Split-Type Mobile Charging Robot Based on Automatic Combination Connection Technology and Its Intelligent Control System

Abstract

The present invention relates to the technical field of split charging robots, and specifically, to a split mobile charging robot and its intelligent control system based on an automatic combination connection technology. It includes: a space modeling unit that uses a cloud large model to perform real-time modeling of a parking lot; a visual positioning unit that uses the PnP algorithm to identify the positions of the vehicle to be charged and the mobile charging robot based on the space modeling unit; an automatic planning and scheduling unit that automatically plans and schedules the nearest mobile charging robot based on the visual positioning unit according to a user call, for quickly responding to the charging needs of the user; a charging control unit that manages the charging process of the mobile charging robot and monitors various parameters during the charging process in real time. The design of the present invention automatically plans and schedules the nearest charging robot, mobile battery box, and automatic plug-and-unplug gun robot through the Dijkstra algorithm, thereby significantly shortening the waiting time of the user.

Description

Technical Field

[0001] The present invention relates to the technical field of split charging robots, and specifically, to a split mobile charging robot based on automatic combination connection technology and its intelligent control system. Background Art

[0002] With the popularization of electric vehicles, traditional fixed charging stations are difficult to meet the growing charging demand. Especially in the central urban areas, due to limited land resources, the cost of building fixed charging stations is high, and often all areas that need charging cannot be covered; many users face difficulties in finding available charging stations. Especially during peak hours, there may be queues at charging stations, resulting in inconvenient charging. Fixed charging stations usually require users to drive their cars to designated locations for charging, which not only increases the waiting time of users, but also requires users to manually operate the charging process in some cases, reducing the charging efficiency; the locations of fixed charging stations are usually fixed, which limits the charging options of users in case of emergencies; most existing charging solutions lack sufficient intelligent functions and cannot perform dynamic scheduling and optimization according to real-time needs. Therefore, a split mobile charging robot based on automatic combination connection technology and its intelligent control system are designed. Summary of the Invention

[0003] The purpose of the present invention is to provide a split mobile charging robot based on automatic combination connection technology and its intelligent control system to solve the problems of inconvenient charging, low charging efficiency, insufficient charging flexibility, and low intelligent level proposed in the above background art.

[0004] To achieve the above purpose, the present invention aims to provide an intelligent control system for a split mobile charging robot based on automatic combination connection technology, including:

[0005] A space modeling unit, which uses a cloud large model to perform real-time modeling on a parking lot;

[0006] A visual positioning unit, which uses the PnP algorithm to identify the positions of the vehicle to be charged and the mobile charging robot based on the space modeling unit, and optimizes the PnP algorithm considering the influence of noise and specular reflection of light on the recognition accuracy under low light conditions;

[0007] An automatic planning and scheduling unit, which automatically plans and schedules the nearest mobile charging robot based on the visual positioning unit according to a user call for quickly responding to the user's charging demand;

[0008] A charging control unit, which manages the charging process of the mobile charging robot and monitors various parameters during the charging process in real time.

[0009] As a further improvement of this technical solution, the spatial modeling unit uses a cloud-based large model to perform real-time modeling of the parking lot, including the following steps:

[0010] S1.1. Define a three-dimensional rectangular coordinate system for positioning all objects in the parking lot, and select the center position of the parking lot as the origin;

[0011] S1.2. Collect the basic structure information of the parking lot, including the positions, sizes, shapes of the parking spaces, and the positions of the entrances and exits. Use the basic structure information data to create a basic map of the parking lot and digitize it to form a three-dimensional model;

[0012] S1.3. Use the three-dimensional modeling software Blender to build a three-dimensional model of the parking lot. Import the digitized basic map data into the three-dimensional modeling software and build a three-dimensional model according to the basic map data, including parking spaces, columns, and walls;

[0013] S1.4. Upload the three-dimensional model to the cloud and use the computing resources of the cloud for large-scale data processing and model updating.

[0014] As a further improvement of this technical solution, the visual positioning unit identifies the positions of the vehicle to be charged and the mobile charging robot, including the following steps:

[0015] S2.1. Continuously capture image data in the parking lot through the visual acquisition module, identify the charging device identifier through the ORB algorithm, and extract the feature points of the vehicle to be charged and the mobile charging robot in the image;

[0016] S2.2. Use OCR technology to automatically identify the license plate number and bind the license plate number to the corresponding vehicle;

[0017] S2.3. Use the known three-dimensional model of the parking lot and the extracted vehicle information to calculate the actual position of the vehicle through the PnP algorithm. At the same time, considering the influence of noise and specular reflection of light on the recognition accuracy under low light conditions, optimize the PnP algorithm;

[0018] Among them, considering the influence of noise and specular reflection of light on the recognition accuracy under low light conditions, the optimization of the PnP algorithm is specifically as follows:

[0019] ;

[0020] Among them, represents the number of feature points of the vehicle to be charged and the mobile charging robot in the image, represents the index variable, represents the minimization objective function, represents the rotation matrix; represents the The weight of a feature point; Denotes the expected value operator; Denotes the two-dimensional coordinates of the ith feature point in the image; Denotes the camera intrinsic matrix; Denotes the rotation matrix from the center position coordinates of the parking lot to the camera coordinates; Denotes the translation vector from the center position coordinates of the parking lot to the camera coordinates; Denotes the three-dimensional coordinates of the ith feature point in the image;

[0021] S2.4. When a vehicle enters, leaves a parking space, or a charging device moves, the position information of the vehicle is updated in real time;

[0022] S2.5. Once the vehicle parks in a parking space, the vehicle is automatically bound to the corresponding parking space according to the license plate recognition result. The mobile charging robot uploads its position to the cloud parking lot model through its built-in GPS and updates the position information in the system in real time.

[0023] As a further improvement of this technical solution, in S2.1, the charging device identifier is recognized by the ORB algorithm, including the following steps:

[0024] S2.11. First, use the FAST method to quickly detect corner points, and then assign a direction to the detected corner points to make the feature description rotation-invariant;

[0025] S2.12. For each detected feature point, use the BRIEF descriptor to generate a fixed-length binary string;

[0026] S2.13. Match by comparing the BRIEF descriptors. If the match is successful, it is considered that the corresponding charging device identifier has been found.

[0027] As a further improvement of this technical solution, in S2.2, the license plate number is automatically recognized using OCR technology, including the following steps:

[0028] S2.21. Use the visual acquisition module to acquire an image containing the license plate, and intercept the frame containing the license plate from the video stream as the input;

[0029] S2.22. Preprocess the video stream;

[0030] S2.23. Use the Canny edge detection algorithm to find the edges in the image, and filter out the license plate area based on the size and ratio features of the license plate;

[0031] S2.24. Calculate the pixel distribution in the vertical direction, find the gaps between characters, and further perform horizontal segmentation on the characters in the license plate based on the vertical segmentation to detect and segment each character on the license plate;

[0032] S2.25. Query the vehicle information in the database using the recognized license plate number, and match the license plate number with the personal information of the vehicle owner.

[0033] As a further improvement of this technical solution, in S2.23, using the Canny edge detection algorithm to find the edges in the image includes the following steps:

[0034] S2.231. Use a Gaussian filter to smooth the grayscale image to remove noise;

[0035] S2.232. Apply the Sobel filter to calculate the gradient magnitude and direction of each pixel in the image. The gradient magnitude represents the edge strength, and the gradient direction points to the direction of the edge;

[0036] S2.233. Perform local maximum detection on the gradient magnitude according to the gradient direction, and only retain the local maximum as the edge candidate, and set the remaining pixels to zero;

[0037] S2.234. Set two thresholds: a low threshold and a high threshold. The high threshold is used to determine strong edge points, and the low threshold is used to determine weak edge points. Strong edge points are directly considered as edge points. Weak edge points are considered as edge points if they are connected to strong edge points; otherwise, they will be eliminated;

[0038] S2.235. Use a tracking algorithm to connect the edge points to form continuous edge segments.

[0039] As a further improvement of this technical solution, the automatic planning and scheduling unit includes an automatic scheduling module and an emergency scheduling module;

[0040] Among them, the automatic scheduling module is used to automatically plan and call the nearest mobile charging robot and automatic plugging and unplugging gun robot to respond quickly and plan the running path through the Dijkstra algorithm, and reach the specified location within the shortest time. Considering the impacts of obstacles, road surface flatness, and road surface width on path planning, substitute the above impact amounts into the Dijkstra algorithm process, and considering the impacts of rainfall and snow depth on the road surface in extreme rain and snow weather, further optimize the Dijkstra algorithm process;

[0041] The emergency dispatch module is used to automatically call service personnel to the designated location for service when there is no automatic plug-in or unplug-out robot, and the charging robot fails to recognize humans when it arrives at the designated location, and no one responds to the operation within 2 minutes.

[0042] As a further improvement of the technical solution, considering the influence of obstacles, road surface flatness and road surface width on path planning, the above influence quantities are substituted into the Dijkstra algorithm process, specifically:

[0043] ;

[0044] Considering the impact of rainfall and snow depth on the road surface in extreme rain and snow weather, the Dijkstra algorithm process is further optimized as follows:

[0045] ;

[0046] in, represents the edge weight after considering the impact of obstacles, road surface flatness and road width on path planning; represents the edge weight after considering the impact of rainfall and snow depth on the road surface in extreme rainy and snowy weather; represents the original edge weight; represents the obstacle influence function; represents the road surface smoothness function; represents the road width influence function; represents an edge; represents the observation conditions; represents the rainfall coefficient; Indicates rainfall; represents the snow depth coefficient; Indicates snow depth.

[0047] As a further improvement of the technical solution, the charging control unit manages the charging process of the mobile charging robot, including the following steps:

[0048] S3.1. The automatic gun insertion and extraction robot inserts the charging gun of the charging robot into the vehicle; the mobile battery box automatically connects to the charging robot; the charging process begins, the charging robot capacity is used first for charging, and the mobile charging robot's bidirectional DC / DC is used to automatically control the input voltage range to achieve wide-range voltage compatibility to meet the use of DC low-voltage piles and DC high-voltage piles; if the capacity of a single mobile battery box is insufficient, the remaining mobile battery boxes are automatically called to come over for replacement;

[0049] S3.2. After the constant power charging phase ends, the charging robot switches to slow charging with its own battery, and the mobile battery box automatically detaches to serve other vehicles.

[0050] S3.3. When the charging is about to end, automatically notify the vehicle owner and notify the mobile automatic plugging and unplugging gun robot to return to the vehicle. Automatically unplug the gun at the end of charging and enter the next round of service.

[0051] On the other hand, the present invention provides a split-type mobile charging robot based on an automatic combination connection technology, and applies the intelligent control system of the split-type mobile charging robot based on the automatic combination connection technology to the split-type mobile charging robot based on the automatic combination connection technology.

[0052] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0053] 1. In the split-type mobile charging robot based on the automatic combination connection technology and its intelligent control system, it can quickly respond to the user's charging request, automatically plan and dispatch the nearest charging robot, mobile battery box and automatic plugging and unplugging gun robot through the Dijkstra algorithm, thus significantly shortening the user's waiting time; by optimizing the weight function in the Dijkstra algorithm and considering factors such as rain, snow, obstacles, road surface flatness and road width, the system can plan the optimal path for the charging robot to ensure a quick and safe arrival at the destination.

[0054] 2. In the split-type mobile charging robot based on the automatic combination connection technology and its intelligent control system, the PnP algorithm is adopted, and the influence of noise and specular reflection of light on position recognition under low light conditions is considered, which enables the system to more accurately identify the positions of the vehicle and the mobile charging robot, reduces the position recognition error, and improves the reliability of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 is the overall flow block diagram of the present invention;

[0056] The meanings of the various reference numerals in the figure are as follows:

[0057] 1. Spatial modeling unit; 2. Visual positioning unit; 3. Automatic planning and scheduling unit; 4. Charging control unit. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0058] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention. Embodiment 1:

[0059] Please refer to Figure 1As shown in the figure, this embodiment provides an intelligent management and control system for a split-type mobile charging robot based on an automatic combination connection technology, including:

[0060] The space modeling unit 1 uses a cloud-based large model to perform real-time modeling of the parking lot;

[0061] In this embodiment, the cloud server usually has powerful computing resources, which can process a large amount of data and complex models. This is crucial for real-time modeling and updating the status of the parking lot. The cloud model can achieve real-time data sharing and synchronization, ensuring that each system component (such as mobile charging robots, automatic plug-and-unplug gun robots, etc.) can obtain the latest environmental information. Through the cloud, it is easy to achieve collaborative work between multiple terminals, such as the coordinated scheduling of multiple robots. The cloud-based modeling method is more easily integrated with new technologies and functions. The cloud model can be updated at any time, which means that the model can be adjusted in a timely manner as the parking lot changes. Using a cloud-based large model to perform real-time modeling of the parking lot includes the following steps:

[0062] S1.1. Define a three-dimensional rectangular coordinate system for positioning all objects in the parking lot, and select the center position of the parking lot as the origin;

[0063] S1.2. Collect the basic structure information of the parking lot, including the positions, sizes, shapes of parking spaces, and the positions of entrances and exits. Use the basic structure information data to create a basic map of the parking lot and digitize it to form a three-dimensional model;

[0064] S1.3. Use the three-dimensional modeling software Blender to build a three-dimensional model of the parking lot. Import the digitized basic map data into the three-dimensional modeling software and build a three-dimensional model according to the basic map data, including parking spaces, columns, and walls;

[0065] S1.4. Upload the three-dimensional model to the cloud and use the computing resources of the cloud for large-scale data processing and model updating.

[0066] The visual positioning unit 2 uses the PnP algorithm to identify the positions of the vehicle to be charged and the mobile charging robot based on the space modeling unit 1;

[0067] In this embodiment, identifying the positions of the vehicle to be charged and the mobile charging robot includes the following steps:

[0068] S2.1. Continuously capture image data in the parking lot through the visual acquisition module, identify the charging device identifier through the ORB algorithm, and extract the feature points of the vehicle to be charged and the mobile charging robot in the image;

[0069] The visual acquisition module includes multi-directional cameras;

[0070] Among them, the ORB algorithm combines the advantages of the FAST corner detection and the BRIEF descriptor, and can effectively detect and describe the feature points in the image. It adopts an efficient corner detection and descriptor generation method. The FAST detection method is very fast, and the BRIEF descriptor only requires simple bit operations to complete, which enables the entire process to be quickly executed in a real-time system and can achieve real-time update of the position information of the charging device. This is very important for the dynamic scheduling of mobile charging robots. Identifying the charging device identifier through the ORB algorithm includes the following steps:

[0071] S2.11. First, quickly detect the corner points using the FAST method, and then assign a direction to the detected corner points to make the feature description rotation-invariant;

[0072] Among them, the FAST method is a corner detection method based on pixel intensity comparison. It checks a certain number of consecutive pixels (usually 16) around a pixel point. If a sufficient number (such as 9 or more) of these pixels are brighter or darker than the central pixel by more than a certain threshold, then the central pixel is considered a corner point; in order to make the detected corner points rotation-invariant, a direction needs to be assigned to each corner point, which is completed by calculating the gradient direction of the pixels around the corner point; specifically, calculate the gradient direction of a group of pixels around each corner point, and then find a dominant direction as the direction of the corner point; in this way, even if the image rotates, the descriptors of the same feature point will remain unchanged;

[0073] S2.12. For each detected feature point, use the BRIEF descriptor to generate a fixed-length binary string, which can describe the appearance information around the feature point;

[0074] Among them, the descriptor generation process is as follows: First, select a group of pixel pairs, which are usually randomly selected in the neighborhood of the feature point; then, for each pair of pixels, compare their gray values. If the gray value of the first pixel is greater than that of the second pixel, record 1 in the binary string; otherwise, record 0; finally, all these bit positions are combined into a binary string, that is, the BRIEF descriptor;

[0075] S2.13. Perform matching by comparing the BRIEF descriptors. If the matching is successful, it is considered that the corresponding charging device identifier has been found;

[0076] Among them, the matching process is specifically as follows: Calculate the Hamming distance between the BRIEF descriptors of the corresponding feature points in two images, that is, the number of non-matching bit positions in the two binary strings. If the Hamming distance between the two descriptors is lower than a preset threshold, then these two feature points are considered to be matched;

[0077] S2.2. Automatically recognize the license plate number using OCR technology and bind the license plate number to the corresponding vehicle;

[0078] Among them, OCR technology can achieve automatic recognition of license plate numbers, reducing the need for manual intervention and improving work efficiency. The OCR system can process a large amount of image data in a short time, quickly recognize license plate numbers, and is also applicable to vehicle recognition on highways. Once the OCR system is installed and configured, it can run continuously without additional labor costs, saving a large amount of labor costs in the long run. The OCR system can easily add new license plate style recognition functions through software upgrades. It is applicable to various occasions, including parking lot management. The steps for automatically recognizing the license plate number using OCR technology are as follows:

[0079] S2.21. Use the visual acquisition module to collect images containing license plates and intercept frames containing license plates from the video stream as input;

[0080] S2.22. Preprocess the video stream, convert the color image to a grayscale image, convert the color image to a grayscale image, convert the color image to a grayscale image, convert the color image to a grayscale image;

[0081] S2.23. Use the Canny edge detection algorithm to find the edges in the image and filter out the license plate area based on the size and proportion characteristics of the license plate;

[0082] Furthermore, the Canny edge detection algorithm can effectively identify the true edges while reducing the situation of mislabeling noise or texture as edges. Through the non-maximum suppression step, the Canny algorithm ensures that each edge is only labeled once, and even if the edge is wide, a single edge response will be generated. This helps to avoid the problem of multiple responses during edge detection. The Canny algorithm can very accurately locate the center position of the edge. This is because it uses a Gaussian filter to smooth the image and uses the gradient direction to determine the exact position of the edge. The steps for using the Canny edge detection algorithm to find the edges in the image are as follows:

[0083] S2.231. Use a Gaussian filter to smooth the grayscale image to remove noise;

[0084] Among them, the Gaussian filter is a linear filter that uses the Gaussian distribution function as the weight coefficient to smooth the image;

[0085] S2.232. Apply the Sobel filter to calculate the gradient magnitude and direction of each pixel in the image. The gradient magnitude represents the edge strength, and the gradient direction points to the direction of the edge;

[0086] Among them, the gradient magnitude represents the intensity of the edge and is calculated from the horizontal and vertical gradient components: ; The gradient direction refers to the direction of the edge and is calculated by the following formula: ;

[0087] S2.233. Perform local maximum detection on the gradient magnitude according to the gradient direction, and only retain the local maximum as the edge candidate, and set the remaining pixels to zero;

[0088] Among them, when performing maximum detection, the gradient magnitude of each pixel is compared with its two adjacent pixels along the gradient direction; only when the gradient magnitude of a certain pixel is larger than both of its neighbors along the gradient direction, this pixel is considered likely to be part of the edge; all pixels that do not meet the conditions are set to zero, that is, they are considered not to be edges;

[0089] S2.234. Set two thresholds: a low threshold and a high threshold. The high threshold is used to determine strong edge points, and the low threshold is used to determine weak edge points. Strong edge points are directly considered as edge points. Weak edge points are considered as edge points if they are connected to strong edge points; otherwise, they will be eliminated;

[0090] S2.235. Use a tracking algorithm to connect the edge points to form continuous edge segments;

[0091] Among them, once the strong edge points and the connected weak edge points are determined, a tracking algorithm can be used to connect these edge points. The tracking algorithm can move along the edge points until it encounters the next strong edge point or reaches the end of the edge. These continuous edge points can form a complete edge segment, thus better depicting the contour of the object;

[0092] S2.24. Calculate the pixel distribution in the vertical direction, find the gaps between characters, and on the basis of vertical segmentation, further perform horizontal segmentation on the characters in the license plate to detect and segment each character on the license plate;

[0093] S2.25. Use the recognized license plate number to query the vehicle information in the database and match the license plate number with the personal information of the vehicle owner.

[0094] S2.3. Use the known three-dimensional model of the parking lot and the extracted vehicle information to calculate the actual position of the vehicle through the PnP algorithm. At the same time, considering the influence of noise under low light conditions and specular reflection of light on the recognition accuracy, optimize the PnP algorithm;

[0095] Among them, the PnP algorithm is used to estimate the pose of the camera relative to the three-dimensional model of the parking lot, so as to calculate the actual position of the object. The PnP algorithm uses the known 3D-2D correspondence to estimate the pose and can provide high-precision position information, which is essential for accurately determining the vehicle position. The PnP algorithm can be applied to different camera types and different scenes, and has high versatility. Whether it is a static scene or a dynamic scene, the PnP algorithm can work well, which is very important for mobile robots because they may need to navigate in a constantly changing environment. It can be combined with other visual algorithms, such as feature point detection and descriptor extraction algorithms, to improve positioning accuracy and robustness.

[0096] By optimizing the PnP algorithm, the position error caused by noise and specular reflection of light can be reduced, thereby improving the accuracy of position recognition. In low-light conditions, image noise is usually high, and specular reflection of light may produce false feature points, which will affect the performance of the PnP algorithm. By optimizing the algorithm, its robustness under complex lighting conditions can be improved; the optimized PnP algorithm can better handle light changes and environmental noise, thereby improving the consistency of recognition in different scenes; the optimized algorithm can better adapt to different environmental conditions, whether indoors or outdoors, day or night, it can maintain a high recognition accuracy. Considering the impact of noise and specular reflection of light under low-light conditions on recognition accuracy, the PnP algorithm is optimized as follows:

[0097] ;

[0098] ;

[0099] in, Indicates the number of feature points of the vehicle to be charged and the mobile charging robot in the image, Represents an index variable, represents the minimization objective function, represents the rotation matrix; Indicates The weight of feature points; Represents the expected value operator; Indicates The two-dimensional coordinates of feature points in the image; Represents the camera internal parameter matrix, including the camera's focal length, principal point position and other information; Represents the rotation matrix from the center coordinates of the parking lot to the camera coordinates; Represents the translation vector from the center coordinates of the parking lot to the camera coordinates; Indicates The three-dimensional coordinates of feature points in the image; represents the deviation between the actual position and the theoretical position of the th feature point, and the deviation is used to reflect the influence of specular reflection of light; represents an additional error term, which is used to compensate for the influence brought by noise and specular reflection of light under low light conditions; represents an adjustment parameter, which is used to balance the influence of the original error and the additional error term; represents the noise variance; represents a coefficient for adjusting the contribution of the reflection effect; represents the additional error caused by specular reflection of light;

[0100] Furthermore, if there is a feature point at its actual position in the three-dimensional space, but due to the specular reflection of light, the position of the observed feature point on the image deviates from the actual position, and the error is gradually approximated in the following way : First, estimate the camera internal parameter matrix (including parameters such as focal length and image center, which are obtained during the calibration process) and establish the rotation and translation relationships (it is necessary to obtain the rotation matrix from the coordinate system of the center position of the parking lot to the camera coordinate system and the translation vector ); Use and to calculate the image position of the feature point under ideal conditions, that is ; In order to find , minimize the error between and , and this error is expressed as ; Then solve the deviation

[0101] S2.4. When a vehicle enters, leaves the parking space, or the charging device moves, the position information of the vehicle is updated in real time;

[0102] S2.5. Once the vehicle parks in the parking space, the vehicle is automatically bound to the corresponding parking space according to the license plate recognition result. The mobile charging robot uploads its position to the cloud parking lot model through its own GPS and updates the position information in the system in real time.

[0103] The automatic planning and scheduling unit 3, based on the visual positioning unit 2, automatically plans and schedules the nearest mobile charging robot according to the user's call to quickly respond to the user's charging needs;

[0104] In this embodiment, the automatic planning and scheduling unit 3 includes an automatic scheduling module and an emergency scheduling module;

[0105] Among them, the automatic scheduling module is used to automatically plan and call the nearest mobile charging robot and automatic plugging and unplugging gun robot to respond quickly and plan the running path through the Dijkstra algorithm, and reach the specified position within the shortest time. Considering the impacts of obstacles, road surface flatness, and road width on path planning, the above impact quantities are substituted into the Dijkstra algorithm process. Also, considering the impacts of rainfall and snow depth on the road surface in extreme rain and snow weather, the Dijkstra algorithm process is further optimized;

[0106] The Dijkstra algorithm is a classic shortest path algorithm, which can ensure that the path from the starting point to any other node is the shortest. This means that the charging robot can reach the destination with the shortest distance, thereby reducing the driving time and energy consumption. If the weights of some edges in the network change, the shortest path can be quickly updated by re-running the Dijkstra algorithm. In many cases, the Dijkstra algorithm can give results within a reasonable time, so it is suitable for real-time path planning scenarios;

[0107] Furthermore, under rain and snow weather conditions, the ground may be slippery or the visibility may be blocked, which will directly affect the driving speed and stability of the robot. By adjusting the weight function to reflect these changes, it can be ensured that the robot selects a safer and more reliable path; if there are obstacles on the path, the robot needs to detour, which will increase the driving distance and time. By adjusting the weight function, the obstacles can be effectively avoided, ensuring that the robot selects a path without obstacles and improving the driving efficiency; uneven road surfaces may cause the robot to drive unstably or even be damaged. By adjusting the weight function, sections with poor road conditions can be avoided, improving the driving safety; narrow roads may not be suitable for the robot to pass, especially when encountering other vehicles or pedestrians. By adjusting the weight function, a wider route can be selected to reduce potential traffic conflicts; Considering these factors comprehensively can make the robot more intelligent when planning the path, not only simply finding the shortest path in terms of physical distance, but also considering various limiting conditions in actual driving, so as to select the optimal driving plan;

[0108] The consideration of the impacts of obstacles, road surface flatness, and road width on path planning and substituting the above impact quantities into the Dijkstra algorithm process specifically means:

[0109] ;

[0110] Considering the impacts of rainfall and snow depth on the road surface in extreme rain and snow weather, the further optimization of the Dijkstra algorithm process is:

[0111] ;

[0112] Among them,

[0113] ;

[0114] ;

[0115] ;

[0116] wherein, represents the edge weight after considering the impacts of obstacles, road surface flatness, and road width on path planning; represents the edge weight after considering the impacts of rainfall amount and snow depth on the road surface in extreme rain and snow weather; represents the original edge weight; represents the obstacle impact function; represents the road surface flatness function; represents the road width impact function; represents an edge; represents the observation condition; represents the rainfall coefficient; represents the rainfall amount; represents the snow depth coefficient; represents the snow depth; represents a constant factor; represents the flatness score of a road section (the flatness is evaluated on a scale from 1 to 5, and the higher the flatness score, the smaller the cost increment); represents a constant factor; represents the width of a road section, and the larger the width, the smaller the cost increment; is expressed as if there is an obstacle impact on edge (i.e., the condition represented by causes edge to be affected by obstacles), then the value of function is ( is a predefined positive value d, used to represent the impact degree of obstacles on the path cost), if there is no obstacle impact (i.e., the condition represented by does not cause edge to be affected by obstacles), then the value of function is 0, indicating that no additional cost is incurred due to obstacles;

[0117] When there is a graph, which contains four vertices: A, B, C, and D (the vertices include parking spaces, charging stations, and the current position of the robot), and the edges between them (connections between two points), as well as a set of weather impact coefficients, obstacle impact coefficients, road surface flatness coefficients, and road width coefficients,

[0118] Set the distances of all vertices to infinity, except the distance from the starting point A to itself, which is set to 0. Create a priority queue that initially contains all vertices and sorts them by distance value.

[0119] Take the vertex with the minimum distance from the priority queue as vertex A; for all neighbors of vertex A (such as B and C), use the new weight function Calculate the total distance of the paths passing through these neighbors; if the path passing through neighbor B is shorter than the currently known shortest path, update B's shortest path distance and update B's priority in the priority queue. Repeat this process until the end point D is found or the priority queue is empty.

[0120] Return the shortest path length from the starting point A to the end point D, and plan the driving path of the mobile charging robot based on the shortest path length.

[0121] The emergency dispatch module is used to automatically call service personnel to the designated location for service when there is no automatic plug-in or unplug-out robot, and the charging robot fails to recognize humans when it arrives at the designated location, and no one responds to the operation within 2 minutes.

[0122] The charging control unit 4 manages the charging process of the mobile charging robot and monitors various parameters during the charging process in real time;

[0123] Among them, the parameters include voltage, current, temperature, etc.;

[0124] In this embodiment, the charging process of managing the mobile charging robot includes the following steps:

[0125] S3.1. The automatic gun insertion and extraction robot inserts the charging gun of the charging robot into the vehicle; the mobile battery box automatically connects to the charging robot; the charging process begins, the charging robot capacity is used first for charging, and the mobile charging robot's bidirectional DC / DC is used to automatically control the input voltage range to achieve wide-range voltage compatibility to meet the use of DC low-voltage piles and DC high-voltage piles; if the capacity of a single mobile battery box is insufficient, the remaining mobile battery boxes are automatically called to come over for replacement;

[0126] S3.2. After the constant power charging phase ends, the charging robot switches to slow charging with its own battery. The mobile battery box automatically detaches to serve other vehicles or charges at nearby charging piles.

[0127] S3.3. When charging is about to end, the vehicle owner is automatically notified and the mobile automatic gun insertion and extraction robot is notified to return to the vehicle. When charging is completed, the gun is automatically extracted and the next round of service or charging is carried out nearby. Embodiment 2:

[0128] This embodiment provides a split-type mobile charging robot based on automatic combination connection technology, and applies the intelligent control system of the split-type mobile charging robot based on automatic combination connection technology to the split-type mobile charging robot based on automatic combination connection technology.

[0129] The foregoing has shown and described the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. The above embodiments and the descriptions in the specification are only preferred examples of the present invention and are not used to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements all fall within the scope of the present invention claimed.

Claims

1. An intelligent management and control system for a split mobile charging robot based on automatic combination connection technology, characterized in that: include: A space modeling unit (1), wherein the space modeling unit (1) uses a cloud-based large model to perform real-time modeling of the parking lot; A visual positioning unit (2), wherein the visual positioning unit (2) uses a PnP algorithm to identify the positions of the vehicle to be charged and the mobile charging robot based on the space modeling unit (1), and optimizes the PnP algorithm by taking into account the influence of noise and specular reflection of light under low light conditions on the recognition accuracy; An automatic planning and scheduling unit (3), wherein the automatic planning and scheduling unit (3) automatically plans and schedules the nearest mobile charging robot based on the visual positioning unit (2) according to the user's call, so as to quickly respond to the user's charging needs; A charging control unit (4), wherein the charging control unit (4) manages the charging process of the mobile charging robot and monitors various parameters during the charging process in real time; The charging control unit (4) manages the charging process of the mobile charging robot, including the following steps: S3.

1. The automatic gun insertion and extraction robot inserts the charging gun of the charging robot into the vehicle; the mobile battery box automatically connects to the charging robot; the charging process begins, the charging robot capacity is used first for charging, and the mobile charging robot's bidirectional DC / DC is used to automatically control the input voltage range to achieve wide-range voltage compatibility to meet the use of DC low-voltage piles and DC high-voltage piles; if the capacity of a single mobile battery box is insufficient, the remaining mobile battery boxes are automatically called to come over for replacement; S3.

2. After the constant power charging phase ends, the charging robot switches to slow charging with its own battery, and the mobile battery box automatically detaches to serve other vehicles. S3.

3. When charging is about to end, the vehicle owner is automatically notified and the mobile automatic gun insertion and extraction robot is notified to return to the vehicle. When charging is completed, the gun is automatically extracted to enter the next round of service.

2. According to claim 1, the intelligent management and control system of the split mobile charging robot based on the automatic combination connection technology is characterized in that: The space modeling unit (1) uses a cloud-based large model to perform real-time modeling of the parking lot, including the following steps: S1.

1. Define a three-dimensional rectangular coordinate system for locating all objects in the parking lot, and select the center of the parking lot as the origin; S1.

2. Collect the basic structural information of the parking lot, including the location, size, shape and entrance and exit locations of the parking spaces, create a basic map of the parking lot using the basic structural information data, and digitize it to form a three-dimensional model; S1.3, use 3D modeling software Blender to build a 3D model of the parking lot, import the digitized basic map data into the 3D modeling software, and build a 3D model based on the basic map data, including parking spaces, pillars, and walls; S1.

4. Upload the 3D model to the cloud and use the computing resources of the cloud to perform large-scale data processing and model updating.

3. The intelligent management and control system of the split mobile charging robot based on the automatic combination connection technology according to claim 1 is characterized in that: The visual positioning unit (2) identifies the positions of the vehicle to be charged and the mobile charging robot, comprising the following steps: S2.

1. Continuously capture image data in the parking lot through the visual acquisition module, identify the charging equipment logo through the ORB algorithm, and extract the feature points of the vehicle to be charged and the mobile charging robot in the image; S2.2, use OCR technology to automatically identify the license plate number and bind the license plate number to the corresponding vehicle; S2.3, using the known parking lot 3D model and the extracted vehicle information to calculate the actual position of the vehicle through the PnP algorithm, and at the same time, considering the influence of noise and specular reflection of light on recognition accuracy under low light conditions, the PnP algorithm is optimized; Among them, considering the influence of noise and specular reflection of light on recognition accuracy under low light conditions, the PnP algorithm is optimized as follows: ； in, Indicates the number of feature points of the vehicle to be charged and the mobile charging robot in the image, Represents an index variable, represents the minimization objective function, represents the rotation matrix; Indicates The weight of feature points; Represents the expected value operator; Indicates The two-dimensional coordinates of feature points in the image; Represents the camera intrinsic parameter matrix; Represents the rotation matrix from the center coordinates of the parking lot to the camera coordinates; Represents the translation vector from the center coordinates of the parking lot to the camera coordinates; Indicates The three-dimensional coordinates of feature points in the image; Indicates The deviation between the actual position and the theoretical position of a feature point, the deviation is used to reflect the effect of specular reflection of light; represents an additional error term used to compensate for the effects of noise and specular reflection of light in low light conditions; S2.

4. When a vehicle enters or leaves a parking space or a charging device moves, the vehicle's location information is updated in real time; S2.

5. Once the vehicle is parked in a parking space, the vehicle is automatically bound to the corresponding parking space based on the license plate recognition result. The mobile charging robot uploads the location to the cloud parking model through its own GPS and updates the location information in the system in real time.

4. The intelligent management and control system of the split mobile charging robot based on the automatic combination connection technology according to claim 3 is characterized in that: In S2.1, identifying the charging device identifier by using the ORB algorithm includes the following steps: S2.11, first use the FAST method to quickly detect corner points, and then assign a direction to the detected corner points; S2.

12. For each detected feature point, use the BRIEF descriptor to generate a fixed-length binary string; S2.

13. Matching is performed by comparing the BRIEF descriptors. If the match is successful, it is considered that the corresponding charging device identifier has been found.

5. The intelligent management and control system of the split mobile charging robot based on the automatic combination connection technology according to claim 3 is characterized in that: In S2.2, the automatic recognition of the license plate number using OCR technology includes the following steps: S2.21, using the visual acquisition module to collect images containing the license plate, and intercepting frames containing the license plate from the video stream as input; S2.22, preprocessing the video stream; S2.23, use the Canny edge detection algorithm to find the edges in the image, and filter out the license plate area based on the size and proportion characteristics of the license plate; S2.24, calculate the pixel distribution in the vertical direction, find the gaps between the characters, and further perform horizontal segmentation on the characters in the license plate based on the vertical segmentation, and detect and segment each character on the license plate; S2.

25. Use the identified license plate number to query the vehicle information in the database and match the license plate number with the personal information of the vehicle owner.

6. The intelligent management and control system of the split mobile charging robot based on the automatic combination connection technology according to claim 5 is characterized in that: In S2.23, the Canny edge detection algorithm is used to find the edges in the image, including the following steps: S2.231, use a Gaussian filter to smooth the grayscale image to remove noise; S2.232, apply the Sobel filter to calculate the gradient magnitude and direction of each pixel in the image, the gradient magnitude represents the edge strength, and the gradient direction points to the direction of the edge; S2.233, perform local maximum detection on the gradient amplitude according to the gradient direction, retain only the local maximum value as the edge candidate, and set the remaining pixels to zero; S2.234, set two thresholds: a low threshold and a high threshold. The high threshold is used to determine strong edge points, while the low threshold is used to determine weak edge points. Strong edge points are directly considered as edge points. If weak edge points are connected to strong edge points, they are considered as edge points; otherwise, they will be eliminated. S2.

235. Use a tracking algorithm to connect edge points to form continuous edge segments.

7. The intelligent management and control system of the split mobile charging robot based on the automatic combination connection technology according to claim 1 is characterized in that: The automatic planning and scheduling unit (3) comprises an automatic scheduling module and an emergency scheduling module; Among them, the automatic scheduling module is used to automatically plan and call the nearest mobile charging robot and automatic plug-in gun robot for quick response and plan the operation path through the Dijkstra algorithm to reach the designated location in the shortest time. Considering the influence of obstacles, road surface flatness and road surface width on path planning, the above influence quantities are substituted into the Dijkstra algorithm process, and considering the influence of rainfall and snow depth on the road surface in extreme rain and snow weather, the Dijkstra algorithm process is further optimized; The emergency dispatch module is used to automatically call service personnel to the designated location for service when there is no automatic plugging and unplugging robot, and the charging robot fails to recognize humans when it arrives at the designated location, and no one responds to the operation within 2 minutes.

8. The intelligent management and control system of the split mobile charging robot based on the automatic combination connection technology according to claim 7 is characterized in that: Taking into account the influence of obstacles, road surface flatness and road surface width on path planning, the above influence quantities are substituted into the Dijkstra algorithm process, specifically: ； Considering the impact of rainfall and snow depth on the road surface in extreme rain and snow weather, the Dijkstra algorithm process is further optimized as follows: ； in, represents the edge weight after considering the impact of obstacles, road surface flatness and road width on path planning; represents the edge weight after considering the impact of rainfall and snow depth on the road surface in extreme rainy and snowy weather; represents the original edge weight; represents the obstacle influence function; represents the road surface smoothness function; represents the road width influence function; represents an edge; represents the observation conditions; represents the rainfall coefficient; Indicates rainfall; represents the snow depth coefficient; Indicates snow depth.

9. A split mobile charging robot based on automatic combination connection technology, characterized in that: The intelligent management and control system of a split mobile charging robot based on automatic combination connection technology described in any one of claims 1 to 8 is applied to a split mobile charging robot based on automatic combination connection technology.

Images