model can learn the complex mapping relationship between interactive signal features and vehicle size parameters. Thus, when faced with new vehicles and key location points, it can quickly and accurately output the size parameters of the vehicle, providing strong support for the precise operation of unmanned forklifts.

[0038] S4. Based on the size parameters of the carriage and the historical temporary map, load the initial temporary map corresponding to the carriage.

[0039] Specifically, it includes: Based on the size parameters of the carriage and the size parameters corresponding to the historical temporary map, multiple matching historical temporary maps are filtered; Based on the time decay coefficients of multiple matched historical temporary maps, the initial temporary map corresponding to the carriage is loaded.

[0040] Specifically, the cosine similarity between the size parameters of the carriage and the size parameters of the carriage corresponding to the historical temporary map can be calculated. Historical temporary maps with a cosine similarity greater than the cosine similarity threshold (e.g., 0.7) are used as the matching historical temporary maps.

[0041] In some embodiments, based on the time decay coefficients of multiple matched historical temporary maps, an initial temporary map corresponding to the carriage is loaded, including: For each matched historical temporary map, calculate the similarity between the matched historical temporary map and other matched historical temporary maps. Based on the similarity between the matched historical temporary map and other matched historical temporary maps and the time decay coefficient of other matched historical temporary maps, calculate the comprehensive priority value of the matched historical temporary map. Based on the overall priority value of each matched historical temporary map, load the initial temporary map corresponding to the carriage.

[0042] Specifically, similarity can be measured from multiple dimensions, such as the spatial layout of the map, the distribution of key landmarks, and route planning information. By using algorithms such as cosine similarity and Euclidean distance, the features of different historical temporary maps in these dimensions can be quantitatively compared to obtain the similarity between the matched historical temporary map and other matched historical temporary maps.

[0043] After obtaining the similarity of each matched historical temporary map to other maps, a comprehensive priority value is calculated by combining the time decay coefficients of other matched historical temporary maps. The time decay coefficient reflects the decrease in the reference value of a historical temporary map over time; the older the time, the smaller the coefficient, meaning that the corresponding historical temporary map is less applicable to the current situation. The comprehensive priority value is calculated by considering both similarity and the time decay coefficient. Specifically, for a matched historical temporary map, its similarity to other historical temporary maps is multiplied by the corresponding time decay coefficients of those other historical temporary maps. These products are then summed, and the sum is normalized to obtain the comprehensive priority value of that matched historical temporary map. This comprehensive priority value comprehensively reflects the timeliness of the historical temporary map in the time dimension and its similarity and relevance to other maps in terms of content, providing a more comprehensive and accurate reflection of its reference value in the current scenario.

[0044] The historical temporary map with the highest overall priority value can be loaded as the initial temporary map for the corresponding carriage.

[0045] During the screening and matching stage of historical temporary maps, the cosine similarity between the carriage size parameters and the corresponding size parameters of the historical temporary maps is calculated. Using a cosine similarity threshold as the standard for screening, historical temporary maps that highly match the current carriage size characteristics can be accurately identified, avoiding interference from irrelevant maps and providing an accurate foundation for subsequent operations, thus improving the targeted nature of map loading. When loading the initial temporary map based on the time decay coefficient, the characteristics of different historical temporary maps are quantitatively compared using cosine similarity algorithms and Euclidean distance algorithms from multiple dimensions, including spatial area layout, key landmark location distribution, and route planning information. The similarity between matching historical temporary maps is calculated, and then a comprehensive priority value is calculated in conjunction with the time decay coefficient. The time decay coefficient considers the decrease in reference value of historical temporary maps over time, giving newer, more timely maps an advantage in the comprehensive evaluation; while the similarity calculation ensures that the selected map is closely related to the current scene in terms of content. The comprehensive priority value comprehensively considers timeliness and content relevance, and can more accurately reflect the reference value of historical temporary maps in the current scene. Selecting the historical temporary map with the highest overall priority value as the initial temporary map loads fully utilizes the most valuable historical data to quickly generate an initial temporary map that fits the current actual conditions of the truck bed. This provides a reliable basis for the operation of unmanned forklifts and other equipment inside the truck bed, helps improve the accuracy and efficiency of operations, reduces errors and mistakes caused by inaccurate maps, lowers operating costs, and enhances the stability and reliability of the overall operation process.

[0046] S5. Based on the dimensions of the carriage and historical temporary maps, determine the corresponding scanning auxiliary information for the carriage.

[0047] Specifically, it includes: Based on multiple matched historical temporary maps, the carriage is divided into multiple sub-regions, and the scanning weight of each sub-region is determined. Based on the scanning weight of each sub-region, the scanning auxiliary information corresponding to the carriage is determined, wherein the scanning auxiliary information corresponding to the carriage includes the scanning parameters of each sub-region.

[0048] Specifically, the scanning weight of a sub-region is related to the differences in the point cloud data of that sub-region across multiple matching historical temporary maps. Point cloud data accurately reflects the three-dimensional spatial information of an object's surface. Differences in the point cloud data of the same sub-region across different matching historical temporary maps indicate that the sub-region has changed at different times or under different operational scenarios, such as changes in cargo stacking or adjustments to the internal structure of the vehicle. The greater the difference, the more frequent and important the changes in that sub-region, thus assigning it a greater scanning weight to highlight its importance in subsequent scanning processes.

[0049] Based on the predetermined scanning weights of each sub-region, the corresponding scanning assistance information for the carriage is determined. This information includes scanning parameters for each sub-region. These parameters cover key indicators such as scanning resolution and scanning frequency, and are set appropriately according to the scanning weights of the sub-regions. For sub-regions with high scanning weights, the scanning resolution and frequency are appropriately increased, and the scanning angle range is expanded to ensure comprehensive and detailed capture of the sub-region's information. Conversely, for sub-regions with low scanning weights, the scanning parameter requirements can be appropriately reduced to improve overall scanning efficiency while meeting basic scanning needs. The scanning assistance information determined in this way is more closely aligned with the actual conditions of the carriage, providing strong support for subsequent accurate and efficient scanning operations.

[0050] S6. Based on the scanning auxiliary information corresponding to the carriage, the internal point cloud data of the carriage is collected by the point cloud acquisition device on the unmanned forklift.

[0051] S7. Based on the initial temporary map and internal point cloud data corresponding to the carriage, construct the temporary map corresponding to the carriage and execute the loading task.

[0052] Specifically, it includes: For each sub-region, the first point cloud data of the sub-region is extracted from the initial temporary map corresponding to the carriage, and the second point cloud data of the sub-region is extracted from the internal point cloud data. The matching degree between the first point cloud data and the second point cloud data of the sub-region is calculated. Based on the matching degree between the first point cloud data and the second point cloud data of the sub-region, the initial temporary map corresponding to the carriage is updated, and a temporary map corresponding to the carriage is constructed.

[0053] Specifically, the matching degree calculation can be considered from several key aspects, such as the similarity of the positional distribution of points in the point cloud data and the degree of fit of the object contours. A quantitative analysis is used to obtain a numerical value that accurately reflects the degree of similarity between the two. Based on the calculated matching degree, the initial temporary map corresponding to the carriage is updated accordingly. If the matching degree is high, it indicates that the information of that sub-region in the initial temporary map is relatively consistent with the actual situation, and the relevant data can be appropriately retained and fine-tuned. If the matching degree is low, it indicates that there is a significant difference between the initial temporary map and the actual situation, and it is necessary to use the second point cloud data in the internal point cloud data as the primary source to significantly correct and supplement the information of that sub-region in the initial temporary map.

[0054] By performing the above operations on each sub-region, the initial temporary map of the entire carriage is gradually updated, ultimately constructing a temporary map that accurately reflects the current actual internal condition of the carriage. This temporary map provides a reliable basis for subsequent loading tasks, enabling unmanned forklifts and other equipment to accurately plan their driving paths and identify the location of goods based on precise map information, thus efficiently and accurately completing loading operations. This effectively improves loading efficiency and accuracy, and reduces errors and risks caused by inaccurate maps.

[0055] S8. After the unmanned forklift completes the loading task and returns to its initial position, it deletes the temporary map corresponding to the truck bed and reloads the fixed map of the warehouse.

[0056] Figure 3 This is a schematic diagram of a long-cycle digital map update system for unmanned forklifts, as shown in some embodiments of this specification. Figure 3 As shown, a long-cycle digital map update system for unmanned forklifts can include a task receiving module, a map loading module, a size prediction module, and a point cloud acquisition module.

[0057] The task receiving module is used to receive loading tasks from the warehouse to the car body; The map loading module is used to load a fixed map of the warehouse and obtain the real-time location of the unmanned forklift until the unmanned forklift moves to the initial position corresponding to the loading task in the truck. The size prediction module is used to determine the size parameters of the truck body by using the mobile base station on the unmanned forklift and multiple positioning tags on the truck body; The map loading module is also used to load the initial temporary map corresponding to the carriage based on the carriage's size parameters and historical temporary maps. The point cloud acquisition module is used to determine the scanning auxiliary information corresponding to the car body based on the size parameters of the car body and the historical temporary map, and to collect the internal point cloud data of the car body through the point cloud acquisition device on the unmanned forklift based on the scanning auxiliary information corresponding to the car body. The map loading module is also used to construct a temporary map corresponding to the carriage based on the initial temporary map and internal point cloud data corresponding to the carriage, and to perform loading tasks. The map loading module is also used to delete the temporary map corresponding to the truck body and reload the fixed map of the warehouse after the unmanned forklift completes the loading task and returns to the initial position.

[0058] The long-cycle digital map update system for unmanned forklifts can be used to execute long-cycle digital map update methods for unmanned forklifts, which will not be elaborated here.

[0059] Finally, it should be understood that the embodiments described in this specification are merely illustrative of the principles of the embodiments described herein. Other variations may also fall within the scope of this specification. Therefore, alternative configurations of the embodiments described herein are intended to be illustrative rather than limiting, and should be considered consistent with the teachings of this specification. Accordingly, the embodiments described herein are not limited to those explicitly introduced and described herein.

Claims

1. A long-cycle digital map update method applied to unmanned forklifts, characterized in that, include: Receive loading tasks from the warehouse to the wagon; Load the fixed map of the warehouse and obtain the real-time position of the unmanned forklift until the unmanned forklift moves to the initial position corresponding to the loading task in the truck bed; The dimensions of the truck bed are determined by using a mobile base station on the unmanned forklift and multiple positioning tags on the truck bed. Based on the size parameters of the carriage and the historical temporary map, load the initial temporary map corresponding to the carriage; Based on the dimensions of the carriage and historical temporary maps, determine the corresponding scanning auxiliary information for the carriage; Based on the scanning auxiliary information corresponding to the carriage, the internal point cloud data of the carriage is collected by the point cloud acquisition device on the unmanned forklift. Based on the initial temporary map and internal point cloud data corresponding to the carriage, a temporary map corresponding to the carriage is constructed, and the loading task is executed. After the unmanned forklift completes the loading task and returns to its initial position, it deletes the temporary map corresponding to the truck bed and reloads the fixed map of the warehouse.

2. The long-cycle digital map update method for unmanned forklifts according to claim 1, characterized in that, The multiple positioning tags on the carriage include at least a first positioning tag, a second positioning tag, a third positioning tag, and a fourth positioning tag respectively set at the four corners of the carriage door, and a fifth positioning tag set along the depth direction of the carriage. The straight-line distance between the fifth positioning tag and one of the first, second, third, and fourth positioning tags is the depth of the carriage.

3. The long-cycle digital map update method for unmanned forklifts according to claim 2, characterized in that, The dimensions of the truck bed are determined using mobile base stations on the unmanned forklift and positioning tags on the truck bed, including: Identify multiple key location points; For each key location point, acquire the characteristics of the interaction signals between the mobile base station on the unmanned forklift and multiple positioning tags on the truck body; Based on the characteristics of the interaction signals between the mobile base station on the unmanned forklift and multiple positioning tags on the truck body corresponding to each key location point, the size parameters of the truck body are determined.

4. The long-cycle digital map update method for unmanned forklifts according to claim 3, characterized in that, Several key location points were identified, including: Randomly sample multiple locations; For each location point, the characteristics of the interaction signals between the mobile base station on the unmanned forklift and multiple positioning tags on multiple test compartments are obtained, wherein the size parameters of any two test compartments are different; For each location point, the response difference value of the location point is calculated based on the characteristics of the interaction signals between the mobile base station on the unmanned forklift and multiple positioning tags on multiple test compartments; Using a genetic algorithm, multiple key location points are determined based on the response difference value of each location point.

5. The long-cycle digital map update method for unmanned forklifts according to claim 4, characterized in that, The characteristics of the interaction signals between the mobile base station on the unmanned forklift and multiple positioning tags on the truck body include at least the strength of the interaction signals between the mobile base station on the unmanned forklift and the first, second, third, fourth, and fifth positioning tags, respectively: Based on the characteristics of the interaction signals between the mobile base station on the unmanned forklift and multiple positioning tags on the truck body corresponding to each key location point, the size parameters of the truck body are determined, including: Based on the characteristics of the interaction signals between the mobile base station on the unmanned forklift corresponding to each key location point and multiple positioning tags on multiple test carriages, a size determination model is constructed. The size determination model determines the size parameters of the truck body based on the characteristics of the interaction signals between the mobile base station on the unmanned forklift and multiple positioning tags on the truck body at each key location point.

6. The long-cycle digital map update method for unmanned forklifts according to any one of claims 1-5, characterized in that, Based on the carriage's size parameters and historical temporary maps, load the initial temporary map corresponding to the carriage, including: Based on the size parameters of the carriage and the size parameters corresponding to the historical temporary map, multiple matching historical temporary maps are filtered; Based on the time decay coefficients of multiple matched historical temporary maps, the initial temporary map corresponding to the carriage is loaded.

7. The long-cycle digital map update method for unmanned forklifts according to claim 6, characterized in that, Based on the time decay coefficients of multiple matched historical temporary maps, the initial temporary map corresponding to the carriage is loaded, including: For each matched historical temporary map, calculate the similarity between the matched historical temporary map and other matched historical temporary maps. Based on the similarity between the matched historical temporary map and other matched historical temporary maps and the time decay coefficient of other matched historical temporary maps, calculate the comprehensive priority value of the matched historical temporary map. Based on the overall priority value of each matched historical temporary map, load the initial temporary map corresponding to the carriage.

8. The long-cycle digital map update method for unmanned forklifts according to claim 6, characterized in that, Based on the dimensions of the carriage and historical temporary maps, the corresponding scanning auxiliary information for the carriage is determined, including: Based on multiple matched historical temporary maps, the carriage is divided into multiple sub-regions, and the scanning weight of each sub-region is determined. Based on the scanning weight of each sub-region, the scanning auxiliary information corresponding to the carriage is determined, wherein the scanning auxiliary information corresponding to the carriage includes the scanning parameters of each sub-region.

9. The long-cycle digital map update method for unmanned forklifts according to claim 8, characterized in that, Based on the initial temporary map and internal point cloud data corresponding to the carriage, a temporary map corresponding to the carriage is constructed, including: For each sub-region, the first point cloud data of the sub-region is extracted from the initial temporary map corresponding to the carriage, and the second point cloud data of the sub-region is extracted from the internal point cloud data. The matching degree between the first point cloud data and the second point cloud data of the sub-region is calculated. Based on the matching degree between the first point cloud data and the second point cloud data of the sub-region, the initial temporary map corresponding to the carriage is updated, and a temporary map corresponding to the carriage is constructed.

10. A long-cycle digital map update system for unmanned forklifts, characterized in that, The long-cycle digital map update method for unmanned forklifts as described in claim 1 includes: The task receiving module is used to receive loading tasks from the warehouse to the car body; The map loading module is used to load a fixed map of the warehouse and obtain the real-time location of the unmanned forklift until the unmanned forklift moves to the initial position corresponding to the loading task in the truck. The size prediction module is used to determine the size parameters of the truck body by using the mobile base station on the unmanned forklift and multiple positioning tags on the truck body; The map loading module is also used to load the initial temporary map corresponding to the carriage based on the carriage's size parameters and historical temporary maps. The point cloud acquisition module is used to determine the scanning auxiliary information corresponding to the car body based on the size parameters of the car body and the historical temporary map, and to collect the internal point cloud data of the car body through the point cloud acquisition device on the unmanned forklift based on the scanning auxiliary information corresponding to the car body. The map loading module is also used to construct a temporary map corresponding to the carriage based on the initial temporary map and internal point cloud data corresponding to the carriage, and to perform loading tasks. The map loading module is also used to delete the temporary map corresponding to the truck body and reload the fixed map of the warehouse after the unmanned forklift completes the loading task and returns to the initial position.

Images