CN-121982361-A - Rail contact net fault positioning device integrating humanoid robot and multisource sensor

CN121982361ACN 121982361 ACN121982361 ACN 121982361ACN-121982361-A

Abstract

The invention relates to the technical field of robots, in particular to a track contact network fault positioning device integrating a humanoid robot and a multisource sensor. According to the invention, autonomous obstacle surmounting and crossing behavior information recording under a structure shielding environment are realized through gait instruction combination, an image acquisition link introduces a gesture guiding alignment control and angle segmentation capturing strategy, clear extraction of a structure outline boundary and image number output are completed, a position mapping relation is established between an image line segment direction and a track coordinate system, a multi-view path segment is dynamically constructed, overlapping verification is carried out on an image edge fracture region and thermal imaging space point data, a space position coordinate of an abnormal component is extracted, a component identification name and a path track file are aligned in information, an execution action index is formed according to time sequence, and structural identification integrity, path matching accuracy and fault part response efficiency are improved.

Inventors

HUANG CHUN
YIN LEI
XIA KE
FAN GUOPENG

Assignees

诚芯智联(武汉)科技技术有限公司

Dates

Publication Date: 20260505
Application Date: 20251215

Claims (10)

1. Track contact net fault locating device of fusion humanoid robot and multisource sensor, its characterized in that, the system includes: the obstacle crossing execution module reads the front distance and the travelling interruption information of the humanoid robot, judges whether a path is blocked, identifies a blocking object by combining an image acquired by a visual sensor, controls the left leg to lift and the right leg to move forwards, executes a crossing action, records the coordinates of a falling point and the positions of soles, and generates a crossing record table; the component recognition module starts forearm shooting according to the crossing record table, adjusts hand gestures to aim at a contact net, collects images in batches, extracts edges and contours, analyzes consistency of line groups, marks components and outputs image numbers, and a shooting image group is generated; The positioning coupling module invokes a head visual angle sequence based on the shooting image group, extracts the image edge direction, maps the image edge direction to a track angle, combines the image displacement and the coordinate deviation, and divides the image displacement and the coordinate deviation into a forward visual angle and a lateral visual angle to generate a visual angle positioning segment; the anomaly tracking module analyzes the visual angle positioning segment, screens edge fracture line segments, positions a thermal center point, performs coordinate comparison, marks a superposition component and outputs an image breakpoint list; And the target verification module extracts a component number and a name, binds an operation number and a path description based on the image breakpoint list, aligns track points and sorts the track points to form track contact net fault positioning information.
2. The system for positioning the rail contact net fault of the fusion humanoid robot and the multisource sensor according to claim 1, wherein the crossing record list comprises a left leg lifting amplitude, a right leg advancing angle, sole falling point coordinates and crossing behavior numbers, the shooting image group comprises an image file number, an image capturing angle, an edge line group and component profile data, the view angle positioning segment comprises an image edge line direction, a rail direction angle mapping value, a component displacement length and an image view angle type, the image breakpoint list comprises a broken line segment number, a broken coordinate point, a thermal image center coordinate and a component positioning label, and the rail contact net fault positioning information comprises a component name number, a component position description, an action track index and a time sequence list.
3. The system for positioning a rail catenary fault of a fusion robot and a multisource sensor according to claim 1, wherein the consistency of the wire sets refers to a state that wires forming the catenary are spatially continuous, the curvature of the center line of the wires is continuous, and the deviation is smaller than a preset threshold value and an interruption.
4. The rail contact net fault locating device system fusing the humanoid robot and the multisource sensor according to claim 1, wherein the edge breaking line segments refer to discontinuous and broken edge segments of component edges in images due to shielding, damage and image quality blurring.
5. The rail catenary fault locating device system of a fusion humanoid robot with a multisource sensor of claim 1, wherein the obstacle crossing execution module includes: The path shielding judging submodule compares a preset path passing threshold value based on the distance information in front of the humanoid robot and the travelling direction interruption information to judge whether a forward channel is blocked or not, and generates a passing shielding judging result; the structure shielding detection submodule invokes the passing shielding judgment result, compares the superposition proportion of the contact area and the structure shielding area in the visual angle, judges the shielding degree and obtains a structure shielding state identification result; And the gait instruction execution sub-module combines the lifting of the left leg and the forward movement of the right leg to be used as instructions according to the structure shielding state recognition result, executes the crossing action, acquires the sole position at the falling point, updates the coordinate data and generates a crossing record table.
6. The rail catenary fault locating device system of a fusion humanoid robot with a multisource sensor of claim 1, wherein the component identification module includes: the shooting structure control submodule detects the boundary value of the current coordinate and the triggering area after approaching the position of the contact net based on the crossing record table, judges whether the contact net enters a shooting triggering area, starts a forearm shooting structure if the condition is met, and controls the gesture of the hand to align to the direction of the lead to adjust the angle range so as to generate a gesture control image sequence set; the image feature extraction submodule calls the attitude control image sequence set, performs edge detection and continuity judgment operation on the image, extracts a wire edge line group and a component closed contour in the image, performs component identification processing according to edge connection rules, and acquires a component feature identification mark set; And the image number output sub-module extracts marked image contents according to the component characteristic identification mark set, distributes image numbers according to the acquisition sequence of images in a sequence, records the information corresponding to the image file names and the numbers, and establishes a shooting image group.
7. The rail catenary fault locating device system of a fusion humanoid robot with a multisource sensor of claim 1, wherein the locating coupling module comprises: The direction matching extraction submodule extracts edge characteristic lines in the image and analyzes the linear direction of the outline area based on the shooting direction sequence recorded by the shooting image group and the head visual angle control, performs angle alignment matching on the line segment direction of the image and the direction sequence in the advancing gesture sequence, and generates an image direction matching angle set; The offset mapping construction submodule calls the image direction matching angle set, extracts the displacement length of a component in an image space, calculates the corresponding relation of the component in the track direction according to the angle offset value between the central position of the component and the track coordinate, and establishes an image track position mapping matrix; And the visual angle classification output sub-module is used for classifying and aggregating the image segment areas of the identification components distributed in the forward visual angle and the lateral visual angle according to the angle range according to the image track position mapping matrix, extracting the image index numbers conforming to the classification rules and obtaining the visual angle positioning segments.
8. The rail catenary fault locating device system of a fusion humanoid robot with a multisource sensor of claim 1, wherein the anomaly tracking module comprises: The edge interruption extraction submodule extracts continuous edge line segment sequences in the images based on each group of images in the visual angle positioning fragments, screens edge areas with pixel interruption, records line segment numbers and coordinate information of breakage occurrence, and generates an image breakage edge coordinate set; The thermal induction coordinate marking submodule calls the image fracture edge coordinate set to obtain the center point coordinate of a corresponding region in thermal induction data, and space coordinate conversion is carried out on the fracture coordinate position and the thermal image point position to obtain a fracture pair Ji Re induction coordinate set; and the component coincidence identification submodule calculates coincidence matching rate between the coordinates of the broken line segments of the image and the coordinates of the heat sensation according to the broken pair Ji Re, screens the layer positions of which the coincidence matching rate exceeds a broken verification threshold, extracts the image numbers of the matched components and the component area indexes, and establishes an image breakpoint list.
9. The rail catenary fault locating device system of a fusion humanoid robot with a multisource sensor of claim 1, wherein the target verification module includes: the component mark matching submodule reads a number field corresponding to the component outline based on the image breakpoint list, extracts a name entry from the image identification name list, performs field comparison operation through the image number and the component outline marking index, and generates a component number mapping table; the numbering track binding submodule invokes the component numbering mapping table, extracts component position description information contained in the numbering field, correspondingly compares the path point position labels in the robot action track path, and fuses the component name field with the path point sequence to obtain a component path point binding list; And the action index construction sub-module sorts the fused name entries and the path sequence according to the component path point binding list and the timestamp field, constructs the relation between component identification information and operation action index based on the time sequence, and acquires the track catenary fault positioning information.
10. The system for positioning a rail catenary fault of a fusion robot and a multisource sensor according to claim 8, wherein the fracture verification threshold refers to a coincidence matching ratio which is required to be achieved by a valid coincidence member when fracture edges in an image are spatially matched with thermal coordinates.

Description

Rail contact net fault positioning device integrating humanoid robot and multisource sensor Technical Field The invention relates to the technical field of robots, in particular to a track contact net fault positioning device integrating a humanoid robot and a multisource sensor. Background The technical field of robots relates to an automatic device for simulating, replacing or enhancing human operation capability, which mainly comprises various fixed robots, movable robots, service robots, special operation robots and the like, and covers a plurality of core links such as structural design, motion control, environment perception, man-machine interaction, intelligent decision and the like. The robot technology is widely applied to the fields of manufacturing, medical treatment, service, traffic, military and the like, a complex system with high degree of freedom operation capability, multi-scene adaptation capability and intelligent perception decision capability is gradually developed, and particularly a robot based on a humanoid structure has remarkable advantages in the aspects of humanoid operation, complex terrain walking, collaborative operation and the like, and is becoming one of the key directions of robot research. The traditional robot system for track contact net fault location is characterized in that a mobile robot, a detection vehicle or an unmanned aerial vehicle and the like are used as an operation platform, data are collected by means of single or limited type sensors such as visible light shooting and infrared thermal imaging, the contact net line and part states in a track power supply system are identified and inspected, a wheel type or crawler type chassis is usually adopted to move along a track or a fixed route, and images or thermal information are collected and then uploaded to a background for manual or semiautomatic analysis. Such systems often rely on track structures or airlines, are difficult to flexibly deploy in complex, obstacle-dense environments, and are limited by perceived dimensions, platform operability, and job interactivity. The existing track contact net fault positioning technology relies on a mechanical chassis to run along a set path, the path rigidity is high, dynamic response capability is lacking in the face of unstructured barriers, sensing data acquisition is limited to a single sensor type, image content and thermal imaging results lack of space fusion foundation, fault part judgment is dependent on excessive manual intervention, angle fixation and range are narrow in the image acquisition process, viewing angle adjustment and region positioning linkage mechanisms are lacked, component contour extraction precision is insufficient, image break points are difficult to track to real space coordinate positions, identification data are not effectively bound with an operation path, executing action lacks dynamic matching support based on image identification results, and the whole fault identification chain has the problems of weak sensing coverage, image positioning separation, information linkage breakage and the like. Disclosure of Invention In order to solve the technical problems in the prior art, the embodiment of the invention provides a track contact net fault positioning device integrating a humanoid robot and a multisource sensor. In one aspect, a track catenary fault locating device that fuses a humanoid robot and a multisource sensor is provided, the system comprising: the obstacle crossing execution module reads the front distance and the travelling interruption information of the humanoid robot, judges whether a path is blocked, identifies a blocking object by combining an image acquired by a visual sensor, controls the left leg to lift and the right leg to move forwards, executes a crossing action, records the coordinates of a falling point and the positions of soles, and generates a crossing record table; the component recognition module starts forearm shooting according to the crossing record table, adjusts hand gestures to aim at a contact net, collects images in batches, extracts edges and contours, analyzes consistency of line groups, marks components and outputs image numbers, and a shooting image group is generated; The positioning coupling module invokes a head visual angle sequence based on the shooting image group, extracts the image edge direction, maps the image edge direction to a track angle, combines the image displacement and the coordinate deviation, and divides the image displacement and the coordinate deviation into a forward visual angle and a lateral visual angle to generate a visual angle positioning segment; the anomaly tracking module analyzes the visual angle positioning segment, screens edge fracture line segments, positions a thermal center point, performs coordinate comparison, marks a superposition component and outputs an image breakpoint list; And the target verification module extracts a component number and a nam