CN-121977537-A - Underwater robot positioning method and system in adsorption process

CN121977537ACN 121977537 ACN121977537 ACN 121977537ACN-121977537-A

Abstract

The invention relates to the technical field of robot positioning, in particular to an underwater robot positioning method and system in an adsorption process, which solve the technical problem of low positioning accuracy in the prior art. The method comprises the steps of obtaining continuous multi-frame initial point cloud data collected by an underwater robot in a current operation time period and real-time motion state information of the underwater robot at each moment in the current operation time period, fusing the continuous multi-frame initial point cloud data collected in the current operation time period based on the real-time motion state information to generate current segment point cloud data, determining attachment confidence of each three-dimensional point in the current segment point cloud data based on the current segment point cloud data, distributing corresponding registration weight for each three-dimensional point according to the attachment confidence of each three-dimensional point, carrying out weighted registration on the current segment point cloud data and ship model point cloud data based on the registration weight, and determining pose information of the underwater robot at the ending moment of the current operation time period.

Inventors

CAO YING
CHEN XIAOBO
FENG CUIZHI

Assignees

北京世航华远科技有限公司

Dates

Publication Date: 20260505
Application Date: 20260408

Claims (10)

1. The underwater robot positioning method in the adsorption process is characterized by comprising the following steps of: Acquiring continuous multi-frame initial point cloud data acquired by an underwater robot in a current operation time period and real-time motion state information of the underwater robot at each moment in the current operation time period; Based on real-time motion state information of the underwater robot at each moment in the current operation time period, fusing continuous multi-frame initial point cloud data acquired in the current operation time period to generate current fragment point cloud data; Determining the attachment confidence of each three-dimensional point in the current fragment point cloud data based on the current fragment point cloud data, wherein the attachment confidence is used for representing the possibility that the three-dimensional point belongs to a non-structural attachment on the surface of a ship body; And distributing corresponding registration weights for each three-dimensional point according to the attachment confidence of each three-dimensional point, carrying out weighted registration on the current fragment point cloud data and the ship model point cloud data based on the registration weights, and determining pose information of the underwater robot at the ending moment of the current operation time period.
2. The method of positioning an underwater robot in an adsorption process of claim 1, further comprising: And determining the ending time of the current operation time period based on the real-time motion state information.
3. The underwater robot positioning method in the adsorption process according to claim 2, wherein determining the end time of the current working period based on the real-time motion state information includes: Calculating the motion state change rate of the underwater robot at the corresponding moment according to the real-time motion state information; And when the motion state change rate is larger than a preset change threshold value or the current operation time period of the underwater robot reaches a preset duration, determining the corresponding time as the ending time of the current operation time period.
4. The method for positioning an underwater robot in an adsorption process according to claim 1, wherein based on real-time motion state information of the underwater robot at each time in the current operation time period, fusing continuous multi-frame initial point cloud data acquired in the current operation time period to generate current fragment point cloud data, comprising: And carrying out coordinate alignment and splicing on the continuous multi-frame initial point cloud data based on the real-time motion state information of the underwater robot at each moment in the current operation time period, and generating the current fragment point cloud data.
5. The method of claim 1, wherein determining the attachment confidence of each three-dimensional point in the current segment point cloud data based on the current segment point cloud data comprises: And aiming at each three-dimensional point in the current fragment point cloud data, calculating the normal vector difference of each three-dimensional point in the preset neighborhood of the three-dimensional point, and determining the attachment confidence of the three-dimensional point.
6. The method of positioning an underwater robot in an adsorption process of claim 5, further comprising: For each three-dimensional point in the current fragment point cloud data, carrying out principal component analysis on normal vectors of all three-dimensional points in a preset neighborhood of the three-dimensional point, and determining a plurality of principal components and corresponding contribution rates of the normal vectors of all three-dimensional points in the preset neighborhood of the three-dimensional point; Determining a marking coefficient corresponding to the three-dimensional point according to the plurality of main components and the corresponding contribution rate, wherein the marking coefficient is used for representing the concentration degree of the point cloud geometric features of the preset neighborhood of the three-dimensional point along the main direction; And correcting the attachment confidence of the three-dimensional point based on the marking coefficient.
7. The method for positioning an underwater robot in an adsorption process according to claim 1, wherein said weighting registration of the current segment point cloud data and the hull model point cloud data based on the registration weights, determining pose information of the underwater robot, comprises: acquiring pose information of the underwater robot at the initial moment of the current operation time period and real-time motion state information of the underwater robot at all moments in the current operation time period; Predicting estimated pose information of the underwater robot at the ending moment of the current operation time period according to the pose information of the underwater robot at the initial moment of the current operation time period and the real-time motion state information of the underwater robot at each moment in the current operation time period; intercepting corresponding local area point cloud data from the ship model point cloud data according to the estimated pose; And carrying out weighted registration on the current segment point cloud data and the local area point cloud data based on the registration weight, and determining pose information of the underwater robot at the ending moment of the current operation time period.
8. The method for positioning an underwater robot in an adsorption process according to claim 1, wherein said weighted registration of the current segment point cloud data with the hull model point cloud data based on the registration weights comprises: The registration optimization function is used for solving pose transformation of the underwater robot by minimizing the error of the weighted point cloud distance of each three-dimensional point, and the weighted point cloud distance is obtained by weighting calculation based on the registration weight of the corresponding three-dimensional point; And carrying out weighted registration on the current fragment point cloud data and the ship model point cloud data based on the registration optimization function.
9. The method of positioning an underwater robot in an adsorption process of claim 1, further comprising: controlling the underwater robot to move to an initial attaching area determined based on the operation path planning; acquiring initial local point cloud data acquired by the underwater robot in the initial attachment area; Registering the initial local point cloud data with the ship model point cloud data to determine initial pose information of the underwater robot, wherein the initial pose information is pose information of the underwater robot at the starting moment of a first operation time period.
10. An underwater robot positioning system in an adsorption process, comprising: The system comprises a data acquisition unit, a control unit and a control unit, wherein the data acquisition unit is used for acquiring continuous multi-frame initial point cloud data acquired by an underwater robot in a current operation time period and real-time motion state information of the underwater robot at each moment in the current operation time period; The point cloud fusion unit is used for fusing continuous multi-frame initial point cloud data acquired in the current operation time period based on real-time motion state information of the underwater robot at each moment in the current operation time period to generate current fragment point cloud data; the device comprises a confidence coefficient determining unit, a confidence coefficient determining unit and a processing unit, wherein the confidence coefficient determining unit is used for determining the attachment confidence coefficient of each three-dimensional point in the current fragment point cloud data based on the current fragment point cloud data, wherein the attachment confidence coefficient is used for representing the possibility that the three-dimensional point belongs to a non-structural attachment on the surface of a ship body; The weighted registration unit is used for distributing corresponding registration weights for each three-dimensional point according to the attachment confidence of each three-dimensional point, carrying out weighted registration on the current fragment point cloud data and the ship model point cloud data based on the registration weights, and determining pose information of the underwater robot at the end moment of the current operation time period.

Description

Underwater robot positioning method and system in adsorption process Technical Field The invention relates to the technical field of robot positioning, in particular to an underwater robot positioning method and system in an adsorption process. Background With the rapid development of the ship industry and the ocean engineering, the inspection, cleaning and maintenance operations of the underwater ship body are gradually upgraded to the automatic and intelligent directions. The adsorption type underwater robot can execute fine operation on the surface of the ship body by means of negative pressure adsorption, magnetic attraction and other stable attachment technologies, effectively replaces traditional manual diving operation, avoids safety risks brought by deep sea high-pressure and low-visibility environments, and improves operation efficiency. In the operation process of the adsorption type underwater robot, accurate space positioning is a core premise for guaranteeing operation quality and safety, and the positioning accuracy directly influences the effects of key operation links such as cleaning path planning, detection area coverage and the like. At present, the positioning of an adsorption type underwater robot mainly depends on a point cloud registration technology, three-dimensional point cloud data of the surface of a ship body are acquired through sonar equipment and are matched with a preset ship model point cloud, and then the real-time pose of the robot is determined. However, the surface of the ship body is in the marine environment for a long time, nonstructural attachments such as barnacles and seaweeds are easy to adhere, so that collected point cloud data contain a large amount of noise and pseudo features, the attachment interference and the real features of the ship body are difficult to effectively distinguish by the traditional point cloud registration method, and the problems of local optimal solution or positioning drift are easy to occur. In addition, factors such as disturbance of water flow in the underwater environment, change of attachment angle of the robot and the like can cause unstable motion state of the robot, so that robustness of a traditional positioning method is further reduced. The problems cause the insufficient precision and stability of the existing positioning technology, are difficult to meet the requirement of continuous operation of the robot under the complex ship bottom environment, and restrict the further development of underwater automatic operation. Disclosure of Invention In order to solve the technical problem of low positioning accuracy in the prior art, the invention aims to provide an underwater robot positioning method and an underwater robot positioning system in an adsorption process, and the adopted technical scheme is as follows: the application provides an underwater robot positioning method in an adsorption process, which comprises the following steps: Acquiring continuous multi-frame initial point cloud data acquired by an underwater robot in a current operation time period and real-time motion state information of the underwater robot at each moment in the current operation time period; Based on real-time motion state information of the underwater robot at each moment in the current operation time period, fusing continuous multi-frame initial point cloud data acquired in the current operation time period to generate current fragment point cloud data; Determining the attachment confidence of each three-dimensional point in the current fragment point cloud data based on the current fragment point cloud data, wherein the attachment confidence is used for representing the possibility that the three-dimensional point belongs to unstructured attachments on the surface of the ship body; And distributing corresponding registration weights for each three-dimensional point according to the attachment confidence of each three-dimensional point, carrying out weighted registration on the point cloud data of the current segment and the point cloud data of the ship model based on the registration weights, and determining pose information of the underwater robot at the ending moment of the current operation time period. The application provides an underwater robot positioning system in an adsorption process, which comprises the following components: the data acquisition unit is used for acquiring continuous multi-frame initial point cloud data acquired by the underwater robot in the current operation time period and real-time motion state information of the underwater robot at each moment in the current operation time period; the point cloud fusion unit is used for fusing continuous multi-frame initial point cloud data acquired in the current operation time period based on real-time motion state information of the underwater robot at each moment in the current operation time period to generate current fragment point cloud data; The device comprises a conf