CN107016706A - A kind of method that application Visual Graph algorithms extract obstacles borders - Google Patents
A kind of method that application Visual Graph algorithms extract obstacles borders Download PDFInfo
- Publication number
- CN107016706A CN107016706A CN201710112558.2A CN201710112558A CN107016706A CN 107016706 A CN107016706 A CN 107016706A CN 201710112558 A CN201710112558 A CN 201710112558A CN 107016706 A CN107016706 A CN 107016706A
- Authority
- CN
- China
- Prior art keywords
- light
- node
- point
- map
- barrier
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20076—Probabilistic image processing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30241—Trajectory
Landscapes
- Control Of Position, Course, Altitude, Or Attitude Of Moving Bodies (AREA)
- Processing Or Creating Images (AREA)
Abstract
The invention discloses a kind of method that application Visual Graph algorithms extract obstacles borders, using the main thought of Visual Graph algorithms, i.e. light along straightline propagation, to carry out the extraction of two-dimentional SLAM maps obstacles borders.Visual Graph are the path planning algorithms proposed in 1979, light is simulated along this characteristic of straightline propagation, the present invention mainly applies the thinking, launch light since any point of map, light is encountered to be stopped after blocking surfaces, record the point of impingement and re-emit light by new transmitting light node of the point of impingement, the whole map of Landfill covering, obtain a series of point in obstacles borders, it is final connect can acquired disturbance thing boundary information, the connection figure between output node coordinate and node.
Description
Technical field
The present invention relates to a kind of auxiliary operation algorithm in intelligent robot motion planning field, two dimension is specifically extracted
The obstacles borders information of SLAM maps, to adapt to the demand of part path planning algorithm.
Background technology
Instantly the robot system research for being capable of independent navigation avoidance is abnormal burning hot, and has had many products applications
In practice.Such as the service robot in restaurant, just possess obstacle recognition, the ability of path planning;And for example family's sweeping robot,
Also the ability that displacement path is planned in clearing is possessed.In terms of public transportation, the application that unmanned plane express delivery is delivered
That has carried out is like a raging fire, path planning must be also used in express delivery delivery, so that unmanned plane can be in avoiding obstacles
On the premise of arrived at most short path and complete deliver.
Path planning is to refer to the ability how robot decision-making moves to another point from the certain point of map.Require first
Robot results in the cartographic information of current environment, and can position the current position of itself, then can just carry out path rule
Draw, position and build figure algorithm it is most practical at present be exactly SLAM algorithms.There are many path planning algorithms, such as RRT, PRM at present
Deng.
Visual Graph are also conventional path planning algorithm, and it mainly applies light along this think of of straightline propagation
Think, using light transmitting is simulated, bump against with barrier and obtain the point of impingement, then light is re-emitted by new transmitting node of the point of impingement
Line, repeats this process to set up a topological diagram using light path as carrier.
In the path planning algorithm or planning problem of part, it is often necessary to the accurate acquired disturbance thing border of robot
Information.Such as when initially proposing Visual Graph algorithms within 1979, it is desirable to which barrier is all convex polygon, and with convex polygon
Line between summit is carrier building topology figure.This can not realize that this is accomplished by over the ground on general SLAM maps
Figure is pre-processed, and complicated landform is changed into the form of convex polygon.Before this, what will be carried out first is exactly obstacle
The extraction on thing border, then fuzzy operation is carried out to border, obtain a series of convex polygon.This is one that obstacles borders are extracted
Individual application scenarios, and for example in semantic imparting and the region segmentation of the semantic map of path planning, it is necessary to be calculated using some morphology
Method is handled map, and these morphologic algorithms first have to obtain the shape of barrier, is further carried out operating, and this
It is accomplished by extracting the boundary information of barrier first, with the shape of this acquired disturbance thing.
The existing algorithm for being usually used in obstacles borders extraction has method of differential operator, watershed algorithm etc., mainly direct
Image is operated, the shade of gray of image is calculated, seeks the violent region of wherein grey scale change, be treated as obstacles borders
And extracted.Its method is to belong to image processing field, many all sufficiently complex, it is difficult to understands and is also difficult to practice and comes out,
It is difficult to be optimized accordingly on specific the problem of, the precision of extraction is difficult artificially to operate.To in path planning problem
Also need to, by a lot of conversion, cause its execution efficiency than relatively low using the result of algorithm.
The content of the invention
The technology of the present invention solves problem:The deficiencies in the prior art are overcome to apply Visual Graph algorithms there is provided one kind
The method for extracting obstacles borders, is extracted applied to the simple map of two dimension and the obstacles borders of SLAM maps, not only can be with
The border of isolated danger is extracted, can also realize and carry well for the vast obstacles borders of the floor spaces such as wall, passage
Take.
The technical solution of the present invention:A kind of method that application Visual Graph algorithms extract obstacles borders, its
Feature is:Applied under the simple map or SLAM maps of two dimension, can effectively extract the boundary information of barrier, its step
It is as follows:
Step 1:Map is converted, the map binary conversion treatment constructed by gray-scale map or SLAM is converted
Map afterwards, this process is referred to as map binarization;
Step 2:For the map after conversion, the domain of passing through on map takes any to be stored in sequence list at random, and
Launch light from that point on, this process is referred to as initialization procedure;
Step 3:When light touches barrier, the coordinate of the point of impingement is recorded, and ask with existing node in sequence list
Distance is taken, gives up the point if apart from less than one threshold value, otherwise retains, the process is referred to as node store decision process;
Step 4:Since the node of reservation, light, and the process of repeat step 3 are re-emitted, this process is referred to as repeating
Light emission process;
Step 5:When all light collision obstacles all can not generate new node, judgement is built figure and terminated;
Step 6:All collision nodes are attached, the boundary information of final output barrier, this process is referred to as connecting
Termination process.
In the step 1, the method for the binaryzation of map is:
Call opencv storehouses to read in given map file, and map matrix is converted into a two-dimentional shaping array, i.e.,
Represent that each element in two dimensional surface space, array represents a pixel of corresponding coordinate on map, if the point is barrier
Hinder object point, then array value is set to 1, if the point is that array value can be set into 0 by point.
In the step 2, the method for transmitting light is:
(1) for certain point, from 0 degree, i.e., horizontal right direction starts to launch light every a fixed step-length, until
Untill 360 degree;
(2) for each direction, light is pushed ahead pixel-by-pixel, often promote a pixel, it will detect the point whether be
The seat for now encountering barrier, if that can continue to promote if, if that can not represent to encounter barrier if, can be recorded by point
Punctuate, into the deterministic process of step 3;
(3) if the light of a direction is apart from too small, then it represents that the direction is extremely pressed close to barrier, be it is irrational,
If therefore light distance is less than a certain threshold value, the light of the direction will be rejected.
In the step 2, the method for storage node is:
Two sequence lists are set, and sequence list allnode stores all nodes, i.e., the emitted light and not yet crossed
Launch all nodes of light, the node of light is not yet launched in another sequence list open storages;During initialization, randomly select
Point is added in open tables.
In the step 3, the decision method of node store is:
When light encounters barrier, all nodes in the point of impingement and allnode are asked for into distance, if this is apart from small
In a certain threshold value, then give up the point, otherwise add the point in allnode and open tables.
In the step 4, the method for repeating light transmitting is:
Take out an element of open gauge outfits, the light emission process shown in repeat step 2.
In the step 5, building the determination methods that figure terminates is:
When the light and the point of impingement of barrier that a certain node is launched are respectively less than threshold value with distance a little, then the section
Point will not generate new node, and hereafter taking out a point from open tables again carries out light transmitting, now the element of open tables
Number will be reduced, and when open tables are space-time, i.e. node generating process terminates.
In the step 6, the connection method of connection procedure is:
Since the gauge outfit of allnode sequence lists, each element asks for distance with all elements thereafter, if its away from
From less than a certain threshold value, then the two nodes are attached.The threshold value has correlation with the above-mentioned threshold value built in figure step 3,
Threshold value should be slightly bigger than the threshold value in step 3 herein, to reach preferable effect.After all elements have been handled in sequence list, connection
Process is to terminate.
The advantage of the present invention compared with prior art is:
(1) present invention has innovatively used the main thought of Visual Graph algorithms to carry out carrying for obstacles borders
Take, the point of impingement, and the annexation set up between the point of impingement are obtained using light collision obstacle, obstacles borders are finally given
Annexation.Therefore the map of differing complexity is adapted to, whether isolated barrier or long straight channel, wall
Wall, can realize preferable Boundary Extraction;
(2) present invention is a kind of pretreatment operation to map, is to provide good support for follow-up path planning, can
Select to use different path planning algorithms, or some path planning algorithms to make with the boundary information according to barrier
More preferable path planning is carried out with the boundary information of barrier;
(3) obstacles borders extraction accuracy of the invention is adjustable, fine or coarse extraction, can be by above-mentioned
Adjusting thresholds in step are very easily adjusted.
Brief description of the drawings
Fig. 1 is that Visual Graph algorithms node generates figure;
Fig. 2 is the flow chart of the inventive method;
Fig. 3 is contour connection figure.
Embodiment
The method that a kind of application Visual Graph algorithms of the present invention extract obstacles borders, is calculated using Visual Graph
The main thought of method, i.e. light are along straightline propagation, to carry out the extraction of two-dimentional SLAM maps obstacles borders.Visual Graph are
The path planning algorithm proposed for 1979, simulates light along this characteristic of straightline propagation, the present invention mainly applies the thinking, from
Any point of map starts to launch light, and light is encountered to be stopped after blocking surfaces, records the point of impingement and with the point of impingement
Light is re-emitted for new transmitting light node, the whole map of Landfill covering obtains a series of point in obstacles borders, most
Connect eventually can acquired disturbance thing boundary information, the connection figure between output node coordinate and node.
As shown in Fig. 2 specific implementation step detailed description of the present invention is as follows:
Step 1:With SLAM algorithms or other build nomography and obtain and be presently in the cartographic information of environment, with .pgm forms
Picture is incoming, calls opencv built-in functions by its binaryzation, obtains opencv matrix variables Mat, then convert it into one
Two-dimensional array, its value is that 0 or 1,0 expression can pass through, and 1 represents there is barrier, and binarization is completed;
Step 2:It is 0 region in the domain of passing through of map, i.e. value, randomly selects a bit, the point will be inserted one first
In individual entitled open sequence list, as the node of light to be launched, while they to be stored in entitled allnode order
Table, allnode is the sequence list for storing all nodes.
Step 3:When open table non-NULLs, since open tables take out a point with originate 0 degree of the angle of departure, i.e. level to
Right transmitting light, when light also will not hit on barrier, light is pushed ahead pixel-by-pixel, often promotes a pixel, all right
Next pixel detected, if can continue to promote if, if can not represent to encounter barrier if, and now light stops
Thrust is entered, and records the collision point coordinates;
Step 4:The point of impingement is detected, if the distance that the point of impingement is sent a little with light is less than a certain threshold value,
Represent that the direction is extremely pressed close to barrier, the light of the direction is irrational, should be given up.If light rationally, is asked for
The Euclidean distance of the point of impingement and all elements in allnode tables, when its distance is less than a certain threshold value, then the point is given up, otherwise
Using the point as new preparation transmitting node, while being stored in the table tail of open tables and allnode tables, and the node and light are sent out
Penetrate node and set up connection, connection is implemented as each self-contained pointer in the structure of two nodes, makes two objects
Comprising pointer coreference, that is, realize that node is connected, that is, the side in topological diagram;
Step 5:Give launch angle one step angle, repeat the process of light transmitting, and light and barrier are touched
Hit a detection for carrying out step 4.The light transmitting of the point terminates if angle reaches 360 degree, repeat step 3, again from open
A point is taken out in table, repeats light transmitting;
Step 6:Can not generate new node after the 360 degree of light transmittings of a certain light terminate, then from this when open
Begin, the content of open tables will be reduced, and when node all in open tables all can not create new node, open tables will be
Sky, now shows node generation completion, has obtained the information of all obstacles borders points of impingement, be stored in allnode sequence lists
In;
Step 7:All nodes in allnode are handled, since gauge outfit, are compared it with all nodes thereafter with this
Compared with asking at the distance between 2 points, if the distance is less than a certain threshold value, the annexation set up between 2 points, visualization is aobvious
It is attached between being shown as at 2 points with straightway, calls opencv line functions, draws the line between 2 points.When all
Node is all after the operation, it is possible to obtain an obstacles borders figure, i.e., as shown in Figure 3.
It it is showing for the node that is obtained by Visual Graph algorithms on a SLAM map and its light path such as Fig. 1
It is intended to.
It is the latter linked design sketch of node generation, it can be seen that the extraction effect of obstacles borders still compares such as Fig. 3
Alright.
Claims (7)
1. a kind of method that application Visual Graph algorithms extract obstacles borders, it is characterised in that:Using light along straight-line transmitting
This characteristic is broadcast, the boundary information of barrier is extracted, its step is as follows:
Step 1:Map is converted, by the map binary conversion treatment constructed by gray-scale map or SLAM, after being converted
Map, this process is referred to as map binarization;
Step 2:For the map after conversion, the random storage node in domain of passing through on map is opened in sequence list, and from this
Originate and penetrate light, this process is referred to as initialization procedure;The method of the storage node is:Two sequence lists, an order are set
Table is allnode tables, stores all nodes, i.e., the emitted all nodes crossed light and not yet launch light, another
Sequence list is open tables, and the node of light is not yet launched in storage;During initialization, the point randomly selected is added in open tables;
Step 3:When light touches barrier, the coordinate of the point of impingement of record light and barrier, and with being had in sequence list
Node ask for distance, if giving up the point if less than given threshold, otherwise retain, the process is referred to as node store and judged
Journey;
Step 4:Since the node of reservation, light, and the process of repeat step 3 are re-emitted, acquisition is a series of to meet node
Judge the node of beachhead demand, this process is referred to as repeating light emission process;
Step 5:When all light collision obstacles all can not generate new node, decision node generation terminates;
Step 6:All nodes are attached, the boundary information of final output barrier, this process is referred to as connection procedure.
2. the method that application Visual Graph algorithms according to claim 1 extract obstacles borders, it is characterised in that:
In the step 1, the method for the binaryzation of map is:
Call opencv storehouses to read in given map file, and map matrix is converted into a two-dimentional shaping array, that is, represent
Each element in two dimensional surface space, array represents a pixel of corresponding coordinate on map, if the pixel is barrier
Hinder object point, then array value is set to 1, if the pixel is that array value can be set into 0 by point.
3. the method that application Visual Graph algorithms according to claim 1 extract obstacles borders, it is characterised in that:
In the step 2, the method for transmitting light is:
(1) for a certain node, from 0 degree, i.e., horizontal right direction starts to launch light every a fixed step-length, until
Untill 360 degree;
(2) for each direction, light is pushed ahead pixel-by-pixel, often promotes a pixel, it will whether detect the pixel
, if that can continue to promote if, if that can not represent to encounter barrier if, can to record by point and now encounter barrier
Coordinate points, into the deterministic process of step 3;
(3) if the light of a direction transmitting is from transmitting starting point to the threshold for being less than setting with the distance between the point of impingement of barrier
Value, then the light of a direction will be rejected.
4. the method that application Visual Graph algorithms according to claim 1 extract obstacles borders, it is characterised in that:
In the step 3, detailed process is as follows:
When light encounters barrier, by all nodes in the point of impingement of the light and barrier and allnode tables ask for away from
From if the distance gives up the point less than a certain threshold value of setting, otherwise by point addition allnode tables and open tables.
5. the method that application Visual Graph algorithms according to claim 1 extract obstacles borders, it is characterised in that:
In the step 4, it is specially:
Take out an element of open gauge outfits, the light emission process shown in repeat step 2.
6. the method that application Visual Graph algorithms according to claim 1 extract obstacles borders, it is characterised in that:
In the step 5, the method that decision node generation terminates is:
When the light and the point of impingement of barrier that a certain node is launched are respectively less than a certain threshold value of setting with distance a little,
Then the node will not generate new node, and hereafter taking out a node from open tables again carries out light transmitting, now open tables
In node number will reduce, when open tables are space-time, i.e. node generating process terminates.
7. the method that application Visual Graph algorithms according to claim 1 extract obstacles borders, it is characterised in that:
In the step 6, connection method is:
Since the gauge outfit of allnode tables, each node asks for distance with all nodes thereafter, is set if its distance is less than
The two nodes, then be attached, the threshold value of the setting has correlation with the threshold value in step 3, herein threshold by fixed a certain threshold value
Value should be slightly bigger than the threshold value in step 3, and after reaching that all node processings are complete in preferable effect, sequence list, connection procedure is
Terminate.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710112558.2A CN107016706B (en) | 2017-02-28 | 2017-02-28 | A method of obstacles borders are extracted using Visual Graph algorithm |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710112558.2A CN107016706B (en) | 2017-02-28 | 2017-02-28 | A method of obstacles borders are extracted using Visual Graph algorithm |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107016706A true CN107016706A (en) | 2017-08-04 |
CN107016706B CN107016706B (en) | 2019-08-06 |
Family
ID=59440527
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710112558.2A Active CN107016706B (en) | 2017-02-28 | 2017-02-28 | A method of obstacles borders are extracted using Visual Graph algorithm |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107016706B (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111148033A (en) * | 2019-12-19 | 2020-05-12 | 广州赛特智能科技有限公司 | Auxiliary navigation method of self-moving equipment |
CN111399507A (en) * | 2020-03-19 | 2020-07-10 | 小狗电器互联网科技(北京)股份有限公司 | Method for determining boundary line in grid map and method for dividing grid map |
CN112180945A (en) * | 2020-10-22 | 2021-01-05 | 南京苏美达智能技术有限公司 | Method for automatically generating barrier boundary and automatic walking equipment |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103365293A (en) * | 2013-05-08 | 2013-10-23 | 北京大学深圳研究生院 | Robot safety path planning method based on dynamic region division |
CN103529843A (en) * | 2013-10-17 | 2014-01-22 | 电子科技大学中山学院 | Lambda path planning algorithm |
CN103884330A (en) * | 2012-12-21 | 2014-06-25 | 联想(北京)有限公司 | Information processing method, mobile electronic device, guidance device, and server |
-
2017
- 2017-02-28 CN CN201710112558.2A patent/CN107016706B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103884330A (en) * | 2012-12-21 | 2014-06-25 | 联想(北京)有限公司 | Information processing method, mobile electronic device, guidance device, and server |
CN103365293A (en) * | 2013-05-08 | 2013-10-23 | 北京大学深圳研究生院 | Robot safety path planning method based on dynamic region division |
CN103529843A (en) * | 2013-10-17 | 2014-01-22 | 电子科技大学中山学院 | Lambda path planning algorithm |
Non-Patent Citations (4)
Title |
---|
MIKE PHILLIPS 等: "Anytime Incremental Planning with E-Graphs", 《2013 IEEE INTERNATIONAL CONFERENCE ON ROBOTICS AND AUTOMATION》 * |
岳荣刚 等: "基于混合整数线性规划的爬壁机器人路径规划", 《北京航空航天大学学报》 * |
张之瑶: "基于地图建立的无人地面车路径规划", 《研究论文》 * |
王正光 等: "基于层次图变换的多Agent组织结构动态重组机制", 《软件学报》 * |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111148033A (en) * | 2019-12-19 | 2020-05-12 | 广州赛特智能科技有限公司 | Auxiliary navigation method of self-moving equipment |
CN111148033B (en) * | 2019-12-19 | 2021-06-18 | 广州赛特智能科技有限公司 | Auxiliary navigation method of self-moving equipment |
CN111399507A (en) * | 2020-03-19 | 2020-07-10 | 小狗电器互联网科技(北京)股份有限公司 | Method for determining boundary line in grid map and method for dividing grid map |
CN111399507B (en) * | 2020-03-19 | 2024-04-02 | 小狗电器互联网科技(北京)股份有限公司 | Method for determining boundary line in grid map and method for dividing grid map |
CN112180945A (en) * | 2020-10-22 | 2021-01-05 | 南京苏美达智能技术有限公司 | Method for automatically generating barrier boundary and automatic walking equipment |
CN112180945B (en) * | 2020-10-22 | 2023-08-04 | 南京苏美达智能技术有限公司 | Method for automatically generating obstacle boundary and automatic walking equipment |
Also Published As
Publication number | Publication date |
---|---|
CN107016706B (en) | 2019-08-06 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN112132893B (en) | Visual SLAM method suitable for indoor dynamic environment | |
Bastani et al. | Roadtracer: Automatic extraction of road networks from aerial images | |
CN108805327B (en) | Method and system for robot path planning and environment reconstruction based on virtual reality | |
CN111666921A (en) | Vehicle control method, apparatus, computer device, and computer-readable storage medium | |
CN112802204B (en) | Target semantic navigation method and system for three-dimensional space scene prior in unknown environment | |
Prieto et al. | As-is building-structure reconstruction from a probabilistic next best scan approach | |
Sales et al. | Adaptive finite state machine based visual autonomous navigation system | |
Quintana et al. | Semantic scan planning for indoor structural elements of buildings | |
KR102372703B1 (en) | Learning method and learning device for integrating object detection information acquired through v2v communication from other autonomous vehicle with object detection information generated by present autonomous vehicle, and testing method and testing device using the same | |
CN108334080A (en) | A kind of virtual wall automatic generation method for robot navigation | |
JP2020038660A (en) | Learning method and learning device for detecting lane by using cnn, and test method and test device using the same | |
CN107016706B (en) | A method of obstacles borders are extracted using Visual Graph algorithm | |
US20200334887A1 (en) | Mobile robots to generate occupancy maps | |
JP2020038661A (en) | Learning method and learning device for detecting lane by using lane model, and test method and test device using the same | |
Zeng et al. | Lookup: Vision-only real-time precise underground localisation for autonomous mining vehicles | |
CN109376677A (en) | A kind of video behavior detection method merged based on shape-movement double fluid information | |
CN110619299A (en) | Object recognition SLAM method and device based on grid | |
CN111507161B (en) | Method and device for heterogeneous sensor fusion by utilizing merging network | |
CN105374030A (en) | Background model and moving object detection method and system | |
EP3686776B1 (en) | Method for detecting pseudo-3d bounding box to be used for military purpose, smart phone or virtual driving based on cnn capable of converting modes according to conditions of objects | |
CN106708043B (en) | A method of improving Visual Graph under complicated map | |
Liu et al. | Dlc-slam: A robust lidar-slam system with learning-based denoising and loop closure | |
Fritsch et al. | Modeling facade structures using point clouds from dense image matching | |
CN106971144B (en) | Method for extracting road center line by applying Visual Graph algorithm | |
Yu et al. | Co-NavGPT: Multi-robot cooperative visual semantic navigation using large language models |
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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |