Parking space cleaning method, device and equipment in garage scene and medium

Abstract

The invention discloses a parking space cleaning method and device in a garage scene, equipment and a medium, and relates to the technical field of parking space cleaning, and the method comprises the steps: carrying out the composition of a target garage scene, so as to generate a scene grid map, and continuously recognizing a plurality of parking spaces in the composition process, so as to generate a parking space semantic map; generating an inspection topology path by using the scene grid map, the parking space semantic map and the target traffic rule, and performing initial inspection path planning to obtain an initial inspection planning result; carrying out initial parking space detection and cleaning operation in the target garage scene according to the initial inspection planning result, and recording inspection related information, so as to carry out reinspection path planning based on the inspection related information after the detection and cleaning operation is completed, and obtaining a reinspection planning result; and performing reinspection based on the reinspection planning result to detect and clean the parking space missing in cleaning, and updating the routing inspection related information, so as to jump to the step of reinspection path planning after the reinspection cleaning is completed until a preset routing inspection stopping condition is met.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of parking space cleaning, and in particular to a parking space cleaning method, device, equipment and medium in a garage scenario. BACKGROUND

[0002] Current automatic parking space cleaning solutions usually directly reuse the cleaning logic of general scenarios to the garage scenario, which has redundant driving paths, more detours, low implementation efficiency, and unsatisfactory cleaning effect. In addition, the current solution adopts a global traversal type of conventional cleaning mode, but due to the dynamic parking of vehicles in the garage, many previously parked vehicles will be missed in the actual application, and the cleaning coverage rate is low and the efficiency is low. In addition, the current solution only considers obstacle avoidance logic, which may cause safety risks such as reverse driving and head-on collision with vehicles entering and exiting the garage.

[0003] In summary, how to improve the coverage rate, cleaning efficiency and safety of parking space cleaning in the garage scenario is a technical problem to be solved at present. SUMMARY

[0004] Therefore, the purpose of the present application is to provide a parking space cleaning method, device, equipment and medium in a garage scenario, which can improve the coverage rate, cleaning efficiency and safety of parking space cleaning in the garage scenario. The specific scheme is as follows:

[0005] In a first aspect, the present application provides a parking space cleaning method in a garage scenario, applied to a cleaning robot, comprising:

[0006] composing a target garage scenario to generate a corresponding scene grid map, and continuously identifying a plurality of parking spaces in the target garage scenario during the composition process to generate a parking space semantic map according to the obtained parking space identification result;

[0007] generating an inspection topology path using the scene grid map, the parking space semantic map and target traffic rules, and performing preliminary inspection path planning according to the inspection topology path to obtain a corresponding preliminary inspection planning result;

[0008] performing a preliminary parking space detection and cleaning operation in the target garage scenario according to the preliminary inspection planning result, and recording inspection related information during the cleaning process, so as to perform a re-inspection path planning based on the inspection related information after completing the preliminary parking space detection and cleaning operation, and obtain a re-inspection planning result;

[0009] Based on the re-inspection planning result, re-inspection is performed in the target garage scene to detect and clean the missed cleaning parking spaces in the re-inspection process, and the patrol related information is updated so as to jump to the step of performing re-inspection path planning based on the patrol related information after completing the re-inspection cleaning this time, until a preset patrol stop condition is met; the missed cleaning parking spaces are the parking spaces that are not cleaned in the initial parking space detection and cleaning operation process.

[0010] Optionally, before the mapping of the target garage scene, the method further comprises:

[0011] determining whether the cleaning robot has a parking space recognition function;

[0012] If the cleaning robot does not have a parking space recognition function, a parking space recognition sensor is installed, and an external parameter calibration is performed on the parking space recognition sensor to obtain a corresponding sensor external parameter; the sensor external parameter is an external parameter of the parking space recognition sensor relative to the cleaning robot.

[0013] Optionally, the mapping of the target garage scene to generate a corresponding scene grid map, and continuously recognizing a plurality of parking spaces in the target garage scene in the mapping process to generate a parking space semantic map according to the obtained parking space recognition result, comprises:

[0014] continuously collecting scene environment data during movement in the target garage scene, and continuously recognizing a plurality of parking spaces in the target garage scene through the parking space recognition sensor to obtain parking space recognition results including parking space positioning information;

[0015] converting the scene environment data into corresponding scene grid data to generate a scene grid map of the target garage scene based on the scene grid data;

[0016] converting the parking space positioning information from a first coordinate system to a second coordinate system through the sensor external parameter to obtain converted parking space positioning information; the first coordinate system is a coordinate system of the parking space recognition sensor; the second coordinate system is consistent with a coordinate system of the scene grid map;

[0017] according to the converted parking space information, superimposing the plurality of parking spaces recognized in the parking space recognition result in the form of a semantic label into the scene grid map to obtain a parking space semantic map.

[0018] Optionally, the generating of the patrol topology path by using the scene grid map, the parking space semantic map and a target traffic rule comprises:

[0019] fusing the scene grid map and the parking space semantic map to obtain a fused scene binary image;

[0020] performing morphological processing on the scene binary image, and performing connected domain analysis on the scene binary image after the morphological processing to identify a plurality of connected domains, and determining a connected domain with a largest area in the plurality of connected domains as a target main passage region;

[0021] determining a target trunk path of the target garage scene from the target main passage region by a preset skeleton extraction algorithm, and performing edge detection on the target main passage region to determine a travel path of the cleaning robot according to a detection result obtained;

[0022] segmenting the travel path based on the target trunk path to obtain a plurality of segmented paths, and determining path travel directions of the plurality of segmented paths according to a target traffic rule and a direction relative to the target trunk path;

[0023] establishing path connection relationships for the plurality of segmented paths according to the path travel directions to generate corresponding segmented connection paths, and associating the plurality of segmented paths with the plurality of parking spaces according to the parking space semantic map to obtain an inspection topology path; the inspection topology path is a directed graph formed based on the segmented connection paths, the path travel directions, and the plurality of segmented paths.

[0024] Optionally, the process of performing initial parking space detection and cleaning operation in the target garage scene according to the initial inspection planning result and recording inspection related information in a cleaning process includes:

[0025] when it is detected that the cleaning robot inspects a target parking space located at a target segmented path, judging a parking space state of the target parking space;

[0026] if the parking space state indicates that the target parking space is not cleaned and not occupied, cleaning the target parking space; or if the parking space state indicates that the target parking space is cleaned, or not cleaned and occupied, skipping the target parking space;

[0027] recording an inspection frequency of the target segmented path and a cleaning condition of the target parking space associated with the target segmented path; the inspection related information includes the inspection frequency and the cleaning condition.

[0028] Optionally, the process of performing re-inspection path planning based on the inspection related information to obtain a re-inspection planning result includes:

[0029] determining a current position of the cleaning robot, and determining a target distance according to the current position and the path travel direction; the target distance is a distance between a path starting point of the segmented path and the cleaning robot;

[0030] The target re-inspection path is determined from the plurality of segmented paths based on the target distance and the inspection-related information;

[0031] The path cost of several topological paths is calculated using a preset path cost formula, and a re-inspection path is planned based on the path cost to generate a re-inspection planning result including the target re-inspection path; the several topological paths include the several segmented paths and the segmented connection paths.

[0032] Optionally, the step of calculating the path cost of several topological paths using a preset path cost formula, and performing re-examination path planning based on the path costs to generate a re-examination planning result including the target re-examination path, includes:

[0033] The path lengths and target inspection-related information of several topological paths are determined, and a preset path cost formula is calculated using the path lengths and target inspection-related information to obtain the path costs of the several topological paths.

[0034] Based on the path cost, a shortest path planning algorithm is used to plan the re-examination path to determine the target arrival route; the target arrival route is the route from the current position to the starting point of the target re-examination path.

[0035] The target arrival route is concatenated with the target re-inspection path to obtain the corresponding re-inspection planning result.

[0036] Secondly, this application provides a parking space cleaning device for a garage scenario, applied to a cleaning robot, comprising:

[0037] The parking space recognition module is used to construct a map of the target garage scene to generate a corresponding scene grid map, and continuously recognize several parking spaces in the target garage scene during the map construction process to generate a parking space semantic map based on the obtained parking space recognition results.

[0038] The path planning module is used to generate an inspection topology path using the scene grid map, the parking space semantic map and the target traffic rules, and to perform preliminary inspection path planning based on the inspection topology path to obtain the corresponding preliminary inspection planning results.

[0039] The information recording module is used to perform initial parking space detection and cleaning operations in the target garage scene according to the initial inspection planning results, and record inspection-related information during the cleaning process, so as to perform re-inspection path planning based on the inspection-related information after the initial parking space detection and cleaning operations are completed, and obtain the re-inspection planning results.

[0040] The scene re-inspection module is used to perform a re-inspection in the target garage scene based on the re-inspection planning results, so as to detect and clean the missed parking spaces during the re-inspection process, and update the inspection-related information, so that after the current re-inspection cleaning is completed, it will jump to the step of re-inspection path planning based on the inspection-related information until the preset inspection stop condition is met; the missed parking spaces are the parking spaces that were not cleaned during the initial parking space detection and cleaning operation.

[0041] Thirdly, this application provides an electronic device, comprising:

[0042] Memory, used to store computer programs;

[0043] A processor is used to execute the computer program to implement the aforementioned parking space cleaning method in a garage scenario.

[0044] Fourthly, this application provides a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, it implements the aforementioned parking space cleaning method in a garage scenario.

[0045] In this application, a target garage scene is mapped to generate a corresponding scene grid map. During the mapping process, several parking spaces in the target garage scene are continuously identified to generate a parking space semantic map based on the parking space identification results. An inspection topology path is generated using the scene grid map, the parking space semantic map, and target traffic rules. An initial inspection path is planned based on the inspection topology path to obtain the corresponding initial inspection planning result. Based on the initial inspection planning result, an initial parking space detection and cleaning operation is performed in the target garage scene. During the cleaning process, inspection-related information is recorded. After completing the initial parking space detection and cleaning operation, a re-inspection path is planned based on the inspection-related information to obtain a re-inspection planning result. Based on the re-inspection planning result, a re-inspection is performed in the target garage scene to detect and clean missed parking spaces and update the inspection-related information. After completing the current re-inspection cleaning, the process jumps to the step of re-inspection path planning based on the inspection-related information until a preset inspection stop condition is met. The missed parking spaces are those that were not cleaned during the initial parking space detection and cleaning operation. As can be seen from the above, this application constructs a scene grid map of the target garage scene, and identifies several parking spaces in the scene during the construction process. Combined with the parking space identification results, a parking space semantic map is generated. Based on the scene grid map, the parking space semantic map, and the target traffic rules, an inspection topology path is generated, and an initial inspection path is planned to obtain an initial inspection planning result. According to the initial inspection planning result, the initial parking space detection and cleaning operation is completed in the target garage scene, and inspection-related information is recorded simultaneously. After the initial cleaning is completed, a re-inspection path is planned based on the inspection-related information to obtain a re-inspection planning result. The re-inspection work is carried out according to the re-inspection planning result. During the re-inspection process, the missed parking spaces that were missed during the initial cleaning are detected and cleaned, and the inspection-related information is updated. Then, the re-inspection path planning and re-inspection operation are repeated until the preset inspection stop condition is met. In this way, through the process described in this application, the combination of scene grid map and parking space semantic map realizes accurate digital modeling of garage scene. Without changing the main hardware of the cleaning vehicle, it endows garage with parking space recognition and semantic map capabilities at extremely low adaptation cost. By generating inspection topology paths using scene grid map, parking space semantic map and target traffic rules, it can achieve high coverage and high efficiency of parking space cleaning while meeting normal traffic habits and safety constraints, ensuring the rationality and efficiency of inspection routes. With the iterative mechanism of initial inspection plus multiple re-inspections, it can effectively reduce the probability of missed parking spaces and improve the effectiveness of parking space cleaning. By continuously optimizing the re-inspection path through dynamic updates of inspection information, it improves the coverage and overall efficiency of garage parking space cleaning, ensuring the comprehensiveness and standardization of garage parking space cleaning work, thereby improving the coverage, cleaning efficiency and safety of parking space cleaning in garage scene. Attached Figure Description

[0046] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0047] Figure 1 This application discloses a flowchart of a parking space cleaning method in a garage setting.

[0048] Figure 2 This is a flowchart illustrating the timing of a parking space cleaning method in a garage setting disclosed in this application.

[0049] Figure 3 This is a schematic diagram of a cleaning robot with parking space recognition function disclosed in this application;

[0050] Figure 4 This is a schematic diagram of a scene grid map and a parking space semantic map disclosed in this application;

[0051] Figure 5 This is a schematic diagram of a path topology for an inspection topology path disclosed in this application;

[0052] Figure 6 This is a schematic diagram of a scene binary image disclosed in this application;

[0053] Figure 7 This is a schematic diagram of a coarse main channel diagram disclosed in this application;

[0054] Figure 8 This is a schematic diagram of a segmented backbone path diagram disclosed in this application;

[0055] Figure 9 This is a schematic diagram of a parking space cleaning device in a garage setting disclosed in this application;

[0056] Figure 10 This is a structural diagram of an electronic device disclosed in this application. Detailed Implementation

[0057] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0058] Current automated parking space cleaning solutions typically reuse cleaning logic from general scenarios directly in garage scenarios, resulting in redundant driving paths, excessive detours, low implementation efficiency, and unsatisfactory cleaning effects. Furthermore, their conventional global traversal cleaning mode means that, due to the dynamic parking spaces in garages, many previously occupied spaces are missed during actual application, failing to be cleaned promptly after subsequent vehicles leave, leading to low cleaning coverage and efficiency. In addition, current solutions only consider obstacle avoidance logic, which can easily lead to situations such as reverse driving, potentially causing safety risks when vehicles encounter each other while entering or exiting the garage.

[0059] To overcome the aforementioned technical problems, this application provides a parking space cleaning method in a garage setting, which can improve the coverage, cleaning efficiency, and safety of parking space cleaning in a garage setting.

[0060] See Figure 1 As shown, this invention discloses a parking space cleaning method in a garage scenario, applied to a cleaning robot, comprising:

[0061] Step S11: Construct a map of the target garage scene to generate a corresponding scene grid map, and continuously identify several parking spaces in the target garage scene during the map construction process to generate a parking space semantic map based on the obtained parking space identification results.

[0062] In this embodiment, a mapping operation is performed on the target garage scene to generate a corresponding scene grid map. Simultaneously, during the mapping process, several parking spaces within the scene are continuously identified, and a parking space semantic map is generated based on the parking space identification results. For example... Figure 2 The diagram shown is a flowchart illustrating the timing of a parking space cleaning method in a garage setting provided in this application.

[0063] It should be noted that before mapping the target garage scene, it is necessary to ensure that the cleaning robot has the ability to identify parking spaces. The processing flow is as follows: determine whether the cleaning robot has parking space recognition functionality; if the cleaning robot does not have parking space recognition functionality, install a parking space recognition sensor and perform extrinsic parameter calibration on the parking space recognition sensor to obtain the corresponding sensor extrinsic parameters; the sensor extrinsic parameters are the extrinsic parameters of the parking space recognition sensor relative to the cleaning robot. The parking space recognition sensor can be implemented using an RGB (Red, Green, Blue) camera combined with the YOLO (You Only Look Once, a deep learning-based object detection algorithm) algorithm. That is, first determine whether the cleaning robot has parking space recognition functionality; if not, install a parking space recognition sensor on it and perform extrinsic parameter calibration on the parking space recognition sensor to obtain the sensor extrinsic parameters relative to the cleaning robot. Figure 3The diagram shown is a schematic of a cleaning robot with parking space recognition function provided in this application.

[0064] Specifically, during movement within the target garage scene, scene environment data is continuously collected, and several parking spaces in the target garage scene are continuously identified through the parking space recognition sensor to obtain parking space recognition results including parking space positioning information; the scene environment data is converted into corresponding scene grid data to generate a scene grid map of the target garage scene based on the scene grid data; the parking space positioning information is converted from a first coordinate system to a second coordinate system through the sensor extrinsic parameters to obtain converted parking space positioning information; the first coordinate system is the coordinate system of the parking space recognition sensor; the second coordinate system is consistent with the coordinate system of the scene grid map; according to the converted parking space information, the several parking spaces identified in the parking space recognition results are superimposed onto the scene grid map in the form of semantic tags to obtain a parking space semantic map. That is, while moving within the target garage scene, scene environment data is continuously collected. Simultaneously, several parking spaces are identified using the parking space recognition sensor to obtain parking space recognition results containing parking space positioning information. The scene environment data is converted into scene grid data, thereby generating a scene grid map of the target garage scene. The parking space positioning information is then transformed from the sensor coordinate system to a coordinate system corresponding to the scene grid map using the sensor extrinsic parameters. Figure 1 The coordinate system is adjusted to obtain the transformed parking space location information. Then, based on the transformed parking space information, the identified parking spaces are superimposed onto the scene grid map in the form of semantic tags, ultimately generating a parking space semantic map. Figure 4 The diagram illustrates a scene grid map and parking space semantic map provided in this application. This embodiment simultaneously performs scene grid modeling and parking space semantic recognition, enabling rapid digitization and structural representation of the garage scene. The grid map accurately reconstructs the spatial topology of the garage, while the semantic map clarifies the specific location and distribution characteristics of parking spaces, achieving the fusion of garage scene spatial information and parking space semantic information. This provides scene data support for subsequent garage inspection path planning and parking space cleaning. By first determining the functionality and then adding equipment as needed and completing external parameter calibration, the smooth progress of subsequent parking space recognition and map construction is ensured. External parameter calibration unifies the coordinate reference of sensors and robots, providing the technical prerequisite for the conversion of parking space positioning information and the generation of semantic maps.

[0065] Step S12: Generate an inspection topology path using the scene grid map, the parking space semantic map, and the target traffic rules, and perform preliminary inspection path planning based on the inspection topology path to obtain the corresponding preliminary inspection planning results.

[0066] In this embodiment, an inspection topology path is generated by combining the scene grid map, the parking space semantic map, and the target traffic rules. Then, preliminary inspection path planning is performed based on the inspection topology path to obtain a preliminary inspection planning result that includes an initial inspection route covering all adjacent parking spaces starting from the current position. The target traffic rule is the right-hand drive rule. Figure 5 The diagram shown is a schematic diagram of a patrol topology path provided in this application.

[0067] It should be noted that the processing flow for generating the inspection topology path using the scene grid map, the parking space semantic map, and the target traffic rules is as follows: The scene grid map and the parking space semantic map are fused to obtain a fused scene binary image; morphological processing is performed on the scene binary image, and connected component analysis is performed on the processed scene binary image to identify several connected components, and the connected component with the largest area among these connected components is determined as the target main passage region; using a preset skeleton extraction algorithm, the target main path of the target garage scene is determined from the target main passage region, and edge detection is performed on the target main passage region to… The driving path of the cleaning robot is determined based on the obtained detection results. The driving path is segmented based on the target main path to obtain several segmented paths. The driving direction of each segmented path is determined according to the target traffic rules and its position relative to the target main path. Path connections are established between the segmented paths based on their driving directions to generate corresponding segmented connection paths. The segmented paths are then associated with the parking spaces according to the parking space semantic map to obtain an inspection topology path. This inspection topology path is a directed graph composed of the segmented connection paths, the driving directions, and the segmented paths. Specifically, the scene grid map and the parking space semantic map are fused. In the scene grid map, passable areas are set to 1, and inaccessible areas are set to 0, converting it into a binary image. Then, parking space frames are drawn in a filled manner on the binary image according to the parking space semantic map to obtain a scene binary image, such as... Figure 6 The image shown is a schematic diagram of a scene binary image provided in this application. Subsequently, morphological processing, such as opening, is performed on the scene binary image to obtain a coarse main channel image, thereby eliminating non-main channel areas in the scene information, such as parking space gaps connected in the front-to-back direction and parking space gaps caused by parking space recognition errors. Figure 7 The diagram shown is a schematic of a coarse main channel map provided in this application. Connectivity analysis is then performed on the coarse main channel map to filter out isolated regions outside the main channel region. The connected region with the largest area among the identified connected regions is then determined as the target main channel region.

[0068] It should be further noted that after determining the target main channel region, the skeleton distribution of the target main channel region is extracted as the target main path using a preset skeleton extraction algorithm (such as the Zhang-Suen thinning algorithm (an iterative algorithm for binary image processing)). The target main path is then segmented based on its path nodes to obtain a segmented main path map, as shown below. Figure 8 The diagram shown is a schematic representation of a segmented main path provided in this application. Figure 8 As shown, the target main path is divided into three segments: a, b, and c. Simultaneously, edge detection is performed on the target main channel area to generate an edge-adjacent path for the main channel area, determining the cleaning robot's travel path. Based on the target main path, the travel paths on both sides of the main channel are segmented to obtain several segmented paths, such as... Figure 5 As shown, paths 1-6 are the segmented paths, i.e., the various segmented paths. The direction of travel is determined by combining the target traffic rules with the relative position of each segmented path to its corresponding target main road, such as... Figure 5 The directions of paths 1-6 shown are used to ensure that the travel direction is to the right. Based on the travel direction, path connections are established for each segmented path to generate segmented connected paths, for example... Figure 5 As shown, edge path segments 1 and 2, 2 and 3, 4 and 5 were originally continuous edge paths. After segmentation, their original connection relationship is maintained by establishing shared nodes. Bidirectional connection relationships are established between edge path segments belonging to both sides of the same main path, which are called transition path segments, i.e., the segmented connection paths (such as...). Figure 5 The path segments ①, ②, and ③ shown are primarily designed to ensure full connectivity in the final topology. For example... Figure 5 Side path segments 2 and 3 belong to the two sides of the main path a, respectively. The nodes at the beginning and end of side path segments 2 and 3 are bidirectionally connected using transition path segment ①. Finally, the segmented paths are associated with the parking spaces using the parking space semantic map, ultimately generating a directed graph-like inspection topology path composed of the segmented connecting paths, the path travel direction, and the segmented paths, for example... Figure 5As shown, the three parking spaces in the upper right corner belong to the edge path segment 2. The essence of the inspection topology path is a directed graph composed of edges and nodes formed by edge path segments and transition path segments. In this way, this embodiment integrates multiple types of map information and traffic rules for path planning, which can take into account the spatial topological constraints of the garage scene, the distribution characteristics of parking spaces, and the traffic regulations during the inspection process. The generated initial inspection path is both comprehensive and reasonable, and safe, ensuring the orderly and efficient conduct of subsequent initial parking space inspection and cleaning work. The systematic path construction method from map fusion, main channel extraction to path segmentation and direction planning can fully combine the spatial structure and parking space distribution characteristics of the garage scene, while following traffic rules to ensure the reasonableness of path traffic. The generated inspection topology path has both comprehensive coverage and execution feasibility.

[0069] Step S13: Based on the initial inspection planning results, perform initial parking space inspection and cleaning operations in the target garage scene, and record inspection-related information during the cleaning process. After completing the initial parking space inspection and cleaning operations, perform re-inspection path planning based on the inspection-related information to obtain the re-inspection planning results.

[0070] In this embodiment, the initial parking space inspection and cleaning operation is carried out in the target garage scene according to the initial inspection planning results. At the same time, inspection-related information is recorded during the cleaning process. After the initial inspection and cleaning is completed, the re-inspection path is planned based on the recorded inspection-related information to generate the corresponding re-inspection planning results.

[0071] It should be noted that the processing flow during the initial parking space detection and cleaning operation in the target garage scenario is as follows: When the cleaning robot detects that it has inspected a target parking space located on the target segment path, the parking space status of the target parking space is determined; if the parking space status indicates that the target parking space is not cleaned and is not occupied, then the target parking space is cleaned; if the parking space status indicates that the target parking space is cleaned, or is not cleaned and is occupied, then the target parking space is skipped; the number of inspections of the target segment path and the cleaning status of the associated target parking spaces are recorded; the inspection-related information includes the number of inspections and the cleaning status. That is, when the cleaning robot patrols to the target parking space corresponding to the target segment path, it first determines the parking space status. If the parking space is uncleaned and not occupied by a car, it cleans it. After cleaning, it marks the parking space as clean and returns to the patrol path to continue patrolling. If the target parking space is cleaned, or uncleaned but occupied, it skips the patrol directly. At the same time, it records the number of patrols of the target segment path and the cleaning status of the associated target parking space to obtain the patrol-related information including the number of patrols and the cleaning status.

[0072] It should be noted that the processing flow for re-inspection path planning based on the inspection-related information to obtain the re-inspection planning result is as follows: Determine the current position of the cleaning robot, and determine the target distance based on the current position and the path travel direction; the target distance is the distance between the starting point of the segmented path and the cleaning robot; determine the target re-inspection path from the segmented paths based on the target distance and the inspection-related information; calculate the path cost of several topological paths using a preset path cost formula, and perform re-inspection path planning based on the path cost to generate a re-inspection planning result including the target re-inspection path; the several topological paths include the segmented paths and the segmented connecting paths. That is, first determine the current position of the cleaning robot, calculate the target distance between the starting point of the segmented path and the robot based on the current position and the path travel direction, and then select the target re-inspection path from the segmented paths based on the target distance and the inspection-related information. The specific formula for determining the target re-inspection path is as follows:

[0073] ;

[0074] Wherein, `path` is the set of path segments with the fewest current inspection counts and containing uncleaned parking spaces among all inspection segments, and is the candidate path set for determining the target re-inspection path; `dist` represents the path segment length; `n_unclean` represents the number of uncleaned parking spaces in the path segment, which is the cleanliness status in the inspection-related information; `arg min` represents the target re-inspection path that minimizes the function value. It can be understood that after selecting the target re-inspection path, the re-inspection path planning can be performed based on the following conditions: quickly reaching the re-inspection target point, covering historically uncovered paths as much as possible, and avoiding repeatedly re-inspecting the same inspection path. Subsequently, using a preset path cost formula, the path cost of several topological paths, including the segmented path and the segmented connecting path, is calculated, and re-inspection path planning is carried out based on the path cost, finally generating a re-inspection planning result containing the target re-inspection path.

[0075] It should be further pointed out that the process of calculating the path cost of several topological paths and planning a re-inspection path based on the path cost is as follows: The path length and target inspection-related information of several topological paths are determined, and a preset path cost formula is calculated using the path length and target inspection-related information to obtain the path cost of the several topological paths; based on the path cost, a preset shortest path planning algorithm is used to plan the re-inspection path to determine the target arrival route; the target arrival route is the route from the current position to the starting point of the target re-inspection path; the target arrival route is concatenated with the target re-inspection path to obtain the corresponding re-inspection planning result. That is, the path length and target inspection-related information of several topological paths are first determined, and the path cost corresponding to each topological path is calculated using a preset path cost formula by combining these two types of data. The specific expression of the preset path cost formula is as follows:

[0076] ;

[0077] Where path is the path of all inspection segments; The weight is the path segment length. The path segment is weighted by its historical cleaning status; len is the length of the path segment; count_cross is the number of times the path segment has been inspected; n_unclean is the number of uncleaned parking spaces on the path segment; and cost represents the path cost. Based on the path cost, a preset shortest path planning algorithm, such as the Floyd-Warshall algorithm, is used to plan the re-inspection path, determining the target arrival route from the current location to the starting point of the target re-inspection path. Finally, the target arrival route is concatenated with the target re-inspection path to generate the corresponding re-inspection planning result. In this way, this embodiment plans the re-inspection path based on the inspection information recorded during the initial inspection. It can formulate a targeted re-inspection path based on the actual situation of the initial inspection, effectively ensuring the targeting of the re-inspection work, avoiding invalid and repeated inspections, and helping to improve the full coverage and overall efficiency of garage parking space cleaning. The differentiated cleaning treatment method based on the real-time status of parking spaces can avoid invalid work on cleaned parking spaces, while avoiding misoperation of occupied parking spaces, ensuring the accuracy and efficiency of the initial parking space inspection and cleaning work. The re-inspection path planning combined with the robot's real-time position, historical inspection information and path cost model can prioritize the path corresponding to uncleaned parking spaces, while taking into account the travel cost of the path, effectively improving the targeting and execution efficiency of the re-inspection work, and avoiding repetitive and inefficient path planning.

[0078] Step S14: Based on the re-inspection planning results, a re-inspection is performed in the target garage scene to detect and clean the missed parking spaces during the re-inspection process, and to update the inspection-related information so that after the current re-inspection and cleaning is completed, the process jumps to the step of re-inspection path planning based on the inspection-related information until the preset inspection stop condition is met; the missed parking spaces are those that were not cleaned during the initial parking space detection and cleaning operation.

[0079] In this embodiment, a re-inspection is conducted in the target garage scenario based on the re-inspection planning results. During the re-inspection, missed parking spaces were detected and cleaned during the initial cleaning, and the inspection-related information was updated. After completing the re-inspection cleaning, a re-inspection path was planned again based on the updated inspection-related information. This process was repeated until a preset inspection stop condition was met, thus ensuring the final cleaning coverage. The preset inspection stop condition could be that each path had reached its maximum number of inspections, all parking spaces had been cleaned, or a cleaning task termination instruction had been received. It is understood that the logic for re-inspection cleaning is the same as the logic for the initial inspection, and will not be elaborated further here. In this way, this embodiment employs an iterative loop mechanism of re-inspection cleaning, inspection-related information updating, and path replanning for parking spaces with changing parking conditions. This mechanism continuously tracks the cleaning status of missed parking spaces and optimizes the re-inspection path through dynamically updated inspection information, ensuring that all parking spaces in the garage are thoroughly cleaned, effectively improving the coverage and precision of the cleaning operation.

[0080] As can be seen from the above, the embodiments of this application construct a scene grid map of the target garage scene, and at the same time identify several parking spaces in the scene during the construction process. Combine the parking space identification results to generate a parking space semantic map. Based on the scene grid map, the parking space semantic map and the target traffic rules, an inspection topology path is generated, and an initial inspection path is planned to obtain an initial inspection planning result. According to the initial inspection planning result, the initial parking space detection and cleaning operation is completed in the target garage scene, and the inspection-related information is recorded simultaneously. After the initial cleaning is completed, a re-inspection path is planned based on the inspection-related information to obtain a re-inspection planning result. The re-inspection work is carried out according to the re-inspection planning result. During the re-inspection process, the missed parking spaces that were missed during the initial cleaning are detected and cleaned, and the inspection-related information is updated. Then, the re-inspection path planning and re-inspection operation are repeated until the preset inspection stop condition is met. In this way, through the process described in the embodiments of this application, the combination of scene grid map and parking space semantic map realizes accurate digital modeling of garage scene. Without changing the main hardware of the cleaning vehicle, it endows garage with parking space recognition and semantic map capabilities at extremely low adaptation cost. By generating inspection topology path using scene grid map, parking space semantic map and target traffic rules, it can achieve high coverage and high efficiency of parking space cleaning while meeting normal traffic habits and safety constraints, ensuring the rationality and efficiency of inspection route. With the iterative mechanism of initial inspection plus multiple re-inspections, it can effectively reduce the probability of missed parking spaces and improve the effectiveness of parking space cleaning. By continuously optimizing the re-inspection path through dynamic updates of inspection information, it improves the coverage and overall efficiency of garage parking space cleaning, ensures the comprehensiveness and standardization of garage parking space cleaning work, and thus improves the coverage, cleaning efficiency and safety of parking space cleaning in garage scene.

[0081] Accordingly, see Figure 9 As shown in the illustration, this application also provides a parking space cleaning device for a garage scenario, applied to a cleaning robot, comprising:

[0082] The parking space recognition module 11 is used to construct a map of the target garage scene to generate a corresponding scene grid map, and to continuously recognize several parking spaces in the target garage scene during the map construction process, so as to generate a parking space semantic map based on the obtained parking space recognition results.

[0083] The path planning module 12 is used to generate an inspection topology path using the scene grid map, the parking space semantic map and the target traffic rules, and to perform preliminary inspection path planning based on the inspection topology path to obtain the corresponding preliminary inspection planning results.

[0084] Information recording module 13 is used to perform initial parking space detection and cleaning operations in the target garage scene according to the initial inspection planning results, and record inspection-related information during the cleaning process, so that after the initial parking space detection and cleaning operations are completed, a re-inspection path planning is performed based on the inspection-related information to obtain the re-inspection planning results.

[0085] The scene re-inspection module 14 is used to perform a re-inspection in the target garage scene based on the re-inspection planning results, so as to detect and clean the missed parking spaces during the re-inspection process, and update the inspection-related information, so that after the current re-inspection cleaning is completed, it will jump to the step of re-inspection path planning based on the inspection-related information until the preset inspection stop condition is met; the missed parking spaces are the parking spaces that were not cleaned during the initial parking space detection and cleaning operation.

[0086] In some specific embodiments, the parking space cleaning device in the garage scenario may further include:

[0087] The condition judgment unit is used to determine whether the cleaning robot has a parking space recognition function;

[0088] An external parameter calibration unit is used to install a parking space recognition sensor if the cleaning robot does not have a parking space recognition function, and to perform external parameter calibration on the parking space recognition sensor to obtain the corresponding sensor external parameters; the sensor external parameters are the external parameters of the parking space recognition sensor relative to the cleaning robot.

[0089] In some specific embodiments, the parking space recognition module 11 may specifically include:

[0090] The parking space recognition unit is used to continuously collect scene environment data during the movement within the target garage scene, and continuously identify several parking spaces in the target garage scene through the parking space recognition sensor to obtain parking space recognition results including parking space positioning information.

[0091] The data conversion unit is used to convert the scene environment data into corresponding scene raster data, so as to generate a scene raster map of the target garage scene based on the scene raster data;

[0092] A coordinate system transformation unit is used to transform the parking space positioning information from a first coordinate system to a second coordinate system using the sensor extrinsic parameters, thereby obtaining the transformed parking space positioning information; the first coordinate system is the coordinate system of the parking space recognition sensor; the second coordinate system is consistent with the coordinate system of the scene grid map;

[0093] The parking space overlay unit is used to overlay the several parking spaces identified in the parking space recognition result into the scene grid map in the form of semantic tags according to the converted parking space information, so as to obtain a parking space semantic map.

[0094] In some specific embodiments, the path planning module 12 may specifically include:

[0095] The map fusion unit is used to fuse the scene raster map and the parking space semantic map to obtain a fused scene binary image;

[0096] The connected component analysis unit is used to perform morphological processing on the scene binary image and perform connected component analysis on the processed scene binary image to identify several connected components, and determine the connected component with the largest area among the several connected components as the target main channel region.

[0097] The edge detection unit is used to determine the target main path of the target garage scene from the target main passage area through a preset skeleton extraction algorithm, and to perform edge detection on the target main passage area in order to determine the driving path of the cleaning robot based on the detection results.

[0098] The direction determination unit is used to segment the driving path based on the target trunk path to obtain several segmented paths, and to determine the driving direction of the several segmented paths according to the target traffic rules and the orientation relative to the target trunk path.

[0099] The path association unit is used to establish path connection relationships for the several segmented paths according to the path driving direction to generate corresponding segmented connection paths, and to associate the several segmented paths with the several parking spaces according to the parking space semantic map to obtain the inspection topology path; the inspection topology path is a directed graph based on the segmented connection path, the path driving direction, and the several segmented paths.

[0100] In some specific embodiments, the information recording module 13 may specifically include:

[0101] The status determination unit is used to determine the parking status of the target parking space when the cleaning robot inspects the target parking space located on the target segment path.

[0102] A parking space cleaning unit is configured to clean the target parking space if the parking space status indicates that the target parking space is neither cleaned nor occupied; and to skip the target parking space if the parking space status indicates that the target parking space is cleaned, or is neither cleaned nor occupied.

[0103] The status recording unit is used to record the number of inspections of the target segmented path and the cleaning status of the associated target parking space; the inspection-related information includes the number of inspections and the cleaning status.

[0104] In some specific embodiments, the information recording module 13 may specifically include:

[0105] A distance determination unit is used to determine the current position of the cleaning robot and determine the target distance based on the current position and the path travel direction; the target distance is the distance between the starting point of the segmented path and the cleaning robot;

[0106] The path determination unit is used to determine the target re-inspection path from the plurality of segmented paths based on the target distance and the inspection-related information;

[0107] The path planning submodule is used to calculate the path cost of several topological paths using a preset path cost formula, and to perform re-examination path planning based on the path cost to generate a re-examination planning result including the target re-examination path; the several topological paths include the several segmented paths and the segmented connection paths.

[0108] In some specific implementations, the path planning submodule may specifically include:

[0109] The formula calculation unit is used to determine the path length and target inspection related information of several topological paths, and to calculate the preset path cost formula using the path length and the target inspection related information to obtain the path cost of the several topological paths.

[0110] The path planning unit is used to perform a re-examination path planning based on the path cost using a preset shortest path planning algorithm to determine the target arrival route; the target arrival route is the route from the current position to the starting point of the target re-examination path;

[0111] The route splicing unit is used to splice the target arrival route with the target re-inspection path to obtain the corresponding re-inspection planning result.

[0112] Furthermore, embodiments of this application also disclose an electronic device, Figure 10 This is a structural diagram of an electronic device 20 according to an exemplary embodiment. The content of the diagram should not be construed as limiting the scope of this application. Specifically, the electronic device 20 may include: at least one processor 21, at least one memory 22, a power supply 23, a communication interface 24, an input / output interface 25, and a communication bus 26. The memory 22 stores a computer program, which is loaded and executed by the processor 21 to implement the relevant steps in the parking space cleaning method in the garage scenario disclosed in any of the foregoing embodiments. Furthermore, the electronic device 20 in this embodiment may specifically be a computer.

[0113] In this embodiment, the power supply 23 is used to provide operating voltage for each hardware device on the electronic device 20; the communication interface 24 can create a data transmission channel between the electronic device 20 and external devices, and the communication protocol it follows can be any communication protocol applicable to the technical solution of this application, and is not specifically limited here; the input / output interface 25 is used to acquire external input data or output data to the outside world, and its specific interface type can be selected according to specific application needs, and is not specifically limited here.

[0114] In addition, the memory 22, as a carrier for resource storage, can be a read-only memory, random access memory, disk or optical disk, etc. The resources stored thereon can include operating system 221, computer program 222, etc., and the storage method can be temporary storage or permanent storage.

[0115] The operating system 221 is used to manage and control the various hardware devices on the electronic device 20 and the computer program 222, which may be Windows Server, Netware, Unix, Linux, etc. In addition to including a computer program capable of performing the parking space cleaning method in a garage scenario as disclosed in any of the foregoing embodiments, the computer program 222 may further include a computer program capable of performing other specific tasks.

[0116] Furthermore, this application also discloses a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, it implements the aforementioned parking space cleaning method in a garage scenario. Specific steps of this method can be found in the corresponding content disclosed in the foregoing embodiments, and will not be repeated here.

[0117] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to in the method section.

[0118] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0119] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly by hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0120] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0121] The technical solutions provided in this application have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for cleaning parking spaces in a garage setting, characterized in that, Applications in cleaning robots, including: The target garage scene is mapped to generate a corresponding scene grid map, and several parking spaces in the target garage scene are continuously identified during the mapping process to generate a parking space semantic map based on the obtained parking space identification results. The inspection topology path is generated using the scene grid map, the parking space semantic map, and the target traffic rules. The initial inspection path is planned based on the inspection topology path to obtain the corresponding initial inspection planning results. Based on the initial inspection planning results, the initial parking space inspection and cleaning operation is carried out in the target garage scene, and inspection-related information is recorded during the cleaning process. After the initial parking space inspection and cleaning operation is completed, a re-inspection path is planned based on the inspection-related information to obtain the re-inspection planning results. Based on the re-inspection planning results, a re-inspection is performed in the target garage scenario to detect and clean missed parking spaces during the re-inspection process, and to update the inspection-related information so that after completing this re-inspection and cleaning, the process jumps to the step of re-inspection path planning based on the inspection-related information until the preset inspection stop condition is met; the missed parking spaces are those that were not cleaned during the initial parking space detection and cleaning operation.

2. The parking space cleaning method in a garage scenario according to claim 1, characterized in that, Before composing the target garage scene, the following steps are also included: Determine whether the cleaning robot has parking space recognition capabilities; If the cleaning robot does not have a parking space recognition function, a parking space recognition sensor is installed, and the external parameters of the parking space recognition sensor are calibrated to obtain the corresponding sensor external parameters; the sensor external parameters are the external parameters of the parking space recognition sensor relative to the cleaning robot.

3. The parking space cleaning method in a garage scenario according to claim 2, characterized in that, The process of composing a scene of the target garage to generate a corresponding scene grid map, and continuously identifying several parking spaces in the target garage scene during the composing process, to generate a parking space semantic map based on the obtained parking space identification results, includes: During the movement within the target garage scene, scene environment data is continuously collected, and several parking spaces in the target garage scene are continuously identified through the parking space recognition sensor to obtain parking space recognition results including parking space positioning information. The scene environment data is converted into corresponding scene raster data, so as to generate a scene raster map of the target garage scene based on the scene raster data; The parking space positioning information is converted from the first coordinate system to the second coordinate system using the sensor extrinsic parameters to obtain the converted parking space positioning information; the first coordinate system is the coordinate system of the parking space recognition sensor; the second coordinate system is consistent with the coordinate system of the scene grid map. Based on the converted parking space information, the parking spaces identified in the parking space recognition result are superimposed onto the scene grid map in the form of semantic tags to obtain a parking space semantic map.

4. The parking space cleaning method in a garage scenario according to claim 1, characterized in that, The step of generating an inspection topology path using the scene grid map, the parking space semantic map, and the target traffic rules includes: The scene raster map and the parking space semantic map are fused to obtain a fused binary image of the scene; Morphological processing is performed on the binary image of the scene, and connected component analysis is performed on the processed binary image of the scene to identify several connected components. The connected component with the largest area among the several connected components is determined as the target main channel region. Using a preset skeleton extraction algorithm, the target main path of the target garage scene is determined from the target main passage area, and edge detection is performed on the target main passage area to determine the driving path of the cleaning robot based on the detection results. The driving route is segmented based on the target main road to obtain several segmented routes. The driving direction of the several segmented routes is determined according to the target traffic rules and the orientation relative to the target main road. Based on the driving direction of the path, establish path connection relationships for the several segmented paths to generate corresponding segmented connection paths, and associate the several segmented paths with the several parking spaces according to the parking space semantic map to obtain the inspection topology path; the inspection topology path is a directed graph based on the segmented connection paths, the driving direction of the path, and the several segmented paths.

5. The parking space cleaning method in a garage setting according to any one of claims 1 to 4, characterized in that, The process of performing initial parking space inspection and cleaning operations in the target garage scenario based on the initial inspection planning results, and recording inspection-related information during the cleaning process, includes: When the cleaning robot is detected to have inspected a target parking space located on the target segment path, the parking status of the target parking space is determined. If the parking space status indicates that the target parking space is not cleaned and is not occupied, then the target parking space is cleaned; if the parking space status indicates that the target parking space is cleaned, or is not cleaned and is occupied, then the target parking space is skipped. Record the number of inspections of the target segmented path and the cleaning status of the associated target parking spaces; the inspection-related information includes the number of inspections and the cleaning status.

6. The parking space cleaning method in a garage scenario according to claim 4, characterized in that, The re-inspection path planning based on the inspection-related information, to obtain the re-inspection planning result, includes: The current position of the cleaning robot is determined, and the target distance is determined based on the current position and the path travel direction; the target distance is the distance between the starting point of the segmented path and the cleaning robot. The target re-inspection path is determined from the plurality of segmented paths based on the target distance and the inspection-related information; The path cost of several topological paths is calculated using a preset path cost formula, and a re-inspection path is planned based on the path cost to generate a re-inspection planning result including the target re-inspection path; the several topological paths include the several segmented paths and the segmented connection paths.

7. The parking space cleaning method in a garage scenario according to claim 6, characterized in that, The step of calculating the path cost of several topological paths using a preset path cost formula, and performing re-inspection path planning based on the path costs to generate a re-inspection planning result including the target re-inspection path, includes: The path lengths and target inspection-related information of several topological paths are determined, and a preset path cost formula is calculated using the path lengths and target inspection-related information to obtain the path costs of the several topological paths. Based on the path cost, a shortest path planning algorithm is used to plan the re-examination path to determine the target arrival route; the target arrival route is the route from the current position to the starting point of the target re-examination path. The target arrival route is concatenated with the target re-inspection path to obtain the corresponding re-inspection planning result.

8. A parking space cleaning device for a garage setting, characterized in that, Applications in cleaning robots, including: The parking space recognition module is used to construct a map of the target garage scene to generate a corresponding scene grid map, and continuously recognize several parking spaces in the target garage scene during the map construction process to generate a parking space semantic map based on the obtained parking space recognition results. The path planning module is used to generate an inspection topology path using the scene grid map, the parking space semantic map and the target traffic rules, and to perform preliminary inspection path planning based on the inspection topology path to obtain the corresponding preliminary inspection planning results. The information recording module is used to perform initial parking space detection and cleaning operations in the target garage scene according to the initial inspection planning results, and record inspection-related information during the cleaning process, so as to perform re-inspection path planning based on the inspection-related information after the initial parking space detection and cleaning operations are completed, and obtain the re-inspection planning results. The scene re-inspection module is used to perform a re-inspection in the target garage scene based on the re-inspection planning results, so as to detect and clean the missed parking spaces during the re-inspection process, and update the inspection-related information, so that after the current re-inspection cleaning is completed, it will jump to the step of re-inspection path planning based on the inspection-related information until the preset inspection stop condition is met; the missed parking spaces are the parking spaces that were not cleaned during the initial parking space detection and cleaning operation.

9. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor for executing the computer program to implement the parking space cleaning method in a garage setting as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, Used to store a computer program; wherein, when the computer program is executed by a processor, it implements the parking space cleaning method in a garage scenario as described in any one of claims 1 to 7.

Images