A positioning method for ceiling vision robot

Abstract

The present invention discloses a positioning method for a ceiling vision robot, comprising: the ceiling vision robot acquires a ceiling image in real time during movement; acquires existing landmark distribution information based on the ceiling image; determines whether to establish a new landmark based on the existing landmark distribution information, and if so, establishes a new landmark at the ceiling image acquisition position; if not, does not establish a new landmark at the ceiling image acquisition position; effectively monitors the positioning of existing landmarks based on the existing landmark distribution information, and if the positioning of the existing landmark is effective, uses the position information corresponding to the existing landmark for positioning; if the positioning of the existing landmark is invalid, does not use the position information corresponding to the existing landmark for positioning. The present invention uses a ceiling vision robot to avoid confusion of landmarks in complex environments, establishes new landmarks based on the existing landmark distribution information, achieves uniform establishment of landmarks, and improves the landmark-based positioning accuracy by effectively detecting existing landmarks.

Description

Technical Field

[0001] The present invention relates to the technical field of robot positioning, and in particular to a positioning method for a ceiling vision robot. Background Art

[0002] With the increasing maturity of robotics technology, the means of robot positioning technology have become more diverse. Currently, robot positioning technologies include positioning based on visual sensors, positioning based on lidar, and positioning based on inertial navigation systems. Among them, the robot positioning technology based on visual sensors mainly uses image recognition landmarks for positioning. The landmarks can be environmental landmarks, robot-built landmarks, or manually arranged landmarks. The current visual robot landmark recognition and construction technology has the problem of uneven landmark construction, resulting in low visual robot positioning accuracy and high positioning error rate. At the same time, there is also the problem that environmental features are easily confused, and the visual robot cannot accurately identify landmarks, resulting in positioning errors. Summary of the Invention

[0003] To address the above issues, the present invention provides a positioning method using a ceiling-mounted vision robot. This method employs a ceiling-mounted vision robot to mitigate the effects of complex and confusing environmental features. It determines whether to establish new landmarks based on existing landmark distribution information to achieve uniform landmark placement. Furthermore, it effectively detects existing landmarks to improve landmark-based positioning accuracy. The specific technical solutions of the present invention are as follows:

[0004] A positioning method for a ceiling vision robot specifically includes: the ceiling vision robot acquires a ceiling image in real time during movement; the ceiling vision robot acquires existing landmark distribution information based on ceiling image recognition; the ceiling vision robot determines whether to establish a new landmark based on the existing landmark distribution information; if so, the ceiling vision robot is controlled to establish a new landmark at the ceiling image acquisition position; if not, the ceiling vision robot is controlled not to establish a new landmark at the ceiling image acquisition position; the ceiling vision robot effectively monitors the positioning of existing landmarks based on the existing landmark distribution information; if the positioning of the existing landmark is valid, the ceiling vision robot is positioned using the posture information corresponding to the existing landmark; if the positioning of the existing landmark is invalid, the ceiling vision robot is not positioned using the posture information corresponding to the existing landmark.

[0005] Furthermore, the method for the ceiling vision robot to obtain the distribution information of existing landmarks based on ceiling image recognition specifically includes: the ceiling vision robot identifies whether there are existing landmarks in the ceiling image; if so, calculates the distances between all existing landmarks in the ceiling image and the ceiling image acquisition position as the existing landmark distribution information; if not, calculates the distances between all established existing landmarks and the ceiling image acquisition position, and obtains the lighting information corresponding to all established existing landmarks when they were established, and uses the distances between all established existing landmarks and the ceiling image acquisition position and the lighting information corresponding to all established existing landmarks when they were established as the existing landmark distribution information.

[0006] Furthermore, the method for the ceiling vision robot to identify whether there are existing landmarks in the ceiling image specifically includes: the ceiling vision robot extracts corner point features from the ceiling image; the corner point features are matched with the features of all established existing landmarks respectively; if the features of the existing landmark successfully match the corner point features, it is confirmed that there are existing landmarks in the ceiling image, and the existing landmarks that successfully match the corner point features are recorded as existing landmarks in the ceiling image; if the features of the existing landmark do not successfully match the corner point features, then there are no existing landmarks in the ceiling image.

[0007] Furthermore, the method for the ceiling vision robot to determine whether to establish a new landmark based on the existing landmark distribution information specifically includes: when there are existing landmarks in the ceiling image, determining whether the distances between all existing landmarks in the ceiling image and the ceiling image acquisition position are greater than a first distance threshold based on the acquired existing landmark distribution information; if so, controlling the ceiling vision robot to establish a new landmark at the ceiling image acquisition position; if not, controlling the ceiling vision robot not to establish a new landmark at the ceiling image acquisition position; when there are no existing landmarks in the ceiling image, determining whether all existing landmarks that have been established are greater than a first distance threshold based on the acquired existing landmark distribution information. Whether there is a landmark in the landmarks whose distance from the ceiling image acquisition position is less than the second distance threshold, if not, the ceiling vision robot is controlled to establish a new landmark at the ceiling image acquisition position, if so, it is determined based on the acquired existing landmark distribution information whether the lighting information corresponding to the ceiling image is different from the lighting information corresponding to all existing landmarks in the circular area with the ceiling image acquisition position as the center and the radius as the second distance threshold when they were established, if so, the ceiling vision robot is controlled to establish a new landmark at the ceiling image acquisition position, if not, the ceiling vision robot is controlled not to establish a new landmark at the ceiling image acquisition position.

[0008] Furthermore, the method for the ceiling vision robot to determine whether to establish a new landmark based on existing landmark distribution information also includes: when the ceiling vision robot observes in real time during movement that the duration of the absence of existing landmarks in the ceiling image reaches a first time threshold, the ceiling vision robot establishes a new landmark at the ceiling image acquisition position.

[0009] Furthermore, the method for the ceiling vision robot to establish a new landmark at the ceiling image acquisition position specifically includes: the ceiling vision robot obtains and records landmark information corresponding to the new landmark at the ceiling image acquisition position; wherein, the landmark information includes: the posture information of the ceiling vision robot at the ceiling image acquisition position, the angle information of the ceiling vision robot at the ceiling image acquisition position, the posture information of the landmark relative to the ceiling vision robot, the corresponding lighting information when the ceiling vision robot obtains the ceiling image, using the corner point features in the ceiling image obtained by the ceiling vision robot as the features of the landmark, and the latest observation time of the landmark.

[0010] Furthermore, the method for the ceiling vision robot to locate and effectively monitor existing landmarks based on the existing landmark distribution information specifically includes: when there is an existing landmark in the ceiling image, the ceiling vision robot selects one of the existing landmarks that has not been effectively detected from all the existing landmarks in the ceiling image as the current landmark, and obtains the first positioning posture corresponding to the current landmark from the landmark information corresponding to the current landmark; the ceiling vision robot obtains the current second positioning posture of the ceiling vision robot based on the visual sensor and the inertial sensor; determines whether the first positioning posture and the second positioning posture are the same; if the first positioning posture and the second positioning posture are the same, confirms that the first positioning posture corresponding to the current landmark is valid; if the first positioning posture and the second positioning posture are not the same, calculates the first positioning posture The first posture difference between the first positioning posture and the second positioning posture is determined; whether the first posture difference between the first positioning posture and the second positioning posture is less than a preset difference threshold; if so, it is confirmed that the first positioning posture positioning corresponding to the current landmark is valid; if not, the landmark positioning consistency of the current landmark is calculated based on all recorded observed landmarks, and it is determined whether the landmark positioning consistency of the current landmark meets the landmark positioning consistency requirement; if the landmark positioning consistency of the current landmark meets the landmark positioning consistency requirement, it is confirmed that the first positioning posture positioning corresponding to the current landmark is valid; if the landmark positioning consistency of the current landmark does not meet the landmark positioning consistency requirement, it is confirmed that the first positioning posture positioning corresponding to the current landmark is invalid; repeat the above steps until the validity of the positioning of all existing landmarks in the ceiling image is confirmed.

[0011] Furthermore, the method for calculating the positioning consistency of the current landmark based on all recorded observed landmarks specifically includes: matching the current landmark with all recorded observed landmarks one by one for positioning consistency, counting the number of landmarks in all recorded observed landmarks that are consistent with the positioning of the current landmark, and confirming the positioning consistency of the current landmark based on the number of landmarks in all recorded observed landmarks that are consistent with the positioning of the current landmark.

[0012] Furthermore, the method of matching the current landmark with all recorded observed landmarks one by one and counting the number of landmarks in all recorded observed landmarks that are consistent with the current landmark's positioning specifically includes: Step 1: selecting one of the observed landmarks that has not been consistently matched with the positioning from all recorded observed landmarks as the current consistently matched landmark; Step 2: Obtaining the observation time of the current consistently matched landmark and obtaining the positioning posture corresponding to the current consistently matched landmark; Step 3: Calculating the difference between the first positioning posture corresponding to the current landmark and the second positioning posture corresponding to the current consistently matched landmark; Step 4: Calculating the first position of the visual robot in the time period from the observation time of the current consistently matched landmark to the observation time of the current landmark based on the inertial sensor of the visual robot Shift distance; Step 5: Determine whether the second posture difference is the same as the first displacement distance. If the second posture difference is the same as the first displacement distance, it is confirmed that the current positioning consistent matching landmark is consistent with the current landmark positioning, and the number of landmarks that are consistent with the current landmark positioning in all recorded observed landmarks is increased by 1. If the second posture difference is not the same as the first displacement distance, it is confirmed that the current positioning consistent matching landmark is inconsistent with the current landmark positioning, and the number of landmarks that are consistent with the current landmark positioning in all recorded observed landmarks remains unchanged; Repeat the above steps 1 to 5 until there are no observed landmarks that have not been consistently matched in all recorded observed landmarks, and obtain the number of landmarks that are consistent with the current landmark positioning in all recorded observed landmarks.

[0013] Furthermore, the current landmark positioning consistency refers to the number of landmarks in all the recorded observed landmarks that are consistent with the current landmark positioning reaching a first quantity threshold; wherein, the first quantity threshold is positively correlated with the first posture difference.

[0014] Furthermore, when all recorded observed landmarks are traversed and positioned in sequence and matched, and the number of landmarks in all recorded observed landmarks that are consistent with the current landmark position is zero, the current landmark is recorded as an observed landmark, and all recorded observed landmarks are updated.

[0015] Furthermore, the method of recording the current landmark as an observed landmark specifically includes: obtaining and recording a first positioning posture corresponding to the current landmark based on the landmark information corresponding to the current landmark, and recording the observation time of the current landmark.

[0016] Furthermore, when the observed landmark is consistent with the current landmark position, the observation time of the observed landmark is updated to the observation time of the current landmark.

[0017] Furthermore, when the time length from the observation time of the recorded observed landmark to the current time reaches a first time threshold, the observed landmark is deleted from all recorded observed landmarks, and all recorded observed landmarks are updated.

[0018] The present invention uses a ceiling vision robot to weaken the influence of complex and confusing environmental features, and determines whether to establish new landmarks based on the distribution information of existing landmarks, thereby achieving uniform establishment of landmarks. By effectively detecting existing landmarks from multiple angles, the effectiveness of landmark-based positioning is ensured, and the accuracy of landmark-based positioning is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 Schematic diagram of the flow of the positioning method of the ceiling vision robot according to the first embodiment of the present invention.

[0020] Figure 2 2 is a flow chart of a method for effectively detecting existing road signs according to a seventh embodiment of the present invention. DETAILED DESCRIPTION

[0021] In order to make the purpose, technical solutions and advantages of the present invention clearer, the present invention will be described and illustrated below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described below are only used to explain the present invention and are not intended to limit the present invention. It is also understood that for those of ordinary skill in the art, making some changes in design, manufacturing or production based on the technical content disclosed in the present invention is merely a routine technical means and should not be interpreted as insufficient disclosure of the present application.

[0022] Unless otherwise defined, technical or scientific terms used in the present invention shall have the ordinary meanings understood by those skilled in the art in the art to which this application belongs. The terms "comprise," "include," "have," and any variations thereof used in this application are intended to cover non-exclusive inclusions, such as a process or method comprising a series of steps that is not limited to the listed steps but may also include steps not listed, or may include other steps inherent to the process or method.

[0023] As a preferred embodiment of the present invention, a first embodiment of the present invention provides a positioning method for a ceiling vision robot, such as Figure 1 As shown, the method specifically includes:

[0024] Control the ceiling vision robot to obtain ceiling images in real time during movement; specifically, the ceiling vision robot refers to a robot body equipped with a visual sensor capable of acquiring ceiling images; the visual sensor can be but is not limited to a monocular camera, a binocular camera, a depth camera or a fisheye camera, etc., which have the function of acquiring ceiling images.

[0025] The ceiling vision robot identifies and obtains the distribution information of existing landmarks in the ceiling image based on the acquired ceiling image; specifically, the ceiling vision robot's recognition of the ceiling image refers to identifying whether there are existing landmarks in the ceiling image; the existing landmarks refer to established landmarks; the landmarks refer to identifiers used to provide position reference information for the ceiling vision robot's positioning.

[0026] The ceiling vision robot determines whether to establish a new landmark based on the existing landmark distribution information in the ceiling image. If so, the ceiling vision robot is controlled to establish a new landmark at the ceiling image acquisition position. If not, the ceiling vision robot is controlled not to establish a new landmark at the ceiling image acquisition position. Specifically, in this step, the ceiling vision robot determines whether to establish a new landmark based on the existing landmark distribution information, so that the construction of the new landmark can refer to the existing landmark distribution information, so that the landmarks can be evenly established, thereby improving the rationality of landmark establishment.

[0027] The ceiling vision robot effectively detects existing landmarks based on their distribution information. If the existing landmark positioning is valid, the ceiling vision robot is positioned using the corresponding position information of the existing landmark. Conversely, if the existing landmark positioning is invalid, the ceiling vision robot is not positioned using the corresponding position information of the existing landmark. The positioning method of the ceiling vision robot provided in this embodiment collects ceiling images and analyzes the existing landmark distribution information during movement to determine whether to establish landmarks. This ensures that landmarks are established uniformly and reasonably. Furthermore, by effectively detecting existing landmarks from multiple angles, the effectiveness of landmark-based positioning is ensured, thereby improving the accuracy of landmark-based positioning.

[0028] Based on the first embodiment described above, as a preferred embodiment of the present invention, the method of the ceiling vision robot in the second embodiment of the present invention for recognizing and acquiring the existing landmark distribution information in the ceiling image based on the ceiling image specifically includes:

[0029] Control the ceiling vision robot to identify whether there are existing road signs in the ceiling image; specifically, the method for the ceiling vision robot to identify whether there are existing road signs in the ceiling image can be, but is not limited to, using an algorithm with image feature recognition and matching functions such as a Harris corner detection algorithm, and detecting whether there are existing road signs in the ceiling image through image feature matching.

[0030] If the ceiling vision robot recognizes that there is at least one existing landmark in the ceiling image, the distances between all existing landmarks in the ceiling image and the ceiling image acquisition position are calculated, and the distances between all existing landmarks in the ceiling image and the ceiling image acquisition position are used as the existing landmark distribution information of the ceiling image; specifically, the method for calculating the distances between all existing landmarks in the ceiling image and the ceiling image acquisition position includes: selecting an existing landmark from all existing landmarks in the ceiling image, obtaining a corresponding positioning posture of the existing landmark, obtaining the distance between the existing landmark and the ceiling image acquisition position based on the corresponding positioning posture of the existing landmark and the ceiling image acquisition position, and repeating the above steps until all existing landmarks in the ceiling image are traversed and calculated.

[0031] If the ceiling vision robot recognizes that there are no existing landmarks in the ceiling image, it calculates the distances between all the established existing landmarks and the ceiling image acquisition position, and at the same time obtains the lighting information corresponding to all the established existing landmarks when they were established, and uses the distances between all the established existing landmarks and the ceiling image acquisition position and the lighting information corresponding to all the established existing landmarks when they were established as the existing landmark distribution information of the ceiling image; specifically, the all established existing landmarks refer to all the existing landmarks that have been established by the ceiling vision robot; the method for calculating the distances between all the established existing landmarks and the ceiling image acquisition position includes: selecting an existing landmark from all the established existing landmarks, obtaining the corresponding positioning posture of the existing landmark, obtaining the distance between the existing landmark and the ceiling image acquisition position according to the corresponding positioning posture of the existing landmark and the ceiling image acquisition position, and repeating the above steps until all the established existing landmarks are traversed and calculated.

[0032] This embodiment obtains the distribution information of existing landmarks corresponding to the ceiling image based on the presence of existing landmarks in the ceiling image. When existing landmarks exist in the ceiling image, the establishment of new landmarks only requires reference to the distance between the existing landmarks in the ceiling image and the ceiling vision robot. When no existing landmarks exist in the ceiling image, the establishment of new landmarks requires reference to the distance between all existing landmarks and the ceiling image acquisition position. This avoids the situation where existing landmarks near the ceiling image acquisition position cannot be found in the ceiling image due to viewing angle limitations. By counting the distances between all existing landmarks and the ceiling image acquisition position, the distances between all existing landmarks and the ceiling vision robot are determined. At the same time, the lighting information corresponding to all existing landmarks at the time of establishment is obtained as the existing landmark distribution information. In this method, the lighting information corresponding to the establishment of the landmarks is used as a reference factor for the establishment of new landmarks to avoid the situation where the reliability of overall landmark positioning is affected by the different lighting information of different landmark points during establishment.

[0033] Based on the above embodiment, as a preferred embodiment, the method for the ceiling vision robot to identify whether there is an existing landmark in the ceiling image in the third embodiment of the present invention specifically includes:

[0034] The ceiling vision robot extracts corner features from the ceiling image; specifically, the method of extracting corner features from the ceiling image by the ceiling vision robot may be, but is not limited to, extracting corner features from the ceiling image based on a Harris corner detection algorithm, or extracting corner features from the ceiling image based on an algorithm with corner feature extraction function such as a FAST algorithm and a SIFT algorithm.

[0035] Corner point features are matched against the features of all established landmarks. If the features of an existing landmark successfully match the corner point features, the landmark is confirmed to exist in the ceiling image and recorded as existing in the ceiling image. Conversely, if the features of an existing landmark successfully match the corner point features, the landmark is confirmed to not exist in the ceiling image. Specifically, corner point features are features that remain stable within the same scene even when the image acquisition perspective changes. A successful match between a corner point feature and an existing landmark occurs when the degree of overlap between the angular features extracted from the ceiling image and the features of the existing landmark reaches a preset matching threshold. The preset matching threshold is a value set based on the positioning accuracy requirements of the ceiling vision robot to determine the degree of match between the corner point features and the features of the existing landmark. It should be noted that the features of the existing landmark are acquired and recorded when the landmark is established. This embodiment improves the reliability of the presence of existing landmarks in ceiling images through feature matching.

[0036] Based on the above embodiment, as a preferred embodiment, in a fourth embodiment of the present invention, the method for the ceiling vision robot to determine whether to establish a new landmark based on existing landmark distribution information specifically includes:

[0037] When there are existing landmarks in the ceiling image, it is determined whether the distances between all existing landmarks in the ceiling image and the ceiling image acquisition position are greater than a first distance threshold based on the acquired existing landmark distribution information; specifically, the first distance threshold is a value set based on the ceiling vision robot's requirements for landmark positioning accuracy to make the landmarks more uniform. By determining the distance between the existing landmarks in the ceiling image and the ceiling image acquisition position, supplemented by distance limitations, it is determined whether new landmarks need to be established, so that the ceiling vision robot can have more uniform and reliable landmarks to achieve precise positioning.

[0038] If the distances between all existing landmarks in the ceiling image and the ceiling image acquisition position are greater than the first distance threshold, the ceiling vision robot is controlled to establish new landmarks at the ceiling image acquisition position; specifically, when the distances between all existing landmarks in the ceiling image and the ceiling image acquisition position are greater than the first distance threshold, it means that there are no landmarks in the circular range with the ceiling image acquisition position as the center and the radius of the first distance threshold to provide positioning assistance for the ceiling vision robot. Therefore, new landmarks need to be established to reduce the area not covered by landmarks and improve the reliability of the ceiling vision robot using landmarks for positioning.

[0039] If the distances between all existing landmarks in the ceiling image and the ceiling image acquisition position are not all greater than the first distance threshold, the ceiling vision robot is controlled not to establish a new landmark at the ceiling image acquisition position; specifically, when the distance between at least one existing landmark among all existing landmarks in the ceiling image and the ceiling image acquisition position is less than or equal to the first distance threshold, it means that there are existing landmarks within a circular range with the ceiling image acquisition position as the center and a radius of the first distance threshold to provide positioning assistance for the ceiling vision robot. Therefore, there is no need to establish new landmarks, avoiding the situation where some areas are densely distributed and some areas are less distributed due to uneven landmark establishment.

[0040] When there are no existing landmarks in the ceiling image, it is determined based on the acquired distribution information of the existing landmarks whether there are any landmarks among all the established existing landmarks whose distance from the ceiling image acquisition position is less than a second distance threshold; specifically, the second distance threshold is a value set based on the ceiling vision robot's requirements for landmark positioning accuracy to make the landmark establishment more uniform. It should be noted that the second distance threshold is less than or equal to the first distance threshold.

[0041] If there is no landmark among all the existing landmarks that have been established, the distance from the ceiling image acquisition position to which is less than the second distance threshold is not present, the ceiling vision robot is controlled to establish a new landmark at the ceiling image acquisition position; specifically, when there is no existing landmark in the ceiling image, when there is no landmark among all the existing landmarks that have been established, the distance from the ceiling image acquisition position to which is less than the second distance threshold is not present, it means that there is no existing landmark to provide positioning assistance to the ceiling vision robot within the circular range centered on the ceiling image acquisition position and with the second distance threshold as the radius. In order to make the landmark establishment more uniform and to provide better positioning assistance to the ceiling vision robot, the ceiling vision robot is controlled to establish a new landmark at the ceiling image acquisition position.

[0042] If a landmark exists among all the established landmarks whose distance from the location where the ceiling image is collected is less than a second distance threshold, then, based on the acquired landmark distribution information, it is determined whether the lighting information corresponding to the ceiling image differs from the lighting information corresponding to all the existing landmarks within a circular area centered on the ceiling image and with a radius of the second distance threshold at the time of establishment. If so, the ceiling vision robot is controlled to establish a new landmark at the location where the ceiling image is collected. If not, the ceiling vision robot is controlled not to establish a new landmark at the location where the ceiling image is collected. Specifically, if there is no existing landmark in the ceiling image, and if a landmark exists among all the established landmarks whose distance from the location where the ceiling image is collected is less than the second distance threshold, then the primary reference factor for establishing the new landmark is converted from distance to lighting information. By determining whether the lighting information corresponding to the ceiling image and the lighting information corresponding to the landmark whose distance from the ceiling vision robot is less than the second distance threshold are all the same, if they differ, a new landmark is established to improve the reliability of the landmark, thereby achieving a landmark with different lighting information at the location.

[0043] Based on the above embodiments, as a preferred embodiment of the present invention, the method for a ceiling vision robot to determine whether to establish a new landmark based on existing landmark distribution information in the fifth embodiment of the present invention further includes: when the ceiling vision robot observes in real time during movement that the absence of existing landmarks in the ceiling image reaches a first time threshold, controlling the ceiling vision robot to establish a new landmark at the ceiling image acquisition location. Specifically, the first time threshold is used to limit the maximum duration that the ceiling vision robot cannot observe existing landmarks, and can be, but is not limited to, 3 seconds, 5 seconds, or 10 seconds. In this embodiment, when the absence of existing landmarks in the ceiling image observed in real time during movement reaches the first time threshold, a new landmark is established at the corresponding ceiling image acquisition location when the first time threshold is reached, in order to ensure more uniform landmark establishment and reduce the area of ​​landmark uncovered.

[0044] Based on the above embodiments, as a preferred embodiment of the present invention, the method for establishing a new landmark at a ceiling image acquisition location described in the sixth embodiment of the present invention specifically includes: the ceiling vision robot acquiring and recording landmark information corresponding to the new landmark at the ceiling image acquisition location; wherein the landmark information includes: the position information of the ceiling vision robot at the ceiling image acquisition location, the angle information of the ceiling vision robot at the ceiling image acquisition location, the position information of the landmark relative to the ceiling vision robot, the lighting information corresponding to the ceiling image acquired by the ceiling vision robot, the use of corner features in the ceiling image acquired by the ceiling vision robot as landmark features, and the observation time of the landmark. It should be noted that the observation time of the landmark refers to the capture time of the ceiling image corresponding to the establishment of the landmark. This embodiment establishes a landmark at a specified location by constructing landmark information, and uses the position information contained in the landmark information to achieve the purpose of landmark-assisted positioning of the ceiling vision robot.

[0045] Based on the above embodiments, as a preferred embodiment of the present invention, the method for the ceiling vision robot to effectively detect existing road signs based on existing road sign distribution information is as follows: Figure 2 As shown, specifically including:

[0046] When there are existing landmarks in the ceiling image, the ceiling vision robot selects one of the existing landmarks in the ceiling image that has not been effectively detected as the current landmark, and obtains the first positioning posture corresponding to the current landmark from the landmark information corresponding to the current landmark; it should be noted that each landmark has a one-to-one corresponding positioning posture, and the positioning posture corresponding to each landmark is recorded when it is established.

[0047] The ceiling vision robot obtains the current second positioning posture of the ceiling vision robot based on the visual sensor and the inertial sensor. Specifically, this step calculates and obtains the current second positioning posture of the ceiling vision robot based on a visual inertial system composed of a ceiling-facing visual sensor and an inertial sensor mounted on the ceiling vision robot body.

[0048] Determine whether the first positioning posture and the second positioning posture are the same. If the first positioning posture and the second positioning posture are the same, it is confirmed that the first positioning posture corresponding to the current landmark is valid. It should be noted that if the first positioning posture corresponding to the current landmark is valid, it means that the current landmark can be used as a positioning reference for the visual robot.

[0049] If the first positioning posture and the second positioning posture are different, the first position difference between the first positioning posture and the second positioning posture is calculated, and it is determined whether the first position difference between the first positioning posture and the second positioning posture is less than a preset difference threshold. If the first position difference between the first positioning posture and the second positioning posture is less than the preset difference threshold, it is confirmed that the first positioning posture corresponding to the current landmark is valid;

[0050] If the first positioning posture difference between the first positioning posture and the second positioning posture is greater than or equal to the preset difference threshold, the landmark positioning consistency of the current landmark is calculated based on all the recorded observed landmarks, and it is determined whether the landmark positioning consistency of the current landmark meets the landmark positioning consistency requirement. If the landmark positioning consistency of the current landmark meets the landmark positioning consistency requirement, it is confirmed that the first positioning posture corresponding to the current landmark is valid. If the landmark positioning consistency of the current landmark does not meet the landmark positioning consistency requirement, it is confirmed that the first positioning posture corresponding to the current landmark is invalid. Specifically, the said all recorded observed landmarks refer to each landmark that meets the recording conditions among the landmarks previously observed by the ceiling vision robot. The landmark will be recorded in all recorded observed landmarks. The recording conditions can be but are not limited to the conditions that the landmark has never been observed, etc., which can limit the recorded observed landmarks to be non-repetitive.

[0051] Repeat the above steps until the validity of the positioning of all existing landmarks in the ceiling image has been confirmed. The landmark validity detection method provided in this embodiment considers the landmark consistency of the positioning pose of the current landmark when the observed positioning pose is different from the positioning pose of the ceiling vision robot obtained based on the visual inertial system. The positioning of landmarks that do not meet the landmark consistency requirements is determined to be invalid, thereby avoiding the problem of ceiling vision robot positioning errors due to incorrect landmark recognition, and ensuring that the ceiling vision robot has higher positioning accuracy and better reliability based on landmarks.

[0052] Based on the above embodiment, as a preferred embodiment of the present invention, the method for calculating the current landmark positioning consistency based on all recorded observed landmarks in the eighth embodiment of the present invention specifically includes: matching the current landmark with all recorded observed landmarks one by one, counting the number of landmarks that are consistent with the current landmark positioning in all recorded observed landmarks, and confirming the current landmark positioning consistency based on the number of landmarks that are consistent with the current landmark positioning in all recorded observed landmarks. Specifically, it should be noted that the landmark positioning consistency is positively correlated with the number of landmarks that are consistent with the current landmark positioning in all recorded observed landmarks. The more landmarks that are consistent with the current landmark positioning in all recorded observed landmarks, the higher the landmark positioning consistency of the current landmark. Therefore, in the method provided in the embodiment, the number of landmarks that are consistent with the current landmark positioning in all recorded observed landmarks is calculated, thereby confirming the landmark positioning consistency of the current landmark. The higher the landmark positioning consistency of the current landmark, the higher the reliability of the first positioning posture corresponding to the current landmark. This embodiment calculates the landmark positioning consistency of the current landmark to understand the reliability of the current landmark and to grasp the landmark situation.

[0053] Based on the above embodiments, as a preferred embodiment of the present invention, the method of performing one-to-one positioning and matching of the current landmark with all recorded observed landmarks, and counting the number of landmarks in all recorded observed landmarks that are consistent with the current landmark's positioning, as described in the ninth embodiment of the present invention, specifically includes:

[0054] Step 1: Select one of the observed landmarks that has not been consistently matched with the positioning from all the recorded observed landmarks as the current consistently matched landmark; specifically, select the observed landmark that has not been consistently matched with the positioning from all the recorded observed landmarks as the current consistently matched landmark, that is, select the observed landmark for consistently matching with the positioning, and by judging the observed landmarks that have not been consistently matched with the positioning among all the recorded observed landmarks, realize traversal of consistently matching with the positioning of all the recorded observed landmarks, without missing any observed landmarks, and ensure the accuracy of landmark consistent matching. It should be noted that the said all recorded observed landmarks refer to that every time a landmark that meets the recording conditions is encountered among the landmarks previously observed by the visual robot, the landmark will be recorded in the all recorded observed landmarks; specifically, the said recording conditions can be, but are not limited to, conditions that can limit the recorded observed landmarks to be non-repeated, such as the landmark that has never been observed.

[0055] Step 2: Obtain the observation time of the currently positioned consistent matching landmark and obtain the positioning posture corresponding to the currently positioned consistent matching landmark; specifically, each landmark has a one-to-one corresponding observation time and a one-to-one corresponding positioning posture, and the observation time refers to the time when the landmark is observed. It should be noted that the observation time corresponding to each landmark is variable and not static. When a landmark is re-observed, the observation time corresponding to the landmark is updated to the current time; the positioning posture refers to the positioning posture recorded when the landmark is first observed, and the positioning posture corresponding to each landmark does not change with the number of observations.

[0056] Step 3: Calculate the difference between the first positioning posture corresponding to the current landmark and the second positioning posture corresponding to the current positioning consistent matching landmark; Step 4: Calculate the first displacement distance of the visual robot in the time period from the observation moment of the current positioning consistent matching landmark to the observation moment of the current landmark based on the inertial sensor of the visual robot; Specifically, the inertial sensor calculates the displacement distance of the visual robot in real time based on the code disk and gyroscope during the movement of the visual robot.

[0057] Step 5: Determine whether the second pose difference is the same as the first displacement distance. If the second pose difference is the same as the first displacement distance, it is confirmed that the current positioning consistent matching landmark is consistent with the current landmark positioning, and the number of landmarks in all recorded observed landmarks that are consistent with the current landmark positioning is increased by 1. If the second pose difference is not the same as the first displacement distance, it is confirmed that the current positioning consistent matching landmark is inconsistent with the current landmark positioning, and the number of landmarks in all recorded observed landmarks that are consistent with the current landmark positioning remains unchanged; repeat the above steps 1 to 5 until there are no observed landmarks in all recorded observed landmarks that have not been consistently matched for positioning, and obtain the number of landmarks in all recorded observed landmarks that are consistent with the current landmark positioning. This embodiment sequentially traverses all recorded observed landmarks to determine whether the positioning of the current landmark is consistent with that of the current landmark. Each time an observed landmark is consistent with the positioning of the current landmark, the number of landmarks in all recorded observed landmarks that are consistent with the positioning of the current landmark is increased by 1. Conversely, when no observed landmark is consistent with the positioning of the current landmark, the number of landmarks in all recorded observed landmarks that are consistent with the positioning of the current landmark is kept unchanged, thereby accurately counting the number of landmarks in all recorded observed landmarks that are consistent with the positioning of the current landmark.

[0058] Based on the above embodiment, as a preferred embodiment of the present invention, the current landmark positioning consistency in the tenth embodiment of the present invention refers to the number of landmarks in the total number of observed landmarks that are consistent with the current landmark positioning reaching a first threshold value; wherein, the first threshold value is positively correlated with the first posture difference. Specifically, the first threshold value is a variable and adjustable value used to determine whether the current landmark consistency meets the requirements; the more landmarks in the total number of observed landmarks that are consistent with the current landmark positioning, the higher the positioning consistency of the current landmark; conversely, the fewer landmarks in the total number of observed landmarks that are consistent with the current landmark positioning, the lower the positioning consistency of the current landmark. A high positioning consistency of the current landmark indicates a high positioning accuracy of the current landmark and a high credibility of the first positioning posture corresponding to the current landmark; a low positioning consistency of the current landmark indicates a low credibility of the first positioning posture corresponding to the current landmark and requires further testing of the accuracy of the current landmark.

[0059] Specifically, the positive correlation between the first quantity threshold and the first posture difference means that when the first posture difference between the first positioning posture and the second positioning posture is large, the first quantity threshold is adjusted to be larger accordingly. Conversely, when the first posture difference between the first positioning posture and the second positioning posture is small, the first quantity threshold is adjusted to be smaller accordingly. For example, when the first posture difference between the first positioning posture and the second positioning posture is small, the first quantity threshold is set to 2, then only the landmarks that are consistent with the current landmark positioning in all the recorded observed landmarks are counted. When the number reaches 2 or more, the signpost consistency of the current signpost meets the signpost consistency requirement; when the first position difference between the first positioning posture and the second positioning posture is large, the first quantity threshold is set to 5, then only when the number of signposts that are consistent with the current signpost positioning in all recorded observed signposts reaches 5 or more, the signpost consistency of the current signpost meets the signpost consistency requirement, if the number of signposts that are consistent with the current signpost positioning in all recorded observed signposts does not reach 5, the signpost consistency of the current signpost does not meet the signpost consistency requirement. It should be noted that the first quantity threshold is configured to a numerical value of different sizes according to the size of the first position difference, and the first quantity threshold can be, but is not limited to, an integer greater than 0. In this embodiment, the landmark positioning consistency of the current landmark is linked to the number of landmarks in all recorded observed landmarks that are consistent with the current landmark positioning, so that the landmark positioning consistency of the current landmark can be directly grasped through the number of landmarks in all recorded observed landmarks that are consistent with the current landmark positioning. At the same time, the landmark consistency requirement is controlled to be positively correlated with the first pose difference, so that during the positioning process, it can be determined whether the landmark consistency requirement needs to be increased based on the size of the preliminary pose difference, and the reliability of landmark positioning is improved through multiple flexible judgments.

[0060] Based on the above embodiment, as a preferred embodiment of the present invention, in the eleventh embodiment of the present invention, when all the recorded observed landmarks are traversed in sequence and the position matching judgment is completed, and the number of landmarks in the counted observed landmarks that are consistent with the current landmark position is zero, the current landmark is recorded as an observed landmark, and all the recorded observed landmarks are updated. In this embodiment, after all the recorded observed landmarks are traversed and judged, the current landmark with the number of landmarks consistent with the current landmark position in the counted observed landmarks is zero is recorded as an observed landmark, so as to include unobserved landmarks in the observed landmarks, increase the total number of recorded observed landmarks, and ensure the comprehensiveness and reliability of all recorded observed landmark data.

[0061] Based on the above embodiments, as a preferred embodiment of the present invention, in the twelfth embodiment of the present invention, the method of recording the current landmark as an observed landmark specifically includes: according to the landmark information corresponding to the current landmark, obtaining and recording the first positioning posture corresponding to the current landmark, and recording the observation time of the current landmark. It should be noted that the information of all the recorded observed landmarks used in the present invention is mainly the positioning posture corresponding to the observed landmark and the observation time corresponding to the observed landmark. In actual application, the method of recording the current landmark as an observed landmark may also include recording other current landmark related information. The observation time of the current landmark is changeable and updateable.

[0062] Based on the above embodiments, as a preferred embodiment of the present invention, in the thirteenth embodiment of the present invention, the method of recording the current landmark as an observed landmark specifically includes: according to the landmark information corresponding to the current landmark, obtaining and recording the first positioning posture corresponding to the current landmark, and recording the observation time of the current landmark. It should be noted that the information of all the recorded observed landmarks used in the present invention is mainly the positioning posture corresponding to the observed landmark and the observation time corresponding to the observed landmark. In actual application, the method of recording the current landmark as an observed landmark may also include recording other current landmark related information. It should be noted that the observation time of the current landmark is changeable and updateable.

[0063] Based on the above embodiments, as a preferred embodiment of the present invention, in the fourteenth embodiment of the present invention, when the observed landmark is consistent with the current landmark's positioning, the observation time of the observed landmark is updated to the observation time of the current landmark. Since the current landmark is consistent with the positioning of the observed landmark, it means that the visual robot has observed the observed landmark again at the current moment, so the observation time of the current landmark is updated to the observation time corresponding to the observed landmark, indicating that the observed landmark has been observed again. This embodiment updates the observation time corresponding to the observed landmark accordingly based on the result of the consistent positioning judgment of the observed landmark and the current landmark, so that the observation time corresponding to all recorded observed landmarks remains the most recent observation time.

[0064] Based on the above embodiments, as a preferred embodiment of the present invention, in the fifteenth embodiment of the present invention, when the time length from the observation moment of the recorded observed landmark to the current moment reaches a first time threshold, the observed landmark is deleted from all recorded observed landmarks, and all recorded observed landmarks are updated. Specifically, the first time threshold is used to limit the maximum length of time that the observed landmark needs to be observed again, and the first time threshold is set based on a comprehensive consideration of multiple factors such as the actual application scenario of the ceiling vision robot and the moving speed of the ceiling vision robot. When the length of time that the observed landmark has not been observed again exceeds the maximum length of time, the observed landmark is considered invalid and needs to be deleted from all recorded observed landmarks, thereby reducing the computational burden of the visual robot algorithm and improving the positioning efficiency of the ceiling vision robot.

[0065] Obviously, the above-mentioned embodiments are only some embodiments of the present invention, rather than all embodiments, and the technical solutions between the various embodiments can be combined with each other. In the above-mentioned embodiments of the present invention, the descriptions of the various embodiments have different focuses. For the parts that are not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments. In the several embodiments provided by the present invention, it should be understood that the disclosed technical contents can be implemented in other ways. It should also be noted that the various specific technical features described in the above-mentioned specific embodiments can be combined in any suitable way unless there is any contradiction. In order to avoid unnecessary repetition, the various possible combinations will not be described separately in the embodiments of the present invention.

[0066] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A positioning method for a ceiling vision robot, characterized in that: The positioning method of the ceiling vision robot specifically includes: The ceiling vision robot acquires ceiling images in real time while moving; The ceiling vision robot obtains the existing landmark distribution information based on ceiling image recognition; The ceiling vision robot determines whether to establish a new landmark based on the existing landmark distribution information. If so, the ceiling vision robot is controlled to establish a new landmark at the ceiling image acquisition position. If not, the ceiling vision robot is controlled not to establish a new landmark at the ceiling image acquisition position. The ceiling vision robot locates and effectively monitors the existing landmarks based on the existing landmark distribution information. If the existing landmark positioning is valid, the ceiling vision robot is positioned using the posture information corresponding to the existing landmark. If the existing landmark positioning is invalid, the ceiling vision robot is not positioned using the posture information corresponding to the existing landmark.

2. The positioning method of the ceiling vision robot according to claim 1, characterized in that: The method for the ceiling vision robot to obtain existing road sign distribution information based on ceiling image recognition specifically includes: The ceiling vision robot identifies whether there are existing landmarks in the ceiling image; If so, calculate the distance between all existing landmarks in the ceiling image and the location where the ceiling image is collected as the existing landmark distribution information; If it does not exist, calculate the distances between all the existing landmarks and the ceiling image acquisition position, and obtain the lighting information corresponding to all the existing landmarks when they were established. The distances between all the existing landmarks and the ceiling image acquisition position and the lighting information corresponding to all the existing landmarks when they were established are used as the existing landmark distribution information.

3. The positioning method of the ceiling vision robot according to claim 2, characterized in that: The method for the ceiling vision robot to identify whether there is an existing road sign in the ceiling image specifically includes: The ceiling vision robot extracts corner features from the ceiling image; The corner point features are matched with the features of all established existing landmarks respectively. If the features of an existing landmark successfully match the corner point features, it is confirmed that there are existing landmarks in the ceiling image, and the existing landmarks that successfully match the corner point features are recorded as existing landmarks in the ceiling image. If the features of no existing landmark successfully match the corner point features, then there are no existing landmarks in the ceiling image.

4. The positioning method of the ceiling vision robot according to claim 3, characterized in that: The method of the ceiling vision robot determining whether to establish a new landmark based on existing landmark distribution information specifically includes: When there are existing landmarks in the ceiling image, it is determined whether the distances between all existing landmarks in the ceiling image and the ceiling image acquisition position are greater than a first distance threshold based on the acquired existing landmark distribution information. If so, the ceiling vision robot is controlled to establish a new landmark at the ceiling image acquisition position. If not, the ceiling vision robot is controlled not to establish a new landmark at the ceiling image acquisition position. When there is no existing landmark in the ceiling image, it is determined based on the acquired existing landmark distribution information whether there is a landmark among all the established existing landmarks whose distance from the ceiling image acquisition position is less than the second distance threshold. If not, the ceiling vision robot is controlled to establish a new landmark at the ceiling image acquisition position. If so, it is determined based on the acquired existing landmark distribution information whether the lighting information corresponding to the ceiling image is different from the lighting information corresponding to all the existing landmarks in the circular area with the ceiling image acquisition position as the center and the radius as the second distance threshold when they were established. If so, the ceiling vision robot is controlled to establish a new landmark at the ceiling image acquisition position. If not, the ceiling vision robot is controlled not to establish a new landmark at the ceiling image acquisition position.

5. The positioning method of the ceiling vision robot according to claim 4, characterized in that: The method for the ceiling vision robot to determine whether to establish a new landmark based on existing landmark distribution information also includes: when the ceiling vision robot observes in real time during movement that the duration of the absence of existing landmarks in the ceiling image reaches a first time threshold, the ceiling vision robot establishes a new landmark at the ceiling image collection position.

6. The positioning method of the ceiling vision robot according to claim 5, characterized in that: The method for the ceiling vision robot to establish a new landmark at a ceiling image acquisition position specifically includes: the ceiling vision robot acquires and records landmark information corresponding to the new landmark at the ceiling image acquisition position; wherein the landmark information includes: the posture information of the ceiling vision robot at the ceiling image acquisition position, the angle information of the ceiling vision robot at the ceiling image acquisition position, the posture information of the landmark relative to the ceiling vision robot, the corresponding lighting information when the ceiling vision robot acquires the ceiling image, using the corner point features in the ceiling image acquired by the ceiling vision robot as the features of the landmark, and the latest observation time of the landmark.

7. The positioning method of the ceiling vision robot according to claim 6, characterized in that: The method for the ceiling vision robot to locate and effectively monitor existing road signs based on existing road sign distribution information specifically includes: When there is an existing landmark in the ceiling image, the ceiling vision robot selects one existing landmark that has not been effectively detected from all existing landmarks in the ceiling image as the current landmark, and obtains the first positioning pose corresponding to the current landmark from the landmark information corresponding to the current landmark; The ceiling vision robot obtains the current second positioning posture of the ceiling vision robot based on the visual sensor and the inertial sensor; Determine whether the first positioning posture and the second positioning posture are the same; If the first positioning posture and the second positioning posture are the same, it is confirmed that the first positioning posture corresponding to the current landmark is valid; If the first positioning posture and the second positioning posture are different, calculating the first positioning posture difference between the first positioning posture and the second positioning posture; Determine whether a first position difference between the first positioning position and the second positioning position is less than a preset difference threshold; If so, it is confirmed that the first positioning posture positioning corresponding to the current landmark is valid; If not, the landmark positioning consistency of the current landmark is calculated based on all recorded observed landmarks, and it is determined whether the landmark positioning consistency of the current landmark meets the landmark positioning consistency requirement. If the landmark positioning consistency of the current landmark meets the landmark positioning consistency requirement, it is confirmed that the first positioning posture positioning corresponding to the current landmark is valid. If the landmark positioning consistency of the current landmark does not meet the landmark positioning consistency requirement, it is confirmed that the first positioning posture positioning corresponding to the current landmark is invalid. Repeat the above steps until the validity of the positioning of all existing landmarks in the ceiling image is confirmed.

8. The positioning method of the ceiling vision robot according to claim 7, characterized in that: The method for calculating the current landmark positioning consistency based on all recorded observed landmarks specifically includes: matching the current landmark with all recorded observed landmarks one by one for positioning consistency, counting the number of landmarks in all recorded observed landmarks that are consistent with the current landmark positioning, and confirming the current landmark positioning consistency based on the number of landmarks in all recorded observed landmarks that are consistent with the current landmark positioning.

9. The positioning method of the ceiling vision robot according to claim 8, characterized in that: The method of performing one-to-one positioning matching of the current landmark with all recorded observed landmarks and counting the number of landmarks in all recorded observed landmarks that are consistent with the current landmark's positioning specifically includes: Step 1: Select one of the observed landmarks that has not been consistently matched from all recorded landmarks as the current consistently matched landmark; Step 2: Obtain the observation time of the current positioning consistent matching landmark and obtain the positioning pose corresponding to the current positioning consistent matching landmark; Step 3: Calculate the difference between the first positioning pose corresponding to the current landmark and the second positioning pose corresponding to the current landmark that matches the landmark; Step 4: Calculate the first displacement distance of the visual robot in the time period from the observation time of the current positioning consistent matching landmark to the observation time of the current landmark based on the inertial sensor of the visual robot; Step 5: Determine whether the second pose difference is the same as the first displacement distance. If the second pose difference is the same as the first displacement distance, it is determined that the current positioning is consistent with the matching landmark and the current landmark is positioned consistently. The number of landmarks in all recorded observed landmarks that are consistent with the current landmark positioning is increased by 1. If the second pose difference is different from the first displacement distance, it is determined that the current positioning is inconsistent with the matching landmark and the current landmark positioning is inconsistent. The number of landmarks in all recorded observed landmarks that are consistent with the current landmark positioning remains unchanged. Repeat steps 1 to 5 above until there are no observed landmarks in all recorded observed landmarks that are not consistently matched, and obtain the number of landmarks in all recorded observed landmarks that are consistently matched with the current landmark position.

10. The positioning method of the ceiling vision robot according to claim 9, characterized in that: The current landmark positioning consistency refers to the number of landmarks in all the recorded observed landmarks that are consistent with the current landmark positioning reaching a first quantity threshold; wherein the first quantity threshold is positively correlated with the first posture difference.

11. The positioning method of the ceiling vision robot according to claim 9, characterized in that: When all recorded observed landmarks are traversed and positioned in sequence and matched, and the number of landmarks in all recorded observed landmarks that are consistent with the current landmark position is zero, the current landmark is recorded as an observed landmark and all recorded observed landmarks are updated.

12. The positioning method of the ceiling vision robot according to claim 11, characterized in that: The method for recording the current landmark as an observed landmark specifically includes: obtaining and recording the first positioning posture corresponding to the current landmark based on the landmark information corresponding to the current landmark, and recording the observation time of the current landmark.

13. The positioning method of the ceiling vision robot according to claim 12, characterized in that: When the observed landmark is consistent with the current landmark position, the observation time of the observed landmark is updated to the observation time of the current landmark.

14. The positioning method of the ceiling vision robot according to claim 12, characterized in that: When the time length between the observation time of the recorded observed landmark and the current time reaches the first time threshold, the observed landmark is deleted from all recorded observed landmarks, and all recorded observed landmarks are updated.

Images