CN106708043A - Method of improving Visual Graph in complicated map - Google Patents
Method of improving Visual Graph in complicated map Download PDFInfo
- Publication number
- CN106708043A CN106708043A CN201611144910.2A CN201611144910A CN106708043A CN 106708043 A CN106708043 A CN 106708043A CN 201611144910 A CN201611144910 A CN 201611144910A CN 106708043 A CN106708043 A CN 106708043A
- Authority
- CN
- China
- Prior art keywords
- light
- map
- point
- node
- visual graph
- 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
- G05—CONTROLLING; REGULATING
- G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
- G05D1/00—Control of position, course or altitude of land, water, air, or space vehicles, e.g. automatic pilot
- G05D1/02—Control of position or course in two dimensions
- G05D1/021—Control of position or course in two dimensions specially adapted to land vehicles
- G05D1/0231—Control of position or course in two dimensions specially adapted to land vehicles using optical position detecting means
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
- G06Q10/047—Optimisation of routes or paths, e.g. travelling salesman problem
Abstract
The invention discloses a method of improving Visual Graph in a complicated map, which is thus applicable to a map with a convex polygon obstacle and also applicable to an SLAM map for a general complicated terrain. Visual Graph is a path planning algorithm brought forward in 1979, the characteristic that light travels in a straight line is simulated, the starting point and the ending point are connected with angular points of the convex polygon obstacle to form paths, but Visual Graph is only limited to the convex polygon obstacle. The main idea that light travels in the straight line is still used, light is emitted from the starting point and the ending point of a path, through reflection of the obstacle, a path connecting the starting point and the ending point with the light path as a carrier is finally acquired, a function of path planning is realized, simply using an obstacle vertex connection line as a carrier does not happen, and thus, the method is not limited to a convex polygon obstacle but is also applicable to a general complicated terrain.
Description
Technical field
It is specifically a kind of for ground the present invention relates to a kind of path planning algorithm in intelligent robot motion planning field
The improvement of Visual Graph algorithms, the map for making it can not only be applied to convex polygonal obstacle under the complicated map of shape
Under, it is also possible to it is used under the map of complexity.
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 to deliver.
It refers to ability that how robot decision-making moves to another point from the certain point of map that path planning is.Require first
Robot is obtained in that 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, but its application has strict limitation to map, it is desirable to
If the barrier convex polygon of map, and each barrier apex coordinate, it is known that then connecting the top of each barrier
Point, and with these lines as carrier, generate path.But in practical application scene, landform is sufficiently complex, in map
Barrier shape is irregular, and its apex coordinate is also unknown, therefore directly to utilize Visual Graph algorithms
It is infeasible.The method of the existing solution problem is that map is pre-processed first, using morphological method, such as point
Water ridge algorithm, Feature Correspondence Algorithm etc., can extract the border of the characteristic point or barrier in map, acquired disturbance thing it is big
Shape is caused, and each characteristic point is connected with straight line, barrier is converted into a convex polygon, reapply Visual Graph calculations
Method.The amount of calculation of this method is not small, and the extra algorithm being related to is more, and efficiency is difficult to estimate, and is transformed into barrier at it
During convex polygon, a certain degree of expansion or corrosion can be carried out to the edge of barrier, script is not barrier in being likely to result in map
Hindering the region of thing turns into barrier, is originally that the region of barrier turns into and can pass through, and may cause that the path for finally giving is illegal.
The content of the invention
Technology solve problem of the invention：Hard requirements of the original Visual Graph to map is overcome, is proposed a kind of
The method that Visual Graph are improved under complicated map, the map for making it no longer be only limitted to convex polygonal obstacle, in complexity
SLAM maps under can also run, and without being pre-processed to map, the density parameter of node is also easy to adjustment, with life
Into accurate path.
Technical solution of the invention：A kind of method that Visual Graph are improved under complicated map, its feature exists
In：Visual Graph can be made to may not only be applied to the map of convex polygonal obstacle, can also to general SLAM maps
It is applicable, its step 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, beginning and end is given, beginning and end is stored in a sequence list,
And launch light since beginning and end, and index path is begun setting up, 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, the point is given up if distance is less than a threshold value, otherwise retained, the process is referred to as node store decision process；
Step 4：Since the node for retaining, re-emit light, and repeat step 3 process, this process is referred to as repeating
Light emission process；
Step 5：When all light collision obstacles all cannot generate new node, judgement is built figure and is terminated；
Step 6：A* pathfinding algorithms are applied on the topological map built up, you can obtain a connection source and terminal, with
Light path is the path of carrier, and this process is referred to as pathfinding 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.,
Two dimensional surface space is represented, each element in 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 launching 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 advance a pixel, it will detect the point whether be
Barrier, if that can continue propulsion if, can be encountered if that can not be represented if by point, the point be recorded, into sentencing for step 3
Disconnected process；
(3) if the light linear distance of a direction is too small, then it represents that the direction is extremely pressed close to barrier, be it is irrational,
If therefore light linear 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 of node, i.e., it is emitted cross light with not yet
Launch all nodes of light, the node of light is not yet launched in another sequence list open storages.During initialization, by starting point and end
Point is all stored in two sequence lists.
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 by point addition 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 point of impingement of light and the barrier of the transmitting of a certain node is 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, that is, are built figure process and terminated.
Present invention advantage compared with prior art is：
(1) present invention continue to use the light straightline propagation thinking of primal algorithm, but breach primal algorithm can be only applied to it is convex
The limitation of polygon barrier map.In complicated map, using the point of impingement of light and blocking surfaces as node, that is, it is similar to
In the summit of convex polygon, and from the node launch new light and generate more nodes, spread all over the whole barrier of map
Surface, and path is formed by carrier of light path, map is had no requirement in itself；
(2) present invention to map without pre-processing, and amount of calculation is minimum.With the side of existing improvement Visual Graph
Method is compared, and prior art needs additional designs algorithm to pre-process map, is fitted with generating a convex polygon map
Original Visual Graph algorithms are answered, very huge amount of calculation is increased.And the present invention has no requirement to map, i.e.,
Map is pre-processed to generate convex polygonal obstacle map without other algorithm for design, directly original place figure is grasped
Make, save a large amount of amounts of calculation, improve operational efficiency；
(3) every threshold parameter of the invention is all extremely easily adjusted, and compared with existing technical method, prior art is needed
Blocking surfaces are expanded or corroded, making the shape of barrier is become regular uniform, and original Visual is adapted to this
Usually there is the illegal situation for passing through barrier in Graph algorithms, the path of gained.And the present invention can be with by adjusting threshold parameter
A more accurate, legal path is obtained, without the edge corrosion to barrier or expansion, the barrier in map will not be caused
Point is corroded as that can be expanded to obstacle object point by point by point, and the path of gained is accurate and legal.
Brief description of the drawings
Fig. 1 is original Visual Graph algorithm effects；
Fig. 2 is the flow chart of the inventive method；
Fig. 3 is design sketch of the present invention.
Specific embodiment
The present invention is generally used in the map of slam algorithms foundation, is actual landform, complex.
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 the matrix variables Mat of opencv, 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：Given beginning and end, if being risen via the position where the location algorithm positioning robot such as SLAM
Point is currently located a little for robot, and terminal is then impact point.Beginning and end is inserted into a sequence list of entitled open first
In, used as the node of light to be launched, while they are stored in into a sequence list of entitled allnode, allnode is for storing up
Deposit the sequence list of all nodes.
Step 3：When open table non-NULLs, taken out since open tables a point with 0 degree of the initial 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 advances a pixel, all right
Next pixel is detected that if that can continue propulsion if, barrier is encountered if that can not be represented if, 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 sends 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 all elements in the point of impingement and allnode tables, when its distance is less than a certain threshold value, then the point is given up, otherwise
Using the point as new prepared 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 two structures of node, 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 is touched with barrier
Hit the detection for a little 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：Cannot generate new node after the transmitting of 360 degree of light of a certain light terminates, then from this when open
Begin, the content of open tables will be reduced, and when all of node all cannot create new node in open tables, open tables will be
Sky, now shows to build figure completion, has obtained a topological map, and its summit is light launch point, its side be node and node it
Between light connection；
Step 7：A* pathfinding algorithms are applied on the topological map of gained, the heuristic cost of A* is set to light between node
The distance of line, from the off, according to the heuristic rule of A*, to terminal, you can obtain the one of connection source and terminal
Path of the bar with light path as carrier.
It is the Visual Graph path planning algorithms before not improving such as Fig. 1, the map of its application can only be as shown in the figure
, barrier is the map of convex polygon, and the summit of each barrier is required for, it is known that by beginning and end and obstacle
Connection is set up on the summit of thing from each other, then can be constituted a communication path from origin-to-destination to be connected as carrier.Road
The advantage in footpath is good linearity, and turn number of times is few, thus speed loss of robot when running is small, but because it can only be used
On simple convex polygonal obstacle map, therefore it is difficult to apply in practical problem.
It is that the present invention improves later Visual Graph method effects such as Fig. 3, the round dot in figure represents light transmitting section
Point, fine rule represents light path, and heavy line represents the path of final acquisition.The present invention inherits former methodical advantage, that is, turn round
Number of times is less, and the speed loss of robot is small, and instant invention overcomes the fatal shortcoming of original algorithm, applies it
In complicated map, and without carrying out pretreatment operation to map, greatly improve its application prospect in practical problem.
Claims (7)
1. a kind of method that Visual Graph are improved under complicated map, it is characterised in that：Visual Graph can be made not
The map of convex polygonal obstacle is applied only for, general SLAM maps can also be applicable, 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, beginning and end is given, beginning and end is stored in a sequence list, and from
Beginning and end starts to launch light, begins setting up index path, and this process is referred to as initialization procedure；
Step 3：When light touches barrier, record the coordinate of the point of impingement, and with existing node in sequence list ask for away from
From, give up the point if distance is less than a threshold value, otherwise retain, the process is referred to as node store decision process；
Step 4：Since the node for retaining, re-emit light, and repeat step 3 process, this process is referred to as repeating light
Emission process；
Step 5：When all light collision obstacles all cannot generate new node, judgement is built figure and is terminated；
Step 6：A* pathfinding algorithms are applied on the topological map built up, you can a connection source and terminal are obtained, with light path
It is the path of carrier, this process is referred to as pathfinding process.
2. the method that Visual Graph are improved under complicated landform according to claim 1, it is characterised in that：The step
In rapid 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
Two dimensional surface space, each element in array represents a pixel of corresponding coordinate on map, if the point is barrier
Point, then be set to 1 by array value, if the point is that array value can be set into 0 by point.
3. the method that Visual Graph are improved under complicated landform according to claim 1, it is characterised in that：The step
In rapid 2, the method for launching light is：
(1) for certain point, from 0 degree, i.e., horizontal right direction starts to launch light every a fixed step-length, until 360
Untill degree；
(2) for each direction, light is pushed ahead pixel-by-pixel, often advances a pixel, it will detect whether the point is to lead to
Cross a little, if propulsion can be continued if, barrier is encountered if that can not be represented if, the point is recorded, into the judgement of step 3
Journey；
(3) if the light linear distance of a direction is too small, then it represents that the direction is extremely pressed close to barrier, is irrational, therefore
If light linear distance is less than a certain threshold value, the light of the direction will be rejected.
4. the method that Visual Graph are improved under complicated landform according to claim 1, it is characterised in that：The step
In rapid 2, the method for storage node is：
Two sequence lists are set, and a sequence list allnode stores all of node, i.e., emitted to cross light and not yet launch
The node of light is not yet launched in all nodes of light, another sequence list open storages；During initialization, by beginning and end all
It is stored in two sequence lists.
5. the method that Visual Graph are improved under complicated landform according to claim 1, it is characterised in that：The step
In rapid 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 the distance is less than certain
One threshold value, then give up the point, otherwise by point addition allnode and open tables.
6. the method that Visual Graph are improved under complicated landform according to claim 1, it is characterised in that：The step
In rapid 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.
7. the method that Visual Graph are improved under complicated landform according to claim 1, it is characterised in that：The step
In rapid 5, building the determination methods that figure terminates is：
When the point of impingement of light and the barrier of the transmitting of a certain node is respectively less than threshold value with distance a little, then the node is not
New node can be generated, hereafter taking out a point from open tables again carries out light transmitting, and now the element number of open tables will
Can reduce, when open tables are space-time, that is, build figure process and terminate.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611144910.2A CN106708043B (en) | 2016-12-13 | 2016-12-13 | A method of improving Visual Graph under complicated map |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611144910.2A CN106708043B (en) | 2016-12-13 | 2016-12-13 | A method of improving Visual Graph under complicated map |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106708043A true CN106708043A (en) | 2017-05-24 |
| CN106708043B CN106708043B (en) | 2019-08-06 |
Family
ID=58936854
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611144910.2A Active CN106708043B (en) | 2016-12-13 | 2016-12-13 | A method of improving Visual Graph under complicated map |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106708043B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108334080A (en) * | 2018-01-18 | 2018-07-27 | 大连理工大学 | A kind of virtual wall automatic generation method for robot navigation |
| CN109799817A (en) * | 2019-01-15 | 2019-05-24 | 智慧航海（青岛）科技有限公司 | A kind of unmanned boat global path planning method based on reflective character |
| CN113108803A (en) * | 2021-04-12 | 2021-07-13 | 北京佰能盈天科技股份有限公司 | Path planning method and device for double-axis positioning system |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101464158A (en) * | 2009-01-15 | 2009-06-24 | 上海交通大学 | Automatic generation method for road network grid digital map based on GPS positioning |
| CN102870430A (en) * | 2010-05-03 | 2013-01-09 | 思科技术公司 | Selecting an optical path for a new connection with the minimum number of optical regenerators |
| US8410920B2 (en) * | 2009-12-04 | 2013-04-02 | Denso Corporation | Proximity notification device, proximity notification program and method for notifying proximity of vehicle |
| CN103679301A (en) * | 2013-12-31 | 2014-03-26 | 西安理工大学 | Complex terrain-based optimal path searching method |
| CN104714555A (en) * | 2015-03-23 | 2015-06-17 | 深圳北航新兴产业技术研究院 | Three-dimensional independent exploration method based on edge |
| CN104737123A (en) * | 2012-10-19 | 2015-06-24 | 哈利伯顿能源服务公司 | Self-defining configuration apparatus, methods, and systems |
| CN105469094A (en) * | 2015-11-28 | 2016-04-06 | 重庆交通大学 | Edge vector line extraction algorithm of binary image of road surface |
| CN105955280A (en) * | 2016-07-19 | 2016-09-21 | Tcl集团股份有限公司 | Mobile robot path planning and obstacle avoidance method and system |
| CN105975975A (en) * | 2016-05-20 | 2016-09-28 | 中国科学技术大学 | Environmental line feature extraction method |
-
2016
- 2016-12-13 CN CN201611144910.2A patent/CN106708043B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101464158A (en) * | 2009-01-15 | 2009-06-24 | 上海交通大学 | Automatic generation method for road network grid digital map based on GPS positioning |
| US8410920B2 (en) * | 2009-12-04 | 2013-04-02 | Denso Corporation | Proximity notification device, proximity notification program and method for notifying proximity of vehicle |
| CN102870430A (en) * | 2010-05-03 | 2013-01-09 | 思科技术公司 | Selecting an optical path for a new connection with the minimum number of optical regenerators |
| CN104737123A (en) * | 2012-10-19 | 2015-06-24 | 哈利伯顿能源服务公司 | Self-defining configuration apparatus, methods, and systems |
| CN103679301A (en) * | 2013-12-31 | 2014-03-26 | 西安理工大学 | Complex terrain-based optimal path searching method |
| CN104714555A (en) * | 2015-03-23 | 2015-06-17 | 深圳北航新兴产业技术研究院 | Three-dimensional independent exploration method based on edge |
| CN105469094A (en) * | 2015-11-28 | 2016-04-06 | 重庆交通大学 | Edge vector line extraction algorithm of binary image of road surface |
| CN105975975A (en) * | 2016-05-20 | 2016-09-28 | 中国科学技术大学 | Environmental line feature extraction method |
| CN105955280A (en) * | 2016-07-19 | 2016-09-21 | Tcl集团股份有限公司 | Mobile robot path planning and obstacle avoidance method and system |
Non-Patent Citations (1)
| Title |
|---|
| JANHARTMANN: "Graph-based visual SLAM and visual odometry using an RGB-D camera", 《PROCEEDINGS OF THE 9TH INTERNATIONAL WORKSHOP》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108334080A (en) * | 2018-01-18 | 2018-07-27 | 大连理工大学 | A kind of virtual wall automatic generation method for robot navigation |
| CN109799817A (en) * | 2019-01-15 | 2019-05-24 | 智慧航海（青岛）科技有限公司 | A kind of unmanned boat global path planning method based on reflective character |
| CN113108803A (en) * | 2021-04-12 | 2021-07-13 | 北京佰能盈天科技股份有限公司 | Path planning method and device for double-axis positioning system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106708043B (en) | 2019-08-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP3690482B1 (en) | Learning method and learning device for integrating an image acquired by a camera and a point-cloud map acquired by radar or lidar in a neural network and testing method and testing device using the same | |
| Bastani et al. | Roadtracer: Automatic extraction of road networks from aerial images | |
| CN110148144B (en) | Point cloud data segmentation method and device, storage medium and electronic device | |
| EP3568334B1 (en) | System, method and non-transitory computer readable storage medium for parking vehicle | |
| CN108334080B (en) | Automatic virtual wall generation method for robot navigation | |
| US20210390329A1 (en) | Image processing method, device, movable platform, unmanned aerial vehicle, and storage medium | |
| CN111094895B (en) | System and method for robust self-repositioning in pre-constructed visual maps | |
| CN106643720B (en) | A kind of map constructing method based on UWB indoor positioning technologies and laser radar | |
| CN107065865A (en) | A kind of paths planning method based on the quick random search tree algorithm of beta pruning | |
| CN105809687A (en) | Monocular vision ranging method based on edge point information in image | |
| CN106708043A (en) | Method of improving Visual Graph in complicated map | |
| Jaspers et al. | Multi-modal local terrain maps from vision and lidar | |
| KR102373492B1 (en) | Method for correcting misalignment of camera by selectively using information generated by itself and information generated by other entities and device using the same | |
| CN111308500B (en) | Obstacle sensing method and device based on single-line laser radar and computer terminal | |
| Rankin et al. | Stereo-vision-based terrain mapping for off-road autonomous navigation | |
| CN110244716A (en) | A method of the robot based on wave front algorithm is explored | |
| CN107016706B (en) | A method of obstacles borders are extracted using Visual Graph algorithm | |
| Zeng et al. | Lookup: Vision-only real-time precise underground localisation for autonomous mining vehicles | |
| KR102372687B1 (en) | Learning method and learning device for heterogeneous sensor fusion by using merging network which learns non-maximum suppression | |
| 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 | |
| CN106971144B (en) | Method for extracting road center line by applying Visual Graph algorithm | |
| CN115307640A (en) | Unmanned vehicle binocular vision navigation method based on improved artificial potential field method | |
| Plachetka et al. | DNN-based recognition of pole-like objects in LiDAR point clouds | |
| Bayerl et al. | Detection and tracking of rural crossroads combining vision and LiDAR measurements | |
| Cho et al. | Sloped terrain segmentation for autonomous drive using sparse 3D point cloud |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |