A collaborative deployment method of mobile storage and charging robots based on specific scenarios

Abstract

The present invention provides a method for collaboratively arranging a mobile storage and charging robot based on a specific scene, which relates to the technical field of mobile storage and charging robots, and includes: identifying characteristic parameters of a target scene by fusing a scene image, a point cloud image and a planning image; determining scene requirements according to scene planning of the target scene; collaboratively planning an arrangement plan of a mobile storage and charging robot according to the characteristic parameters and the scene requirements; guiding the mobile storage and charging robot to reach a designated location according to the arrangement plan; automatically identifying a specific scene, and collaboratively planning an arrangement plan of the robot according to the scene characteristics and arrangement requirements, so as to improve the arrangement efficiency and practicality of the mobile storage and charging robot in a specific scene.

Description

Technical Field

[0001] The present invention relates to the technical field of mobile storage and charging robots, and more particularly, to a collaborative layout method for mobile storage and charging robots based on specific scenarios. Background Art

[0002] With the widespread application of new energy vehicles, mobile storage and charging robots, as a new type of charging equipment, have gradually attracted attention. Mobile storage and charging robots can carry multiple rechargeable battery packs and provide instant charging services where charging is needed. However, in specific scenarios, such as large-scale events and temporary charging requirements, how to efficiently and practically arrange mobile storage and charging robots is an urgent problem to be solved.

[0003] In view of this, the present invention provides a collaborative layout method for mobile storage and charging robots based on specific scenarios, which can automatically identify specific scenarios and, according to the scenario characteristics and layout requirements, collaboratively plan the layout scheme of the robots to improve the layout efficiency and practicality of mobile storage and charging robots in specific scenarios. Summary of the Invention

[0004] The object of the present invention is to provide a collaborative layout method for mobile storage and charging robots based on specific scenarios, including: identifying the characteristic parameters of the target scenario by fusing the scenario image, point cloud image, and planning image; the characteristic parameters include the area, shape, and obstacle position of the scenario; wherein, when performing image fusion, the point cloud features of the point cloud image are extracted through the scenario features of the scenario image; determining the scenario requirements according to the scenario planning of the target scenario; the scenario planning includes the activity duration, peak period, vehicle ratio, and number of participants; the scenario requirements include the number, location, and power requirements of the charging piles; collaboratively planning the layout scheme of the mobile storage and charging robots according to the characteristic parameters and the scenario requirements; the layout scheme includes the number of mobile storage and charging robots, the designated locations, and the number and power of the battery packs carried by each mobile storage and charging robot; guiding the mobile storage and charging robots to reach the designated locations according to the layout scheme; the designated locations refer to the service waiting locations assigned to the mobile storage and charging robots in the layout scheme.

[0005] Further, the characteristic parameters of the target scenario are identified, including: extracting features from the planning image to obtain planning features; aligning and fusing the scenario image and the point cloud image to obtain a first fused image; aligning and fusing the planning image and the first fused image to obtain a second fused image; performing image reconstruction based on the second fused image to obtain a reconstructed scenario model; extracting features from the reconstructed scenario model to obtain reconstructed characteristic parameters. Based on the differences between the planning features and the reconstructed characteristic parameters, the reconstructed scenario model and the reconstructed characteristic parameters are adjusted to obtain an initial scenario model and the characteristic parameters.

[0006] Further, obtaining the first fused image includes: extracting features from the scenario image to obtain scenario features; projecting the scenario features into 3D space to obtain three-dimensional scenario features; calculating the distance between each point cloud point in the point cloud image and the three-dimensional scenario feature points; based on the distance between the point cloud points and the three-dimensional scenario feature points, determining the probability of each point cloud point in the point cloud image being selected; the calculation formula for the probability of a point cloud point being selected is: where P sample (p) represents the probability of point cloud point p being selected; ρ p represents the point cloud density at point cloud point p; e * represents the exponential function; D p represents the minimum distance between point cloud point p and the three-dimensional scenario feature points; σ represents the adjustment parameter; σ p represents the Gaussian scale of the three-dimensional scenario feature point with the minimum distance from point cloud point p; based on the probability of a point cloud point being selected, initial point cloud features are selected, and a point cloud feature vector for each initial point cloud feature point is constructed; the three-dimensional feature vector of the three-dimensional scenario feature points is determined; the feature distance between the point cloud feature vector and the three-dimensional feature vector is calculated, and the point cloud features matching the three-dimensional scenario are determined to obtain a set of feature pairs; based on the set of feature pairs, a transformation matrix is determined; based on the transformation matrix, the point cloud image is mapped onto the scenario image to obtain the first fused image.

[0007] Further, determining the scenario requirements includes: predicting the maximum number of new energy vehicles parked during the process of the activity based on the scenario plan; the maximum number of parked vehicles refers to the maximum number of new energy vehicles parked at the parking lot during the duration of the activity; calculating the minimum entrance distance between the charging pile and the parking lot entrance and the minimum exit distance from the parking lot exit; the minimum entrance distance refers to the shortest distance between the charging pile and the parking lot entrance; the minimum exit distance refers to the shortest distance between the charging pile and the parking lot exit; based on the entrance distance, the exit distance, and the maximum number of parked vehicles, determining the number and location of the charging piles; based on the activity duration, the peak period, and the charging demand distribution, determining the power demand of the charging piles.

[0008] Further, determining the positions of the charging piles includes: selecting a plurality of charging piles as initial clustering centers based on the sum of the minimum entrance distance and the minimum exit distance; the number of the initial clustering centers is the maximum parking number; the distance between each initial clustering center is greater than the service distance of the mobile storage and charging robot; the service distance refers to the farthest moving distance set for the mobile storage and charging robot; clustering the charging piles in the parking lot based on the initial clustering centers to obtain a plurality of initial clustering clusters; the distance between the charging piles within each group of the initial clustering clusters is less than the service distance; for each of the initial clustering clusters, calculating the average position of all the charging piles within the cluster and taking the average position as a new clustering center; clustering the charging piles in the parking lot again based on the new clustering center to obtain new clustering clusters; repeating the update process of the clustering clusters until the clustering clusters are stable or the maximum number of iterations is reached to obtain a clustering result; determining the score of each clustering cluster based on the entrance distance and the exit distance of the charging piles in each clustering cluster in the clustering result; the entrance distance refers to the distance between the charging pile and the multiple entrances of the parking lot; the exit distance refers to the distance between the charging pile and the multiple exits of the parking lot; sorting the clustering clusters in the clustering result based on the score of each clustering cluster to obtain a sorting result; the sorting result is used to represent the convenience degree of the parking spaces at the charging piles entering and leaving the parking lot; selecting the charging piles in one or more clustering clusters as the charging piles required by the scenario based on the sorting result, and obtaining the positions of the charging piles.

[0009] Further, taking the average value of the charging pile scores in each clustering cluster as the score of each clustering cluster, and the calculation formula of the charging pile score is: where S core represents the score; im represents the entrance variable; IM represents the total number of the parking lot entrances; P im represents the probability that a new energy vehicle enters from the entrance im; e * represents the exponential function; λ represents the attenuation coefficient related to the size of the parking lot; d im，ch represents the distance between the entrance im and the charging pile ch; ch represents the charging pile variable; ex represents the exit variable; EX represents the total number of the parking lot exits; P ex represents the probability that a new energy vehicle leaves from the exit ex; d ex，ch represents the distance between the exit ex and the charging pile ch.

[0010] Further, a layout plan of the mobile storage and charging robot is collaboratively planned through a genetic algorithm, including: taking the number of the mobile storage and charging robots as the chromosome length, taking the specified positions as genes, initializing a population to obtain a variety of first layout plans; calculating the fitness value of each first layout plan based on a fitness function; screening to obtain a variety of second layout plans in a roulette wheel selection manner based on the fitness value; performing crossover and mutation operations on the second layout plans to obtain third layout plans; taking the third layout plans as new first layout plans, repeating the screening process and the crossover and mutation processes until a stop condition is reached; taking the second layout plans that reach the stop condition as the layout plan.

[0011] Further, the calculation formula of the fitness function is: where Fi represents the fitness function, and the lower the value, the higher the fitness; C represents an adjustment parameter; P p represents the predicted peak power demand of electric energy; T p represents the predicted duration of the peak electric energy; v represents the moving speed of the mobile storage and charging robot; represents the minimum average value of the distances from the charging piles to the specified positions of the mobile storage and charging robots; E b represents the capacity of the battery; T s represents the time required for the mobile storage and charging robot to replace the battery; M represents the number of battery packs carried by the robot; T represents the activity duration.

[0012] Further, guiding the mobile storage and charging robot to reach the specified position includes: determining the initial map of the target scenario and the initial pose of the mobile storage robot; collecting environmental data through sensors arranged on the mobile storage robot; performing a SLAM algorithm to update the initial map and the initial pose to obtain the current map and the current pose; calculating the optimal path for the mobile storage robot to reach the specified position from the current position based on the current map and the current pose through a path planning algorithm; the optimal path includes multiple moving positions; controlling the mobile storage and charging robot to reach the next moving position based on the optimal path and obtaining a new current position; judging whether the distance between the new current position and the specified position is less than a preset distance threshold; if so, determining that the mobile storage and charging robot reaches the specified position and stopping the iteration; if not, repeating the operation of updating the current position based on the current map and the current pose.

[0013] Further, the initial map is an initial scene model, and the initial scene model is a modeling of the target scenario obtained by fusing a scene image, a point cloud image, and a planning image.

[0014] The technical solutions of the embodiments of the present invention have at least the following advantages and beneficial effects:

[0015] The present invention improves the layout efficiency and practicability of mobile storage and charging robots in specific scenarios by automatically identifying specific scenarios and collaborating to plan the layout scheme of the robots according to the scenario characteristics and layout requirements.

[0016] The present invention obtains a reconstructed scene model through the second fused image, and can correct the first fused image through more accurate features in the planning image, so that the obtained second fused image can more accurately represent the layout of the actual scene, thereby improving the accuracy of the reconstructed scene model.

[0017] The present invention can improve the efficiency of extracting point cloud features by extracting point cloud features from the extracted scene features, and avoid spending unnecessary time on mismatched features.

[0018] The present invention clusters based on the distance between the charging piles and the entrances and exits to obtain the positions of multiple charging piles, enabling participants to enter and exit the parking lot more conveniently and reducing the time for participants to detour in the parking lot.

[0019] The present invention determines the layout scheme through a genetic algorithm, which can make the positions of the mobile storage and charging robots more reasonable, ensure the charging efficiency, reduce the number of mobile storage and charging robots, and reduce the cost. Brief Description of the Drawings

[0020] Figure 1 It is an exemplary flowchart of a collaborative layout method for mobile storage and charging robots based on specific scenarios provided by the present invention. Detailed Embodiment

[0021] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. Usually, the components of the embodiments of the present invention described and illustrated herein can be arranged and designed in various different configurations.

[0022] Figure 1 It is an exemplary flowchart of a collaborative layout method for mobile storage and charging robots based on specific scenarios provided by the present invention. As Figure 1 shown, the collaborative layout method for mobile storage and charging robots based on specific scenarios provided by the present invention includes the following content:

[0023] Step 110: Identify the characteristic parameters of the target scenario by fusing the scene image, point cloud image, and planning image; the characteristic parameters include the area, shape, and obstacle positions of the scene; wherein, when performing image fusion, the point cloud features of the point cloud image are extracted from the scene features of the scene image. The target scenario refers to the scenario where mobile energy storage and charging robots need to be deployed; for example, scenarios such as concerts and exhibitions. The characteristic parameters can be related to the characteristics of the building. The scene image can refer to the image obtained by an image acquisition device such as a camera; for example, visible light images and depth images. The point cloud image refers to the point cloud data obtained by a Light Detection and Ranging (LiDAR); for example, 2D point cloud data and 3D point cloud data. The planning image can refer to the image for venue planning during scene layout; for example, architectural drawings, etc.

[0024] Step 120: Determine the scene requirements according to the scene plan of the target scenario; the scene plan includes the activity duration, peak hours, vehicle ratio, and number of participants; the scene requirements include the number, location, and power requirements of the charging piles. The scene plan can be related to the usage purpose of the scene.

[0025] Step 130: Collaboratively plan the layout plan of the mobile energy storage and charging robots according to the characteristic parameters and the scene requirements; the layout plan includes the number of mobile energy storage and charging robots, the designated locations, and the number and power of the battery packs carried by each mobile energy storage and charging robot. The mobile energy storage and charging robot refers to an automated device for providing mobile charging services for new energy vehicles.

[0026] Step 140: Guide the mobile energy storage and charging robots to reach the designated locations according to the layout plan; the designated locations refer to the service waiting locations assigned to the mobile energy storage and charging robots in the layout plan. The robots can be deployed to the designated locations through the Simultaneous Localization and Mapping (SLAM) algorithm.

[0027] Identify the characteristic parameters of the target scene, including: extracting features from the planning image to obtain planning features. The planning features can refer to features related to building information. Image processing and feature extraction can be performed on the planning image through a computer vision library, a deep learning model, etc. to obtain planning features. For example, the Scale Invariant Feature Transform (SIFT) or the Speeded-Up Robust Features (SURF) algorithm can be used to extract features from the planning image. After aligning and fusing the scene image and the point cloud image, a first fused image is obtained. The first fused image refers to a new scene image with point cloud information obtained after mapping the point cloud image onto the scene image. After aligning and fusing the planning image and the first fused image, a second fused image is obtained. The second fused image refers to an image with scene and point cloud information obtained after mapping the first fused image onto the planning image. Since the planned features are not easily changed, such as the position of the building wall, etc., by performing image fusion on these relatively accurate features, the obtained second fused image can more accurately represent the layout of the actual scene. The planning image and the first fused image can be fused by fusing the scene image and the point cloud image to obtain the second fused image. Based on the second fused image, image reconstruction is performed to obtain a reconstructed scene model. The reconstructed scene model can refer to a virtual scene model obtained by modeling the scene based on the data acquired by the sensor. Specifically, 3D reconstruction can be performed through the point cloud information in the second fused image to obtain the reconstructed scene model; of course, the second fused image can also be processed by other methods to obtain the reconstructed scene model, such as various image-based modeling methods, etc. Feature extraction is performed on the reconstructed scene model to obtain reconstructed characteristic parameters. The reconstructed characteristic parameters can refer to the characteristic parameters extracted from the reconstructed scene model. Feature extraction can be performed on the reconstructed scene model through various feasible methods, including but not limited to 3D descriptors, global feature extraction, and deep learning methods. Based on the difference between the planning features and the reconstructed characteristic parameters, the reconstructed scene model and the reconstructed characteristic parameters are adjusted to obtain an initial scene model and the characteristic parameters. For example, based on the distance between the position of the reconstructed characteristic parameter and the position of the corresponding planning feature point, the positions of other pixel points can be corrected.

[0028] Obtaining the first fused image includes: extracting features from the scene image to obtain scene features. The scene features can be used to represent the features within the scene image. Projecting the scene features into 3D space to obtain three-dimensional scene features. The three-dimensional scene features can be used to represent the features of the scene features in 3D space. Among them, the calculation formula for the three-dimensional scene features is: Where X i' represents the abscissa of the three-dimensional scene feature; Y i ' represents the ordinate of the three-dimensional scene feature; Z i ' represents the vertical coordinate of the three-dimensional scene feature; fx represents the focal length of the camera in the horizontal axis direction; fy represents the focal length of the camera in the y-axis direction; cx represents the horizontal coordinate offset of the camera's principal optical center in the image coordinate system; cy represents the vertical coordinate offset of the camera's principal optical center in the image coordinate system; x i represents the abscissa of the scene feature; y i represents the ordinate of the scene feature; depth(x i , y i ) represents the depth value of the feature point in the scene feature, which can be obtained at least from the depth image obtained by recognizing the depth camera; i represents the three-dimensional scene feature variable. Calculate the distance between each point cloud point in the point cloud image and the three-dimensional scene feature point. The distance can be calculated by means such as Euclidean distance and cosine similarity. Based on the distance between the point cloud point and the three-dimensional scene feature point, determine the probability of each point cloud point in the point cloud image being selected; the calculation formula for the probability of the point cloud point being selected is: where, P sampe (p) represents the probability of the point cloud point p being selected; ρ p represents the point cloud density at the point cloud point p; e * represents the exponential function; D p represents the minimum distance between the point cloud point p and the three-dimensional scene feature point; σ represents the adjustment parameter; σ p represents the Gaussian scale of the three-dimensional scene feature point with the smallest distance from the point cloud point p; Based on the probability of the point cloud point being selected, select the initial point cloud feature and construct the point cloud feature vector of each initial point cloud feature point. The point cloud feature can refer to the feature of the selected point cloud image, and the initial point cloud feature can refer to the point cloud feature selected for the first time. The initial point cloud feature needs to be processed to obtain the final point cloud feature. The point cloud feature vector can be used to reflect the information of the selected point cloud feature. For example, the point cloud feature vector can be determined by the fast point feature histogram. Specifically, the point cloud points with a probability greater than the preset probability threshold can be selected as the initial point cloud feature. The preset probability threshold can refer to the minimum value preset for selecting the point cloud points. If it exceeds the preset probability threshold, it is selected as the initial point cloud point. The FPFH feature of the point cloud point is (X j , Y j , Z j , P sample (j), r), P sample (j) represents the probability of the point cloud point j being selected, which is calculated by P sample (p); X j represents the abscissa of the point cloud point j in the point cloud image Y j represents the ordinate of the point cloud point j in the point cloud image; Zj Denote the vertical coordinate of the point cloud point j in the point cloud image; [*] T Denote the transpose of the matrix; r represents the local neighborhood radius for calculating FPFH. Determine the three-dimensional feature vector of the three-dimensional scene feature point. The three-dimensional feature vector can be used to reflect the information of the three-dimensional scene feature point. For example, the three-dimensional feature vector can be obtained by the SIFT algorithm (feature detection algorithm). Specifically, the three-dimensional feature vector is (x i , y i , σ i , d i ), d i Denote the feature descriptor of the three-dimensional scene feature point i; σ i Denote the Gaussian scale of the three-dimensional scene feature point i. Calculate the feature distance between the point cloud feature vector and the three-dimensional feature vector, determine the point cloud feature that matches the three-dimensional scene, and obtain the feature pair set. The feature pair set includes multiple pairs of corresponding three-dimensional scene features and point cloud features. Specifically, the initial point cloud feature points with a feature distance less than the preset feature distance can be used as point cloud points, and the set of point cloud points can be used as the point cloud feature. The preset feature distance refers to the maximum distance between the matching point cloud feature point and the three-dimensional scene feature point set in advance. When it is greater than this distance, the two points are considered unmatched. Based on the feature pair set, determine the transformation matrix. Based on the transformation matrix, map the point cloud image onto the scene image to obtain the first fusion image.

[0029] Determine the scene requirements, including: based on the scene plan, predict the maximum number of new energy vehicles parked during the activity; the maximum number of parked vehicles refers to the maximum number of new energy vehicles parked at the parking lot during the activity duration. The maximum number of parked vehicles can be obtained through various feasible methods, including but not limited to predicting through historical data or deep learning models. Calculate the minimum entrance distance between the charging pile and the parking lot entrance and the minimum exit distance from the parking lot exit; the minimum entrance distance refers to the shortest distance between the charging pile and the parking lot entrance; the minimum exit distance refers to the shortest distance between the charging pile and the parking lot exit. Based on the entrance distance, the exit distance, and the maximum number of parked vehicles, determine the number and location of the charging piles.

[0030] Based on the activity duration, the peak period, and the charging demand distribution, determine the power demand of the charging piles. The charging demand distribution can refer to the power demand distribution of new energy vehicles. The peak period and its charging demand distribution can be obtained by analyzing historical data. For example, create a charging demand curve based on the data to understand the demand distribution at different times; estimate the required average power and peak power based on the total energy demand and peak demand; select the appropriate power charging pile model and calculate the required quantity.

[0031] Determining the positions of charging piles includes: selecting multiple charging piles as initial clustering centers based on the sum of the minimum entrance distance and the minimum exit distance; the number of the initial clustering centers is the maximum parking number; the distance between each initial clustering center is greater than the service distance of the mobile storage and charging robot; the service distance refers to the farthest moving distance set for the mobile storage and charging robot. Based on the initial clustering centers, clustering the charging piles in the parking lot to obtain multiple initial clustering clusters; the distance between the charging piles within each group of the initial clustering clusters is less than the service distance. For each of the initial clustering clusters, calculate the average position of all the charging piles within the cluster and use the average position as the new clustering center. Based on the new clustering center, cluster the charging piles in the parking lot again to obtain new clustering clusters. Repeat the update process of the clustering clusters until the clustering clusters are stable or the maximum number of iterations is reached to obtain the clustering result. Based on the entrance distance and the exit distance of the charging piles in each clustering cluster in the clustering result, determine the score of each clustering cluster; the entrance distance refers to the distance between the charging pile and multiple entrances of the parking lot; the exit distance refers to the distance between the charging pile and multiple exits of the parking lot. Based on the scores of each clustering cluster, sort the clustering clusters in the clustering result to obtain a sorting result; the sorting result is used to represent the convenience degree of the parking spaces at the charging piles to enter and exit the parking lot. Based on the sorting result, select the charging piles in one or more clustering clusters as the charging piles required by the scenario and obtain the positions of the charging piles.

[0032] Take the average value of the charging pile scores in each clustering cluster as the score of each clustering cluster, and the calculation formula of the charging pile score is: where, S core represents the score; im represents the entrance variable; IM represents the total number of parking lot entrances; P im represents the probability that a new energy vehicle enters from entrance im; e * represents the exponential function; λ represents the attenuation coefficient related to the size of the parking lot; d im，ch represents the distance between entrance im and charging pile ch; ch represents the charging pile variable; ex represents the exit variable; EX represents the total number of parking lot exits; P ex represents the probability that a new energy vehicle leaves from exit ex; d ex，ch represents the distance between exit ex and charging pile ch.

[0033] Collaboratively plan the layout scheme of mobile energy storage and charging robots through a genetic algorithm, including: taking the number of the mobile energy storage and charging robots as the chromosome length and the specified positions as genes, initializing the population to obtain multiple first layout schemes. The first layout scheme can refer to the parental chromosome for genetic algorithm processing. Calculate the fitness value of each first layout scheme based on the fitness function. Based on the fitness value, screen to obtain multiple second layout schemes by means of roulette wheel selection. Perform crossover and mutation operations on the second layout schemes to obtain third layout schemes. Take the third layout schemes as the new first layout schemes, and repeat the screening process and the crossover and mutation process until the stop condition is reached. The stop condition includes but is not limited to that the fitness value is less than the preset fitness threshold or the number of iterations reaches the threshold, etc. Take the second layout scheme that reaches the stop condition as the layout scheme.

[0034] The calculation formula of the fitness function is: In it, Fi represents the fitness function, and the lower the value, the higher the fitness; C represents the adjustment parameter; P p represents the predicted peak power demand of electric energy; T p represents the predicted peak duration of electric energy; v represents the moving speed of the mobile energy storage and charging robot; represents the minimum average value of the distance from the charging pile to the specified position of the mobile energy storage and charging robot; F b represents the capacity of the battery; T s represents the time required for the mobile energy storage and charging robot to replace the battery; M represents the number of battery packs carried by the robot; T represents the activity duration.

[0035] Guiding the mobile storage and charging robot to reach a specified position includes: determining the initial map of the target scenario and the initial pose of the mobile storage robot. The initial map may refer to the map before updating the map based on the robot's traveling process. The initial map may be an initial scene model, which is a modeling of the target scenario obtained by fusing the scene image, the point cloud image, and the planning image. The initial pose may refer to the position and attitude information of the robot before moving. Collect environmental data through sensors installed on the mobile storage robot. The environmental data may include point cloud data, acceleration, angular velocity, and moving distance obtained by the robot, etc. Execute the SLAM (Simultaneous Localization and Mapping) algorithm to update the initial map and the initial pose to obtain the current map and the current pose. The current map refers to the map of the parking lot updated according to the environmental data obtained by the robot. The current pose refers to the current position and attitude of the robot. Based on the current map and the current pose, calculate the optimal path for the mobile storage robot to reach the specified position from the current position through a path planning algorithm; the optimal path includes multiple moving positions. The optimal path may refer to the most reasonable driving route for the robot to reach the specified position. The most reasonable may mean the fastest and safest. Based on the optimal path, control the mobile storage robot to reach the next moving position and obtain the new current position. The next moving position may refer to the position of the next point that the robot needs to reach on the optimal path. Determine whether the distance between the new current position and the specified position is less than a preset distance threshold; if so, determine that the mobile storage robot has reached the specified position and stop the iteration; if not, repeat the operation of updating the current position based on the current map and the current pose. The preset distance threshold may refer to the minimum error distance when the robot reaches the specified position set in advance.

[0036] The above are only the preferred embodiments of the present invention and are not used to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for collaborative deployment of mobile storage and charging robots based on specific scenarios, characterized in that: include: Identify the characteristic parameters of the target scene by fusing the scene image, point cloud image and planning image; The characteristic parameters include the area, shape and obstacle position of the scene; wherein, when image fusion is performed, the point cloud features of the point cloud image are obtained by extracting the scene features of the scene image; Determine the scenario requirements according to the scenario planning of the target scenario; the scenario planning includes the duration of the activity, peak hours, vehicle ratio and number of participants; the scenario requirements include the number, location and power requirements of charging piles; According to the characteristic parameters and the scenario requirements, collaboratively plan the layout of the mobile storage and charging robots; the layout plan includes the number of the mobile storage and charging robots, the designated locations, and the number and power of the battery packs carried by each mobile storage and charging robot; According to the arrangement plan, guiding the mobile storage and charging robot to arrive at a designated position; the designated position refers to a service waiting position allocated to the mobile storage and charging robot in the arrangement plan; Identify the characteristic parameters of the target scene, including: Extract features from the planning image to obtain planning features; Aligning the scene image and the point cloud image and fusing them to obtain a first fused image; Aligning the planning image with the first fused image and fusing them to obtain a second fused image; Performing image reconstruction based on the second fused image to obtain a reconstructed scene model; Extracting features from the reconstructed scene model to obtain reconstruction feature parameters; Based on the difference between the planning feature and the reconstruction feature parameter, the reconstruction scene model and the reconstruction feature parameter are adjusted to obtain the feature parameters of the initial scene model and the target scene; The first fused image is obtained, including: Extracting features from the scene image to obtain scene features; Projecting the scene features into 3D space to obtain three-dimensional scene features; Calculating the distance between each point cloud point in the point cloud image and a feature point of the three-dimensional scene; Determining the probability of each point cloud point in the point cloud image being selected based on the distance between the point cloud point and the feature point of the three-dimensional scene; Based on the probability of point cloud points being selected, the initial point cloud features are selected, and the point cloud feature vector of each initial point cloud feature point is constructed; Determine a three-dimensional feature vector of the three-dimensional scene feature point; Calculating a feature distance between the point cloud feature vector and the three-dimensional feature vector, determining a point cloud feature that matches the three-dimensional scene feature, and obtaining a feature pair set; Based on the set of feature pairs, determining a transformation matrix; Based on the transformation matrix, the point cloud image is mapped onto the scene image to obtain the first fused image.

2. The method for collaborative deployment of mobile storage and charging robots based on specific scenarios according to claim 1 is characterized in that: The calculation formula for the probability of a point cloud point being selected is: ； in, Represents point cloud points The probability of being selected; Represents point cloud points Point cloud density at ; represents the exponential function; Represents point cloud points Minimum distance to feature points of a 3D scene; Indicates adjustment parameters; Representation and point cloud The Gaussian scale of the 3D scene feature point with the smallest distance.

3. The method for collaborative deployment of mobile storage and charging robots based on specific scenarios according to claim 1 is characterized in that: Determine scenario requirements, including: Based on the scenario planning, predict the maximum number of new energy vehicles parked during the activity; the maximum number of parking refers to the maximum number of new energy vehicles parked in the parking lot during the activity; Calculate the minimum entrance distance between the charging pile and the parking lot entrance and the minimum exit distance between the charging pile and the parking lot exit; the minimum entrance distance refers to the shortest distance between the charging pile and the parking lot entrance; the minimum exit distance refers to the shortest distance between the charging pile and the parking lot exit; Determine the number and location of charging piles based on the entrance distance, the exit distance and the maximum parking number; Based on the activity duration, the peak period and the charging demand distribution, a power demand of the charging pile is determined.

4. The method for collaboratively deploying mobile storage and charging robots based on specific scenarios according to claim 3 is characterized in that: Determining the location of the charging station includes: Based on the sum of the minimum entrance distance and the minimum exit distance, multiple charging piles are selected as initial cluster centers; the number of the initial cluster centers is the maximum parking number; the distance of each initial cluster center is greater than the service distance of the mobile storage and charging robot; the service distance refers to the farthest moving distance of the mobile storage and charging robot; Based on the initial cluster center, the charging piles in the parking lot are clustered to obtain a plurality of initial clusters; the distance between the charging piles in each group of the initial clusters is less than the service distance; For each of the initial clusters, respectively calculate the average position of all charging piles in the cluster, and use the average position as a new cluster center; Based on the new cluster center, the charging piles in the parking lot are clustered again to obtain a new cluster; Repeat the clustering update process until the clustering is stable or the maximum number of iterations is reached to obtain the clustering result; Determine the score of each cluster based on the entrance distance and exit distance of the charging pile in each cluster in the clustering result; the entrance distance refers to the distance between the charging pile and multiple entrances of the parking lot; the exit distance refers to the distance between the charging pile and multiple exits of the parking lot; Based on the score of each cluster, the clusters in the cluster result are sorted to obtain a sorting result; the sorting result is used to indicate the convenience of the parking space at the charging pile to enter and exit the parking lot; Based on the sorting result, charging piles in one or more clusters are selected as charging piles required by the scenario, and the locations of the charging piles are obtained.

5. The method for collaboratively deploying mobile storage and charging robots based on specific scenarios according to claim 4 is characterized in that: The average value of the charging pile scores in each cluster is taken as the score of each cluster. The calculation formula of the charging pile score is: ； in, Indicates the rating; Indicates the entry variable; Indicates the total number of parking lot entrances; represents the probability of new energy vehicles entering from entrance im; represents the exponential function; represents the attenuation coefficient related to the parking lot size; Indicates the distance between the entrance im and the charging pile ch; ch represents the charging pile variable; represents the export variable; Indicates the total number of parking lot exits; represents the probability of new energy vehicles leaving from exit ex; Indicates the distance between the exit ex and the charging station ch.

6. The method for collaborative deployment of mobile storage and charging robots based on specific scenarios according to claim 1 is characterized in that: The deployment plan of the mobile storage and charging robot is collaboratively planned through genetic algorithms, including: Using the number of the mobile storage and charging robots as the chromosome length and the designated positions as genes, the population is initialized to obtain a plurality of first arrangement schemes; Based on the fitness function, calculating the fitness value of each first arrangement scheme; Based on the fitness value, multiple second arrangement schemes are screened by roulette; Performing a crossover mutation operation on the second arrangement scheme to obtain a third arrangement scheme; The third arrangement scheme is used as a new first arrangement scheme, and the screening process and the crossover mutation process are repeated until a stop condition is reached; The second arrangement scheme that reaches the stopping condition is used as the arrangement scheme.

7. The method for collaborative deployment of mobile storage and charging robots based on specific scenarios according to claim 6 is characterized in that: The calculation formula of the fitness function is: ； middle, Represents the fitness function, the lower the value, the higher the fitness; Indicates adjustment parameters; represents the predicted peak power demand for electrical energy; Indicates the predicted duration of the peak power; Indicates the moving speed of the mobile storage and charging robot; Indicates the minimum average value of the distance between the charging pile and the designated location of the mobile storage and charging robot; Indicates the capacity of the battery; Indicates the time required for the mobile storage and charging robot to replace the battery; Indicates the number of battery packs carried by the robot; Indicates the duration of the activity.

8. The method for collaborative deployment of mobile storage and charging robots based on specific scenarios according to claim 1 is characterized in that: Guide the mobile storage and charging robot to reach the designated location, including: Determining an initial map of the target scene and an initial position and posture of the mobile storage and charging robot; Collecting environmental data through sensors arranged on the mobile storage and charging robot; Execute the SLAM algorithm, update the initial map and the initial pose, and obtain the current map and the current pose; Based on the current map and the current position, the optimal path for the mobile storage and charging robot to reach the designated location from the current location is calculated by a path planning algorithm; the optimal path includes a plurality of mobile locations; Based on the optimal path, control the mobile storage and charging robot to reach the next moving position and obtain a new current position; Determine whether the distance between the new current position and the designated position is less than a preset distance threshold; if so, determine that the mobile charging robot has reached the designated position and stop iteration; if not, repeat the operation of updating the current position based on the current map and the current posture.

9. The method for collaboratively deploying mobile storage and charging robots based on specific scenarios according to claim 8, characterized in that: The initial map is an initial scene model, and the initial scene model is a modeling of the target scene obtained by fusing a scene image, a point cloud image and a planning image.

Images