CN117132882A - Four-wheel track robot spike identification and positioning method based on deep learning depth camera - Google Patents
Four-wheel track robot spike identification and positioning method based on deep learning depth camera Download PDFInfo
- Publication number
- CN117132882A CN117132882A CN202310332159.2A CN202310332159A CN117132882A CN 117132882 A CN117132882 A CN 117132882A CN 202310332159 A CN202310332159 A CN 202310332159A CN 117132882 A CN117132882 A CN 117132882A
- Authority
- CN
- China
- Prior art keywords
- spike
- wheel track
- track robot
- point
- depth camera
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 29
- 238000013135 deep learning Methods 0.000 title claims abstract description 24
- 238000001514 detection method Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000013473 artificial intelligence Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000007689 inspection Methods 0.000 description 2
- 238000012423 maintenance Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/176—Urban or other man-made structures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/20—Ensemble learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/07—Target detection
Abstract
The application provides a spike identification positioning method based on a deep learning depth camera. At a speed V 0 Driving four-wheel track robot to reach X 0 The point is that the spike enters the shooting area of the depth camera, the spike is identified and primarily positioned, and the spike is positioned at a slightly slower speed V according to the primarily positioned information 1 Driving four-wheel track robot to reach X near working point 1 And (3) point, identifying the spike again and accurately positioning, driving the four-wheel track robot to reach the working point X at a slower speed V according to the accurate positioning information, and then starting the operation. Four-wheel rail combined by two-step positioning method of preliminary positioning and accurate positioningThe way of controlling the speed of the track robot provides accurate position information of the track spike for the four-wheel track robot, and improves the operation precision of the four-wheel track robot.
Description
Technical Field
The application relates to a spike identification and positioning method.
Background
At present, the railway development of China is already in the front of the world, brings convenience to us, and has a plurality of hidden dangers. In the case of a rail, because the train is operated frequently, a huge load is brought to the rail, so that the railway spike deforms, damages, falls off and the like, and great potential safety hazards are caused. Therefore, daily maintenance work of the rail is particularly important. In the past, the traditional mode is adopted to carry out inspection one by workers, so that huge social resources are consumed, and huge hidden danger exists in life safety of the inspection workers. With the development of artificial intelligence and machine vision technologies, the application range is wider, and the research on rail spike detection, identification and positioning methods is particularly important.
Disclosure of Invention
When the four-wheel track robot works on a rail, the spike needs to be accurately identified and positioned to determine a working point, so that more accurate spike position information is provided for the four-wheel track robot, the working efficiency of the track robot is improved, and the maintenance cost is reduced.
In order to solve the problems, the application is realized by the following technical scheme: a four-wheel track robot spike identification positioning method based on a deep learning depth camera comprises the following steps:
step one: four-wheel track robot moves on the rail at a speed V 0 Moving from the initial point O to the working point X.
Step two: the surrounding environment of the rail is captured by a depth camera and input into a computer, and then the computer performs spike identification according to a deep learning algorithm from the content captured by the camera.
Step three: when the four-wheel track robot arrives at X 0 When the point is reached, the spike appears in the field of view of the depth camera, the spike is identified and initially positioned by the deep learning algorithm, the collected spike position information is fed back to the computer, and after the computer receives the initial positioning information, the computer sends out a driving instruction to make the spike at a slightly slower speed V 1 Driving the four-wheel track robot to continuously move towards the working point X; x is X 0 Point located between the initial point O and the operation point X.
Step four: four-wheel track robot moves to X 1 When the track is in a point, the depth camera accurately identifies and positions the track spike, and feeds back accurate positioning information of the track spike to the computer, and after receiving the accurate positioning information, the computer sends out a driving instruction again to drive the four-wheel track robot to slowly move to an operation point X at a slower speed V; x is X 1 The point is at X 0 Between the point and the working point X.
Step five: after the four-wheel track robot reaches the working point X, determining the position of the spike according to a deep learning algorithm, reducing the speed of the four-wheel track robot to 0, and then starting the operation.
Step six: after the four-wheel track robot finishes the operation. Repeating the first step to the fifth step, and carrying out the next group of spike operation.
Wherein V is 0 >V 1 >V>0。
In the third step and the fourth step, the four-wheel track robot moves forwards, the depth camera starts to capture the surrounding environment of the rail and identify and detect rail spikes, and when the camera captures the rail spikes for the first time, the rail spikes are identified, detected and positioned. Then the data is fed back to the computer, the computer sends out a driving instruction to drive the four-wheel track robot to move towards the working point X, and the speed of the four-wheel track robot is changed from V in the process 0 Decelerating to V 1 Then from V 1 And the speed is reduced to V, the final speed is reduced to 0, and the primary identification and the accurate identification of the spike are combined, so that the identification precision of the spike is improved.
And thirdly, identifying and detecting the target by using a deep learning algorithm, and determining the distance between the camera and the object by using a depth image shot by the depth camera.
Compared with the prior art, the four-wheel track robot spike identification and positioning method based on the deep learning depth camera has the following beneficial effects: according to the application, the spike is identified and detected by adopting a deep learning algorithm, the spike is positioned by adopting a depth camera, and accurate position information of the spike is provided for the four-wheel track robot by combining a two-step positioning method of preliminary positioning and accurate positioning with a speed control mode of the four-wheel track robot, so that the operation precision of the four-wheel track robot is improved.
Drawings
Fig. 1 is a schematic diagram of the object recognition and positioning operation of the depth camera according to the present application.
In fig. 1, the depth camera is attached to a four-wheeled orbital robot, which is at a speed V 0 Moving from an initial point to a working point X, and reaching the point X by a four-wheel track robot 0 At the time, the speed is reduced to V 1 . Continue moving and reach point X 1 At this point, the velocity drops to V. When the four-wheel track robot moves to the working point X, the speed is reduced to 0, and the work is started.
Description of the embodiments
The following description of the embodiments of the present application will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present application, but not all embodiments. The following description of at least one exemplary embodiment is merely exemplary in nature and is in no way intended to limit the application, its application, or uses. All other embodiments, which can be made by one of ordinary skill in the art without undue burden on the person of ordinary skill in the art based on embodiments of the present application, are within the scope of the present application.
Techniques, methods, and apparatus known to those of ordinary skill in the relevant art may not be discussed in detail, but are intended to be part of the specification where appropriate.
The embodiment of the application provides a four-wheel track robot spike identification and positioning method based on a deep learning depth camera, which comprises the following steps:
step one: four-wheel track robot moves on the rail at a speed V 0 Moving from the initial point O to the operation point X;
step two: capturing the surrounding environment of the rail through a depth camera and inputting the surrounding environment into a computer, and then carrying out spike identification from the content captured by the camera by the computer according to a deep learning algorithm;
step three: when the four-wheel track robot arrives at X 0 When the point is reached, the spike appears in the field of view of the depth camera, the spike is identified and initially positioned by the deep learning algorithm, the collected spike position information is fed back to the computer, and after the computer receives the initial positioning information, the computer sends out a driving instruction to make the spike at a slightly slower speed V 1 Driving the four-wheel track robot to continuously move towards the working point X;
step four: four-wheel track robot moves to X 1 When the depth camera accurately identifies and positions the rail spike, and feeds back the accurate positioning information of the spike to the computer, and after receiving the accurate positioning information, the computer sends out a driving instruction again, and the four-wheel track robot is driven to slowly move to the working point X at a slower speed V.
Step five: after the four-wheel track robot reaches the working point X, determining the position of the spike according to a deep learning algorithm, reducing the speed of the four-wheel track robot to 0, and then starting the operation.
Step six: after the four-wheel track robot finishes the operation. Repeating the first step to the fifth step, and carrying out the next group of spike operation.
In the third and fourth steps of the method, the four-wheel track robot moves forwards, the depth camera starts to capture the surrounding environment of the rail and identify and detect rail spikes, and after the camera captures the spikes for the first time, the depth camera identifies, detects and positions the rail spikes. Then the data is fed back to the computer, the computer sends out a driving instruction to drive the four-wheel track robot to move towards the working point X, and the speed of the four-wheel track robot is changed from V in the process 0 Decelerating to V 1 Then from V 1 Decelerating to V, decelerating to 0, and combining preliminary spike identification and preliminary spike identificationAccurate recognition, the recognition accuracy of the spike is improved. According to the method, accurate position information of the spike is provided for the four-wheel track robot according to the deep learning algorithm, the operation efficiency of the robot is improved, and the cost is saved.
In the third step of the method, a deep learning algorithm is used for identifying and detecting the target, and a depth image shot by a depth camera is used for determining the distance between the camera and the object.
In the method, target distance information acquired by a depth camera is fed back to a computer, and the computer sends out a corresponding instruction to enable the four-wheel track robot to move towards a working point X until the four-wheel track robot reaches the working point X to finish the operation.
In the third and fifth steps of the method, the four-wheel track robot reaches the point X 0 And point X 1 After the computer obtains the preliminary positioning information and the accurate positioning information of the spike respectively, the driving instruction sent by the computer changes the moving speed of the four-wheel track robot.
For simplicity of description, specific artificial intelligence algorithms in the above embodiments are not described. At present, the target detection algorithm mainly comprises Yolo, SSD, RCNN, fast R-CNN and the like. The method is mainly based on the Yolo algorithm, and the idea of the method is that the whole image is taken as input, and the position of the binding box and the category to which the binding box belongs are directly returned at the output layer. Compared with other methods, the Yolo has the advantages of high detection speed, direct prediction of the spike based on the image information acquired by the camera, and great improvement of working efficiency. The application does not need to improve the image recognition technology, and only needs to adopt the existing image recognition technology.
The foregoing description of the exemplary embodiments of the application is not intended to limit the application to the particular embodiments disclosed, but on the contrary, the intention is to cover all modifications, equivalents, and alternatives falling within the spirit and scope of the application.
Claims (4)
1. A four-wheel track robot spike identification positioning method based on a deep learning depth camera is characterized by comprising the following steps:
step one: fourth, fourthWheel track robots perform a moving operation on rails at a speed V 0 Moving from the initial point O to the operation point X;
step two: capturing the surrounding environment of the rail through a depth camera and inputting the surrounding environment into a computer, and carrying out spike identification from the content captured by the camera by the computer according to a deep learning algorithm;
step three: when the four-wheel track robot arrives at X 0 When the point is reached, the spike appears in the field of view of the depth camera, the spike is identified and initially positioned by the deep learning algorithm, the collected spike position information is fed back to the computer, and after the computer receives the initial positioning information, the computer sends out a driving instruction to make the spike at a slightly slower speed V 1 Driving the four-wheel track robot to continuously move towards the working point X; x is X 0 The point is located between the initial point O and the operation point X;
step four: four-wheel track robot moves to X 1 When the track is in a point, the depth camera accurately identifies and positions the track spike, and feeds back accurate positioning information of the track spike to the computer, and after receiving the accurate positioning information, the computer sends out a driving instruction again to drive the four-wheel track robot to slowly move to an operation point X at a slower speed V; x is X 1 The point is at X 0 Between the point and the working point X;
step five: after the four-wheel track robot reaches the working point X, determining the position of the spike according to a deep learning algorithm, reducing the speed of the four-wheel track robot to 0, and then starting the operation;
step six: and after the four-wheel track robot finishes the operation, repeating the step one to the step five, and carrying out the next group of spike operation.
2. The four-wheeled orbital robot spike identification and positioning method based on the deep learning depth camera according to claim 1, wherein the method comprises the following steps: in the third step and the fourth step, the four-wheel track robot moves forwards, the depth camera starts to capture the surrounding environment of the rail and identify and detect rail spikes, and when the camera captures the rail spikes for the first time, the rail spikes are identified, detected and positioned; then the data is fed back to the computer, the computer sends out a driving instruction to drive the four-wheel track robot to the working pointX moves, in the process, the speed of the four-wheel track robot is changed from V 0 Decelerating to V 1 Then from V 1 And the speed is reduced to V, the final speed is reduced to 0, and the primary identification and the accurate identification of the spike are combined, so that the identification precision of the spike is improved.
3. The four-wheeled orbital robot spike identification and positioning method based on the deep learning depth camera according to claim 2, wherein the method is characterized by comprising the following steps: and thirdly, identifying and detecting the target by using a deep learning algorithm, and determining the distance between the camera and the object by using a depth image shot by the depth camera.
4. The four-wheel track robot spike identification positioning method based on the deep learning depth camera, which is characterized by comprising the following steps of: v (V) 0 >V 1 >V>0。
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310332159.2A CN117132882A (en) | 2023-03-31 | 2023-03-31 | Four-wheel track robot spike identification and positioning method based on deep learning depth camera |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310332159.2A CN117132882A (en) | 2023-03-31 | 2023-03-31 | Four-wheel track robot spike identification and positioning method based on deep learning depth camera |
Publications (1)
Publication Number | Publication Date |
---|---|
CN117132882A true CN117132882A (en) | 2023-11-28 |
Family
ID=88860615
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202310332159.2A Pending CN117132882A (en) | 2023-03-31 | 2023-03-31 | Four-wheel track robot spike identification and positioning method based on deep learning depth camera |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN117132882A (en) |
-
2023
- 2023-03-31 CN CN202310332159.2A patent/CN117132882A/en active Pending
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109580657B (en) | Crack detection method in bridge quality detection | |
CN109754368B (en) | Crack splicing method in bridge quality detection | |
CN103464383A (en) | Industrial robot sorting system and method | |
CN111389763A (en) | Coal gangue sorting system and method based on image data processing | |
CN112253168B (en) | Automatic assembling method and device for segments of shield tunneling machine | |
CN109159113B (en) | Robot operation method based on visual reasoning | |
CN101098465A (en) | Moving object detecting and tracing method in video monitor | |
CN206276628U (en) | A kind of express mail automatic sorting device of view-based access control model | |
CN102806560A (en) | Method capable of automatically eliminating motion accumulated error of robot | |
CN109434800A (en) | A kind of radio patrol checking machine people system and control method | |
CN103722468B (en) | That recognizes face machine for taper roll bearing automatic charging recognizes plane system | |
CN113982690A (en) | Control method, device and system based on anchoring and protecting operation and tunneling, anchoring and protecting all-in-one machine | |
CN111582135B (en) | Excavator hand operation proficiency evaluation method and device based on working stage identification | |
CN1766907A (en) | Multi-target image recognition method based on cluster genetic algorithm | |
CN113927601B (en) | Method and system for realizing precise picking of mechanical arm based on visual recognition | |
CN117132882A (en) | Four-wheel track robot spike identification and positioning method based on deep learning depth camera | |
CN107992061A (en) | A kind of the wisdom laboratory machine people means of delivery and system | |
CN109692826A (en) | A kind of chip on-line sorting platform | |
CN106379784A (en) | Visual detection method and system for mine reinforcing steel bar inhaul cable | |
CN104281832A (en) | Visual identity industrial robot | |
CN207923179U (en) | Intelligent patrol detection vehicle | |
CN103148783B (en) | A kind of automatic testing method of valve rocker installation site | |
CN214577157U (en) | Intelligent dust source tracking, spraying and dust falling system of mining coal mining machine | |
Gietler et al. | Forestry Crane Automation using Learning-based Visual Grasping Point Prediction | |
CN209615485U (en) | A kind of radio patrol checking machine people's system |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |