CN104407616B - Dynamic path planning method for mobile robot based on immune network algorithm - Google Patents
Dynamic path planning method for mobile robot based on immune network algorithm Download PDFInfo
- Publication number
- CN104407616B CN104407616B CN201410729927.9A CN201410729927A CN104407616B CN 104407616 B CN104407616 B CN 104407616B CN 201410729927 A CN201410729927 A CN 201410729927A CN 104407616 B CN104407616 B CN 104407616B
- Authority
- CN
- China
- Prior art keywords
- antibody
- path
- robot
- affinity
- network algorithm
- 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.)
- Expired - Fee Related
Links
Landscapes
- Control Of Position, Course, Altitude, Or Attitude Of Moving Bodies (AREA)
- Feedback Control In General (AREA)
- Manipulator (AREA)
Abstract
The invention discloses a dynamic path planning method for a mobile robot based on an immune network algorithm. The method comprises the following steps: firstly, acquiring information of working environment of the mobile robot, processing the acquired information, and expressing a static obstacle in the surrounding environment by using a rectangular bounding box; secondly, planning a global path by applying an improved immune network algorithm on the basis of the collected static environment information; thirdly, allowing the robot to move forward along the planned path, if the unknown obstacle is found, predicting whether the unknown obstacle hampers normal running of the robot or not, if the unknown obstacle influences the normal running of the robot, predicting a dangerous area with possible collision; fourthly, forming a new local environment view according to the predicated result, and planning a new local path by applying the immune network algorithm; fifthly, when the robot reaches a temporary target point of local planning, allowing the robot to return to the planned global path for continuous advancement. According to the dynamic path planning method disclosed by the invention, a global optimal path can be solved under the condition that a complex environment is guaranteed, and the optimum of the local dynamic area path planning can also be guaranteed.
Description
Technical field：The present invention relates to a kind of mobile robot dynamic path planning method based on Immune network algorithm.
The method can be widely used in and solve mobile robot path planning problem under relative complex dynamic environment.
Background technology：Path planning is one of core content of mobile robot research, be its objective is with obstacle
In the working environment of thing, an appropriate safety movement path from given starting point to impact point is found, make robot in motion
During can without collision bypass all barriers.In the relation technological researching of mobile robot, Path Planning Technique is
One important field of research.
In most cases mobile robot is in dynamic uncertain environments in reality, to the dynamic disorder in this kind of environment
Thing, mobile robot is difficult to be possessed priori and only has component environment information.In this case, mobile robot can only
Local paths planning is carried out according to the environmental information for real-time detecting.In order to solve this kind of path planning problem, typically will pass
In combination with the method for prediction of collision, a kind of such as pre-collision sensing method is in combination with Artificial Potential Field Method for the global path planning algorithm of system
Algorithm, the speed of service of the algorithm according to the position that collides with dynamic barrier of prediction to change mobile robot carry out
Hide, but it does not account for the global path planning ability of mobile robot under complex environment.To solve above-mentioned deficiency, this
Invention is applied to the mobile robot path planning under the circumstances not known of part using a kind of improved Immune network algorithm.
Immune network algorithm is an important method in Artificial Immune Algorithm.Proposed in immune system based on Jerne
The distinct network that portion is adjusted is theoretical, and de Castro and Timmis proposed Immune network algorithm (opt- in 2002
AiNet), it has a fast convergence rate, and solving precision is high, and stability is good, and can efficiently against " precocity " the characteristics of,
Make it that there is advantage in terms of global optimizing and optimal speed, be thus suitable for solving the complexity in mobile robot path planning
Many obstacle Global motion planning problems and quick dynamic obstacle avoidance planning problem.
The content of the invention：
Goal of the invention：In order to improve the solution quality of mobile robot path planning problem and the problem of solution efficiency, this
Invention provides a kind of mobile robot dynamic path planning method based on Immune network algorithm, and the method not only can solve ring
Mobile robot global path planning under the complex situations of border, the local paths planning being also applied under component environment unknown situation.
Technical scheme：
A kind of mobile robot dynamic path planning method based on Immune network algorithm, it is characterised in that：It is primarily based on
Visual Graph method carries out pretreatment to global static environment space, cooks up global path using Immune network algorithm afterwards, is expert at
Real-time detection ambient condition information during entering, to the unknown dynamic disorder into sensing range risk prediction is carried out, according to
The result of prediction forms new local environment Visual Graph and cooks up new local path using Immune network algorithm, works as robot
Continue in the global path of the preplanning returned to after the transient target point for reaching sector planning.
(1), Immune network algorithm realizes process：
Through the improvement to Immune network algorithm and and Visual Graph combination, algorithm concrete implementation step is as follows：
(1) pretreatment is carried out to the environment space of mobile robot operation, builds Visual Graph；
(2) according to the foundation and analysis of Visual Graph, summarize priori and extract vaccine, including antibody or the path of generation
There is no setback in the bounding box for not passing through barrier and the antibody for generating or path；
(3) initialization of population：First vaccination, it is m bars road to produce m antibody by implantation vaccine in search space
Footpath, the generating mode in path is progressively to extend to impact point from starting point, is that m paths constitute initial parent by this m antibody
PopulationWherein subscript i represented for the i-th generation, and subscript m represents antibody number；
(4) stop condition：It is anti-finally to determine according to the concentration between the fitness and antibody between antibody and antigen
The affinity of body, calculate current parent for population genetic algebra andIn all antibody affinity, if meeting condition, eventually
Only computing and output result, otherwise continue；
(5) clonal expansion：N clone body is produced respectively to each antibody (path) in above-mentioned antibody population, is now expanded
For m*n clonal antibody (path), now antibody population is changed intoN times is all replicated per paths, the path variation of next step
More fully will increased the multiformity of per generation antibody；
(6) high frequency closedown：Clonal antibody groupIn each clone antibody be intended to experience high frequency closedown obtainReferring here to antagonist path is increased or deletion of node；
(7) adaptation value is recalculated：All individualities in antagonist group, including each father's antibody and its variant clone be anti-
Body, recalculates its affinity；
(8) memory antibody group updates：From each father's antibody and its clonal antibodyIn, select that there is highest parent
With the variant clone antibody of degree, the m new elite antibody variants group with highest affinity is collectively constitutedAnd made
For update memory antibody group candidate, then fromWithIn select m affinity highest and constitute new elite antibody variants
GroupCalculate the average affinity in memory antibody group again afterwards；
(9) immune self-regulation and receptor editing：It is random to generate r (r<M) individual brand-new antibodyReplace now antibody populationThe antibody of middle r with minimum affinity, to simulate Immune System in receptor editing process, increase many of antibody population
Sample, consequently, it is possible to algorithm generates follow-on antibody populationReturn to step 4 afterwards；
(2), the mobile robot active path planning step based on Immune network algorithm：
It is the unknown situation in environmental information part for mobile robot working space, specific path planning step is such as
Under：
(1) global path planning is carried out to the Immune network algorithm of known static context information application enhancements；
(2) robot advances along the path cooked up；
(3) judge whether to reach impact point, arrival then terminates, be less than and advance up to then continuing on the path cooked up；
Due to the information of the static-obstacle thing in environment space it is previously known, then whenever the step of robot ambulation one will be sentenced
Whether the distance between disconnected robot and known each static-obstacle thing are less than the half of the sensing range of scrolling windows, i.e. sensor
Footpath, thus can show whether these static-obstacle things are had occurred in the range of scrolling windows；
(4) judge whether the dyskinesia hinders the normally travel of robot；
(5) predict labelling unknown dynamic barrier can with along the robot advanced of the good path of preplanning connecing
Certain T moment of getting off collides, then the location records at this Unknown Motion barrier T moment are got off, as " static-obstacle
Thing " process, it is contemplated that the uncertain factor in the error and actual environment of prediction, expands the region of this " static-obstacle thing "
Some, are denoted as deathtrap；
(6) after deathtrap is completely within scrolling windows, robot is by the preplanning therewith of current time scrolling windows
Path intersection point as interim localized target point, the position of current robot as starting point, the static-obstacle thing in perceived window and
New path, road of the robot along new planning are cooked up as local environmental information using Immune network algorithm in deathtrap
Footpath travels, return to step 3.
(2) robot is first without considering scrolling windows during traveling in (4th) step of step, until sensor
It was found that new unknown barrier is entered within scrolling windows, i.e., unknown barrier is less than the radius of scrolling windows with the distance of robot
When, the labelling unknown dynamic barrier and start to predict the movement locus of unknown barrier；The result of prediction is divided into following two
The situation of kind：One is the normally travel that unknown barrier will not hinder robot, then the good road of the preplanning that robot is continued on
Footpath is advanced, that is, return to step 2；Two is to predict the normally travel that the dyskinesia may interfere with robot, then go to
(2) (5th) step of step.
The design of antibody in Immune network algorithm：
After Visual Graph method is to the process of mobile work robot environment space, constitutes antibody path and only need to consider each
Element in the set V on individual barrier summit, it is not necessary to any point outside considering in environment space, such hunting zone will be aobvious
Write and reduce so that efficiency of algorithm is improved.The structure of each node is as follows：
| IDpoint | IDobject | x | y |
Wherein IDpoint represents node serial number, and IDobject represents the affiliated barrier numbering of node, and x, y represent respectively section
The transverse and longitudinal coordinate of point；In mobile robot path planning problem, antigen represents problem to be solved i.e. optimal path, this Shen
Please be set as the Euclidean distance of starting point s and impact point g, antibody represents the solution of required problem, that is, the feasible path for searching out, it
A series of broken line connected into by the starting point of mobile robot, impact point and intermediate nodes is constituted, and is clearly to describe road
Footpath is moved towards, and through the order of node when the order of element represents moveable robot movement in antibody, antibody adopts string encoding
Mode, its length can change with the number of node, the data structure of antibody enter it is lower shown in：
| Affinity | Density | Fitness | Length | s | … | v | … | g |
Wherein Affinity represents the affinity of antibody, evaluates the quality of antibody with affinity here；Fitness represents anti-
The fitness of body, the antibody with shorter path has higher fitness；Density represents the concentration of antibody, i.e., identical anti-
Ratio in antibody population shared by body, Length represents length of the antibody by s to g, and remainder is the node for constituting path, this
Storage organization when antibody changes, the length of flexibly changing antibody；
The calculating of affinity in Immune network algorithm：
The application adds here the calculating of antibody concentration to suppress and avoiding the appearance of precocious phenomenon, concrete affine
The computing formula of degree is as follows：
Wherein Affinity (i) represents the affinity of antibody i；Fitness (i) represents the fitness of antibody i, and distance has
Close, the distance in antibody path is shorter, and the fitness of its antibody is higher；Density (i) represents the concentration of antibody i, i.e. identical
Antibody path is more, and the concentration of antibody is bigger；α is a positive constant, can be seen that fitness is bigger by this formula, parent
It is bigger with spending；Concentration is higher, and affinity is less.
Advantage and effect：The present invention provides a kind of mobile robot active path planning side based on Immune network algorithm
Method, the application using path length as weighing the whether excellent main standard in path, known to static context information, unknown dynamic
Have studied under obstacle information circumstances not known Idiotype immune network theory in opt-aiNet algorithms in combination with Visual Graph
Paths planning method.It is primarily based on Visual Graph method to be described global static environment space, afterwards using Immune network algorithm
Cook up global path.The real-time detection ambient condition information in robot moving process, to into the unknown dynamic of sensing range
State obstacle carries out risk prediction, forms new local environment Visual Graph according to the result of prediction and plans using Immune network algorithm
Go out new local path, so as to realize local dynamic station planning and the combination of global static programming.Additionally, in order to avoid immunological network
The problems such as carrying out the evolution precocity, the antibody deficiency multiformity that easily occur during robot path planning, introduces the meter of antibody concentration
Calculation method.The present invention can also ensure the optimum that local dynamic station zone routing is planned on the basis of global optimum path is ensured.
Description of the drawings：
Fig. 1 robots working environment schematic diagram；
Fig. 2 local paths planning flow charts；
Fig. 3 Immune network algorithm flow charts；
The relative complex path planning of Fig. 4 working environments；
The statistics of Fig. 5 evolutionary generations；
The analysis of Fig. 6 population antibody affinity；
There is the path planning of unknown dynamic barrier in Fig. 7.
Specific embodiment：Below in conjunction with the accompanying drawings the present invention is described further：
Fig. 1 represents mobile work robot environment space, and the dolly advanced along path represents mobile robot, its periphery
Dashed circle be mobile robot local sensing scope, i.e., the effectively perceive scope of sensor is (such as laser sensor, ultrasound
Sensor or panoramic vision sensor etc.), s represents the starting point of mobile robot, and g represents the final goal of mobile robot
Point, the static-obstacle thing that black rectangle is represented in environment space (uses OjRepresent), another dolly represents dynamic in environment space
Barrier, arrow represents its direction of motion, and four summits of the Rectangular Bounding Volume of each barrier have fixed numbering (to use vi
To represent), numbering i on its summit is i=1,2 ..., j*4 with the corresponding relation of numbering j of barrier.
Herein in view of the actual size of mobile robot, reserve when barrier bounding box is built certain safety away from
From the dotted line frame of i.e. barrier bounding box periphery, all end points that condition is met in Fig. 1 are connected, that is, Visual Graph are constituted, in figure
A paths (also representing an antibody) are per bar from starting point s to the line of final goal point g.The method for expressing in path is
Through the order on summit, such as path1=s → v during moveable robot movement9→v2→g。
Through pretreatment of the Visual Graph method to environment space, scope all of point reduction from working environment of algorithm search
To the node and impact point of each barrier, the hunting zone huge compression of algorithm is made, the convergence of optimized algorithm greatly improved
Speed, enables to adapt to real-time local paths planning.
A kind of mobile robot path planning step based on Immune network algorithm：
It is the unknown situation in environmental information part for mobile robot working space, as shown in Fig. 2 specific path
Planning step is as follows：
Step 1：Obtain the information of mobile work robot environment and it is processed, by the static state barrier in surrounding
The mode for hindering thing Rectangular Bounding Volume is represented.
In view of mobile work robot environment polytropy caused by algorithm search complexity and real-time requirement,
The application describes the working environment at mobile robot place using Visual Graph method.Note Visual Graph VG=(V, L), wherein V be by
Each summit of each barrier bounding box and starting point s, impact point g and each barrier Rectangular Bounding Volume, the application is adopted
The set constituted with each summit of Rectangular Bounding Volume, L represents " visual " line between each summit, if i.e. these lines
Do not intersect with barrier, think that line segment is " visual ", then the combination of all " visual " lines is current environment space
Visual Graph, what every line also illustrated that mobile robot can operating path.
Planning process is carried out to the working environment space of mobile robot using Visual Graph method, the problem of core is to be built into
This matrix, i.e., the matrix that the size of distance is constituted between each node.With node serial number, the four of each barrier Rectangular Bounding Volume
Individual summit and starting point, impact point to represent Cost matrix in row and column, content in row and column is with two for can directly connecting
Euclidean distance between individual node represents that wherein ∞ represents that the line between two nodes is infinitely great through barrier, i.e. cost.Profit
With Cost matrix, the cost needed for either path can be quickly calculated；After Cost matrix construction complete, will to a certain extent
Immune network algorithm is affected to solve the extraction of mobile robot path planning problem time vaccines and the determination of antibody coding mode.
Step 2：Global path planning is carried out on the basis of the static context information collected using Immune network algorithm.
It is that optimal path is considered as antigen by problem to be solved, the solution (optimal path obtained) of problem is considered as into antibody, with
Machine generates the initial path (antibody) of given quantity, and evolution behaviour is carried out to population using operators such as selection, duplication, transform paths
Make, produce the filial generation path better than parent path.Here the interphase interaction to antibody in network or immune self-regulation are added
The simulation of function, by the similarity degree between calculating antibody, dynamically in balance antibody group antibody number.
This step is carried out to the calculation of the affinity in Immune network algorithm with the coded system of antibody and its gene
Improve.
Using the flow chart of Immune network algorithm realizing route planning as shown in figure 3, its algorithm steps is as follows：
(1) according to the foundation and analysis of Visual Graph, summarize priori and extract vaccine, including the antibody (path) for generating
There is no setback in the bounding box for not passing through barrier and the antibody (path) for generating.
After Visual Graph method is to the process of mobile work robot environment space, constitutes antibody path and only need to consider each
Element in the set V on individual barrier summit, it is not necessary to any point outside considering in environment space, such hunting zone will be aobvious
Write and reduce so that efficiency of algorithm is improved.The structure of each node is as shown in table 1：
| IDpoint | IDobject | x | y |
The nodes encoding form of table 1
Wherein IDpoint represents node serial number, and IDobject represents the affiliated barrier numbering of node, and x, y represent respectively section
The transverse and longitudinal coordinate of point.In mobile robot path planning problem, antigen represents problem to be solved i.e. optimal path, this Shen
Please be set as the Euclidean distance of starting point s and impact point g.Antibody represents the solution of required problem, that is, the feasible path for searching out, it
A series of broken line connected into by the starting point of mobile robot, impact point and intermediate nodes is constituted.Clearly to describe road
Footpath is moved towards, through the order of node when the order of element represents moveable robot movement in antibody.Antibody adopts string encoding
Mode, its length can change with the number of node, shown in the data structure table 2 of antibody：
| Affinity | Density | Fitness | Length | s | … | v | … | g |
The antibody coding form of table 2
Wherein Affinity represents the affinity of antibody, evaluates the quality of antibody with affinity here；Fitness represents anti-
The fitness of body, the antibody with shorter path has higher fitness；Density represents the concentration of antibody, i.e., identical anti-
Ratio in antibody population shared by body, Length represents length of the antibody by s to g, and remainder is the node for constituting path, this
Storage organization when antibody changes, the length of flexibly changing antibody.
In Immune network algorithm, the affinity of antibody plays an important role, and especially updates and exempts from memory antibody group
The value of antibody is weighed in epidemic disease self-regulation step with affinity, so as to antagonist is screened.
The affinity of antibody refers to the bond strength of antibody and antigen, and the application evaluates the quality of antibody with affinity.For
Suppress and avoid the appearance of precocious phenomenon, add the calculating of antibody concentration here.The computing formula of concrete affinity is as follows：
Wherein Affinity (i) represents the affinity of antibody i；Fitness (i) represents the fitness of antibody i, and distance has
Close, the distance in antibody path is shorter, and the fitness of its antibody is higher；Density (i) represents the concentration of antibody i, i.e. identical
Antibody path is more, and the concentration of antibody is bigger；α is a positive constant.Can be seen that fitness is bigger by this formula,
Affinity is bigger；Concentration is higher, and affinity is less.
(2) initialization of population：First vaccination, m antibody (m bars road is produced in search space by implantation vaccine
Footpath), the generating mode in path is progressively to extend to impact point from starting point, and by this m antibody (m paths) initial parent is constituted
PopulationWherein subscript i represented for the i-th generation, and subscript m represents antibody number.
(3) stop condition：It is anti-finally to determine according to the concentration between the fitness and antibody between antibody and antigen
The affinity of body, calculate current parent for population genetic algebra andIn all antibody affinity, if meeting condition, eventually
Only computing and output result, otherwise continue.
(4) clonal expansion：N clone body is produced respectively to each antibody (path) in above-mentioned antibody population, is now expanded
For m*n clonal antibody (path), now antibody population is changed intoN times is all replicated per paths, the path variation of next step
More fully will increased the multiformity of per generation antibody.
(5) high frequency closedown：Clonal antibody groupIn each clone antibody be intended to experience high frequency closedown obtainReferring here to antagonist path is increased or deletion of node.
(6) adaptation value is recalculated：All individualities in antagonist group, including each father's antibody and its variant clone be anti-
Body, recalculates its affinity.
(7) memory antibody group updates：From each father's antibody and its clonal antibodyIn, select that there is highest parent
With the variant clone antibody of degree, the m new elite antibody variants group with highest affinity is collectively constitutedAnd made
For update memory antibody group candidate, then fromWithIn select m affinity highest and constitute new elite antibody variants
GroupCalculate the average affinity in memory antibody group again afterwards.
(8) immune self-regulation and receptor editing：It is random to generate r (r<M) individual brand-new antibodyReplace now antibody populationThe antibody of middle r with minimum affinity, to simulate Immune System in receptor editing process, increase many of antibody population
Sample.Consequently, it is possible to algorithm generates follow-on antibody populationReturn (3) afterwards.
Step 3：Robot advances along the path cooked up.
Step 4：Judge whether to reach impact point, arrival then terminates, be less than and advance up to then continuing on the path cooked up.
Due to the information of the static-obstacle thing in environment space it is previously known, then whenever the step of robot ambulation one will be sentenced
Whether the distance between disconnected robot and known each static-obstacle thing are less than the half of scrolling windows (sensing range of sensor)
Footpath, thus can show whether these static-obstacle things are had occurred in the range of scrolling windows.
Step 5：Judge whether the dyskinesia hinders the normally travel of robot.
Robot is first without considering scrolling windows during traveling, until sensor finds that new unknown barrier is entered
To within scrolling windows when (radius of the distance of i.e. unknown barrier and robot less than scrolling windows), the labelling unknown dynamic disorder
Thing and start to predict the movement locus of unknown barrier.The result of prediction is divided into following two situations：One is unknown barrier
The normally travel of robot will not be hindered, then the good path of the preplanning that robot is continued on is advanced, that is, return to
Step 3.Two is to predict the normally travel that the dyskinesia may interfere with robot, then go to step 6.
Step 6：Predict labelling unknown dynamic barrier can with along the robot advanced of the good path of preplanning
Collide at following certain T moment, then the location records at this Unknown Motion barrier T moment are got off, as " static state barrier
Hinder thing " process, it is contemplated that the uncertain factor in the error and actual environment of prediction, expands the region of this " static-obstacle thing "
It is larger, it is denoted as deathtrap.
Step 7：After deathtrap is completely within scrolling windows, robot is by the preplanning therewith of current time scrolling windows
Path intersection point as interim localized target point, the position of current robot is used as starting point, the static-obstacle thing in perceived window
With deathtrap as local environmental information, new path is cooked up using Immune network algorithm, robot is along new planning
Route, return to step 4.
Through above-mentioned steps description, draw the present invention can not only more efficient realize that mobile robot is complete under complex environment
Office's path planning, can also complete the local paths planning that there is the unknown dynamic disorder in local.
Fig. 4 represents the relative complex global path planning of environment space, is provided with multiple in the movement environment of robot
Intensive obstacle.Each 2 line for arbitrarily not passing through obstacle represents the feasible path line segment of robot in figure, and s to g is any
One paths are all the combinations of these feasible path line segments.With increasing for barrier, the path from s to g is increasing at double
Plus.It can be seen that path number feasible in Fig. 4 is very more, algorithm finally will select cost most from the various path of this number
That little paths, the line that thick segment is added from s to g means that the final path that Algorithm for Solving goes out, and can solve as seen from Figure 4
The path for going out is optimum.Fig. 5 represents the variation relation of path planning number of times and evolutionary generation.Carry out in this working environment
500 planning, randomly generates due to planning that initial antibodies (path) group is every time, adds the randomness of high frequency closedown and per generation
There is new antibody (path) to add so that the evolutionary generation that every time planning is obtained required for optimal path has uncertainty.
Through to 500 times planning statistics show that average evolutionary generation only had for 248 generations, hence it is evident that less than because search space it is excessive cause into
Change the substantially higher traditional immunization network algorithm of algebraically.An evolutionary generation (245 generation) is have selected from this 500 times planning and is put down
Used as analysis, successive dynasties antibody population is affine in the given current planning of Fig. 6 for the data that evolutionary generation (248 generation) is closer to
Degree represents evolutionary generation with the ever-increasing situation of change of evolutionary generation, abscissa, and vertical coordinate represents affinity.From Fig. 6
As can be seen that the optimum antibody affinity of population is presented progressively increasing trend with the increase of evolutionary generation, although algorithm enters
Changed for 500 generations, but highest affinity has just been reached when evolving to for 245 generation；When optimum affinity reaches peak,
Also there is certain fluctuation always in average affinity, this be due to the multiformity of antibody in order to keep population, per generation have new
Random antibody generate and replace the minimum antibody of equal number affinity in population.
Local paths planning is built upon on the basis of global path planning unknown not true for working environment presence dynamic
Determine the planning of barrier, Fig. 7 shows final planning effect, the direction of motion for setting unknown barrier is unknown, but its along
Straight line does at the uniform velocity one-way movement.It can be seen that the global path cooked up and local path are optimum.In the case of the overwhelming majority,
The complaint message that local environment is included is less than or significantly less than global quantity of information, the local path so from for amount of calculation
Planning be equivalent to environment it is simple when global path planning, its convergence of algorithm speed can be competent at the requirement of real-time.
Claims (3)
1. a kind of mobile robot dynamic path planning method based on Immune network algorithm, it is characterised in that：Being primarily based on can
Sight method carries out pretreatment to global static environment space, cooks up global path using Immune network algorithm afterwards, is advancing
During real-time detection ambient condition information, risk prediction is carried out to the unknown dynamic disorder into sensing range, according to pre-
The result of survey forms new local environment Visual Graph and cooks up new local path using Immune network algorithm, when robot is arrived
Continue in the global path of the preplanning returned to up to after the transient target point of sector planning；
(1), Immune network algorithm realizes process：
Through the improvement to Immune network algorithm and and Visual Graph combination, algorithm concrete implementation step is as follows：
(1) pretreatment is carried out to the environment space of mobile robot operation, builds Visual Graph；
(2) according to the foundation of Visual Graph and analysis, summarize priori and simultaneously extract vaccine, including the antibody that generates or path are not worn
There is no setback in the bounding box of obstacle-overpass thing and the antibody for generating or path；
(3) initialization of population：First vaccination, by be implanted into vaccine to produce m antibody is m paths in search space, road
The generating mode in footpath is progressively to extend to impact point from starting point, is that m paths constitute initial parent population by this m antibodyWherein subscript i represented for the i-th generation, and subscript m represents antibody number；
(4) stop condition：Antibody is finally determined according to the concentration between the fitness and antibody between antibody and antigen
Affinity, calculate current parent for population genetic algebra andIn all antibody affinity, if meeting condition, terminate fortune
Calculate and output result, otherwise continue；
(5) clonal expansion：N clone body is produced respectively to each antibody in above-mentioned population, now amplification is anti-for m*n clone
Body, now antibody population be changed intoN times is all replicated per paths, the path variation of next step more fully will increased per generation anti-
The multiformity of body；
(6) high frequency closedown：Clonal antibody groupIn each clone antibody be intended to experience high frequency closedown obtainThis
In refer to antagonist path and increased or deletion of node；
(7) adaptation value is recalculated：All individualities in antagonist group, including each father's antibody and its variant clone antibody, weight
Newly calculate its affinity；
(8) memory antibody group updates：From each father's antibody and its clonal antibodyIn, select that there is highest affinity
Variant clone antibody, collectively constitute the new elite antibody variants groups with highest affinity of mAnd as more
The candidate of new memory antibody group, then fromWithIn select m affinity highest and constitute new elite antibody variants group
Calculate the average affinity in memory antibody group again afterwards；
(9) immune self-regulation and receptor editing：It is random to generate r (r<M) individual brand-new antibodyReplace now antibody populationMiddle r
The individual antibody with minimum affinity, to simulate Immune System in receptor editing process, increase antibody population multiformity,
Consequently, it is possible to algorithm generates follow-on antibody populationReturn to step (4) afterwards；
(2), the mobile robot active path planning step based on Immune network algorithm：
It is the unknown situation in environmental information part for mobile robot working space, specific path planning step is as follows：
(1) global path planning is carried out to the Immune network algorithm of known static context information application enhancements；
(2) robot advances along the path cooked up；
(3) judge whether to reach impact point, arrival then terminates, be less than and advance up to then continuing on the path cooked up；
Due to the information of the static-obstacle thing in environment space it is previously known, then whenever the step of robot ambulation one will judge machine
Whether the distance between device people and known each static-obstacle thing are less than the radius of the sensing range of scrolling windows, i.e. sensor,
Thus can show whether these static-obstacle things are had occurred in the range of scrolling windows；
(4) judge whether the dyskinesia hinders the normally travel of robot；
(5) predict labelling unknown dynamic barrier can with along the robot advanced of the good path of preplanning following
Certain T moment collides, then the location records at this Unknown Motion barrier T moment are got off, as " static-obstacle thing " place
Reason, it is contemplated that the uncertain factor in the error and actual environment of prediction, expands the region of this " static-obstacle thing ",
It is denoted as deathtrap；
(6) after deathtrap is completely within scrolling windows, robot is by the path of current time scrolling windows preplanning therewith
, used as interim localized target point, the position of current robot is used as starting point, the static-obstacle thing and danger in perceived window for intersection point
New path, path row of the robot along new planning are cooked up as local environmental information using Immune network algorithm in region
Sail, return to step (3).
2. the mobile robot dynamic path planning method based on Immune network algorithm according to claim 1, its feature
It is：(2) robot is first without considering scrolling windows during traveling in (4th) step of step, until sensor finds
New unknown barrier is entered within scrolling windows, i.e., when unknown barrier is less than the radius of scrolling windows with the distance of robot,
The labelling unknown dynamic barrier and start to predict the movement locus of unknown barrier；The result of prediction is divided into following two feelings
Condition：One is the normally travel that unknown barrier will not hinder robot, then the good path row of the preplanning that robot is continued on
Enter, that is, return to step (2)；Two is to predict the normally travel that the dyskinesia may interfere with robot, then go to
(2) (5th) step of step.
3. the mobile robot dynamic path planning method based on Immune network algorithm according to claim 1, its feature
It is：
The design of antibody in Immune network algorithm：
After Visual Graph method is to the process of mobile work robot environment space, constitutes antibody path and only need to consider each barrier
Hinder the element in the set V on thing summit, it is not necessary to any point outside considering in environment space, such hunting zone will significantly subtract
It is few so that efficiency of algorithm is improved；The structure of each node is as follows：
Wherein IDpoint represents node serial number, and IDobject represents the affiliated barrier numbering of node, and x, y represent respectively node
Transverse and longitudinal coordinate；In mobile robot path planning problem, antigen represents problem to be solved i.e. optimal path, and the application sets
It is set to the Euclidean distance of starting point s and impact point g, antibody represents the solution of required problem, that is, the feasible path for searching out, it is by moving
A series of broken line that the starting point of mobile robot, impact point and intermediate nodes are connected into is constituted, and is clearly to describe path to walk
To, through the order of node when the order of element represents moveable robot movement in antibody, antibody adopts string encoding mode,
Its length can change with the number of node, the data structure of antibody enter it is lower shown in：
Wherein Affinity represents the affinity of antibody, evaluates the quality of antibody with affinity here；Fitness represents antibody
Fitness, the antibody with shorter path has higher fitness；Density represents the concentration of antibody, i.e. same antibody institute
The ratio accounted in antibody population, Length represents length of the antibody by s to g, and remainder is the node for constituting path, this storage
Structure when antibody changes, the length of flexibly changing antibody；
The calculating of affinity in Immune network algorithm：
The application adds here the calculating of antibody concentration to suppress and avoiding the appearance of precocious phenomenon, concrete affinity
Computing formula is as follows：
Wherein Affinity (i) represents the affinity of antibody i；Fitness (i) represents the fitness of antibody i, and distance dependent,
The distance in antibody path is shorter, and the fitness of its antibody is higher；Density (i) represents that the concentration of antibody i, i.e. identical resist
Body path is more, and the concentration of antibody is bigger；α is a positive constant, can be seen that fitness is bigger by this formula, affine
Degree is bigger；Concentration is higher, and affinity is less.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410729927.9A CN104407616B (en) | 2014-12-03 | 2014-12-03 | Dynamic path planning method for mobile robot based on immune network algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410729927.9A CN104407616B (en) | 2014-12-03 | 2014-12-03 | Dynamic path planning method for mobile robot based on immune network algorithm |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104407616A CN104407616A (en) | 2015-03-11 |
| CN104407616B true CN104407616B (en) | 2017-04-26 |
Family
ID=52645254
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410729927.9A Expired - Fee Related CN104407616B (en) | 2014-12-03 | 2014-12-03 | Dynamic path planning method for mobile robot based on immune network algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104407616B (en) |
Families Citing this family (32)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109791406B (en) * | 2016-07-06 | 2022-06-07 | 劳伦斯利弗莫尔国家安全有限责任公司 | Autonomous carrier object awareness and avoidance system |
| CN106292673B (en) * | 2016-09-29 | 2019-02-12 | 深圳大学 | A kind of method for optimizing route and system |
| JP6450737B2 (en) | 2016-12-08 | 2019-01-09 | ファナック株式会社 | Robot system |
| CN108268031A (en) * | 2016-12-30 | 2018-07-10 | 深圳光启合众科技有限公司 | Paths planning method, device and robot |
| CN106873600A (en) * | 2017-03-31 | 2017-06-20 | 深圳市靖洲科技有限公司 | It is a kind of towards the local obstacle-avoiding route planning method without person bicycle |
| CA3094649C (en) | 2017-04-12 | 2023-03-28 | David W. Paglieroni | Attract-repel path planner system for collision avoidance |
| CN107168054B (en) * | 2017-05-10 | 2020-11-10 | 沈阳工业大学 | Multi-robot task allocation and path planning method |
| CN107309876B (en) * | 2017-07-14 | 2021-04-13 | 广西师范大学 | Control method for picking by mechanical arm |
| CN107578427B (en) * | 2017-07-31 | 2021-05-18 | 深圳市易成自动驾驶技术有限公司 | Method and device for detecting dynamic obstacle and computer readable storage medium |
| JP7147142B2 (en) * | 2017-09-15 | 2022-10-05 | ソニーグループ株式会社 | CONTROL DEVICE, CONTROL METHOD, PROGRAM, AND MOVING OBJECT |
| CN107728613A (en) * | 2017-09-19 | 2018-02-23 | 海南职业技术学院 | A kind of intelligent robot kinetic control system based on immune algorithm |
| WO2019139815A1 (en) | 2018-01-12 | 2019-07-18 | Duke University | Apparatus, method and article to facilitate motion planning of an autonomous vehicle in an environment having dynamic objects |
| CN108268042A (en) * | 2018-01-24 | 2018-07-10 | 天津大学 | A kind of path planning algorithm based on improvement Visual Graph construction |
| PL3769174T3 (en) | 2018-03-21 | 2022-10-24 | Realtime Robotics, Inc. | Motion planning of a robot for various environments and tasks and improved operation of same |
| CN108549383B (en) * | 2018-05-17 | 2020-06-09 | 电子科技大学 | Real-time multi-sensor community robot navigation method |
| CN112601641B (en) * | 2018-08-23 | 2024-03-08 | 实时机器人有限公司 | Collision detection for robotic motion planning |
| CN109062215A (en) * | 2018-08-24 | 2018-12-21 | 北京京东尚科信息技术有限公司 | Robot and barrier-avoiding method, system, equipment and medium are followed based on its target |
| CN111136651A (en) * | 2018-11-01 | 2020-05-12 | 锥能机器人（上海）有限公司 | Control system, driving device, and method and device for processing operation of driving device |
| CN109341698B (en) * | 2018-11-29 | 2019-12-27 | 深圳市银星智能科技股份有限公司 | Path selection method and device for mobile robot |
| CN110196588A (en) * | 2019-03-28 | 2019-09-03 | 陕西理工大学 | Method for planning path for mobile robot based on networks decomposition |
| US11966852B2 (en) | 2019-12-11 | 2024-04-23 | Shanghai United Imaging Intelligence Co., Ltd. | Systems and methods for situation awareness |
| CN111026115A (en) * | 2019-12-13 | 2020-04-17 | 华南智能机器人创新研究院 | Robot obstacle avoidance control method and device based on deep learning |
| CN111142529B (en) * | 2019-12-31 | 2023-06-09 | 达闼机器人股份有限公司 | Virtual wall decision method and device and robot |
| CN111337042B (en) * | 2020-03-13 | 2021-11-02 | 湖北大学 | Vehicle path planning method and system |
| CN111283688B (en) * | 2020-03-23 | 2021-08-10 | 深圳市科瑞技术科技有限公司 | Obstacle avoidance method of robot and robot equipment |
| US11927972B2 (en) | 2020-11-24 | 2024-03-12 | Lawrence Livermore National Security, Llc | Collision avoidance based on traffic management data |
| RU2749202C1 (en) * | 2020-12-07 | 2021-06-07 | Общество с ограниченной ответственностью «РобоСиВи» | Method for planning robot movement and mobile robot |
| CN113009918B (en) * | 2021-03-09 | 2023-12-05 | 京东鲲鹏(江苏)科技有限公司 | Path planning method, device, system and readable storage medium |
| CN113721615B (en) * | 2021-08-27 | 2023-08-15 | 广州大学 | Navigation path planning method and system based on machine vision |
| CN114199266A (en) * | 2021-11-25 | 2022-03-18 | 江苏集萃智能制造技术研究所有限公司 | Path planning method for occupied target based on diagnosis guide service robot |
| CN114415665A (en) * | 2021-12-17 | 2022-04-29 | 新疆钵施然智能农机股份有限公司 | Algorithm for obstacle avoidance path planning |
| CN114460933B (en) * | 2021-12-30 | 2023-11-03 | 南京理工大学 | Dynamic environment-oriented mobile robot local path planning algorithm |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN100468265C (en) * | 2007-08-24 | 2009-03-11 | 北京航空航天大学 | Combined type vision navigation method and device |
| CN101887271A (en) * | 2010-07-19 | 2010-11-17 | 东莞职业技术学院 | Mobile robot path planning method |
| CN103412490B (en) * | 2013-08-14 | 2015-09-16 | 山东大学 | For the polyclone Algorithm of Artificial Immune Network of multirobot active path planning |
| CN103529843B (en) * | 2013-10-17 | 2016-07-13 | 电子科技大学中山学院 | Lambda path planning algorithm |
-
2014
- 2014-12-03 CN CN201410729927.9A patent/CN104407616B/en not_active Expired - Fee Related
Also Published As
| Publication number | Publication date |
|---|---|
| CN104407616A (en) | 2015-03-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104407616B (en) | Dynamic path planning method for mobile robot based on immune network algorithm | |
| Sadeghian et al. | Sophie: An attentive gan for predicting paths compliant to social and physical constraints | |
| CN110737968B (en) | Crowd trajectory prediction method and system based on deep convolutional long and short memory network | |
| CN107607120B (en) | Unmanned aerial vehicle dynamic track planning method based on improved restoration type Anytime sparse A algorithm | |
| CN106444769A (en) | Method for planning optimal path for incremental environment information sampling of indoor mobile robot | |
| CN105717929A (en) | Planning method for mixed path of mobile robot under multi-resolution barrier environment | |
| CN108153867B (en) | User trajectory prediction technique and device based on temporal regularity | |
| CN110412877A (en) | A kind of carrier-borne aircraft deck path planning method for optimally controlling based on NSP algorithm | |
| CN106714110A (en) | Auto building method and system of Wi-Fi position fingerprint map | |
| CN104121903B (en) | A kind of rolling Route planner based on boundary value problem | |
| US9460540B2 (en) | Method for animating characters, with collision avoidance based on tracing information | |
| CN110009150B (en) | The compact reservoir physical parameter intelligent Forecasting of data-driven | |
| CN106774327A (en) | A kind of robot path planning method and device | |
| Sarmady et al. | Modeling groups of pedestrians in least effort crowd movements using cellular automata | |
| CN102057368A (en) | Distribution of properties in a 3D volumetric model using a maximum continuity field | |
| Bruneau et al. | Energy-efficient mid-term strategies for collision avoidance in crowd simulation | |
| Huang et al. | Multimodal trajectory prediction: A survey | |
| CN110530373A (en) | A kind of robot path planning method, controller and system | |
| CN113391633A (en) | Urban environment-oriented mobile robot fusion path planning method | |
| Li et al. | Dimensionally aware multi-objective genetic programming for automatic crowd behavior modeling | |
| Ning‐bo et al. | Destination and route choice models for bidirectional pedestrian flow based on the social force model | |
| CN113515131A (en) | Mobile robot obstacle avoidance method and system based on condition variation automatic encoder | |
| Bruneau et al. | Eacs: Effective avoidance combination strategy | |
| US20240176989A1 (en) | Trajectory predicting methods and systems | |
| CN113033954A (en) | Intelligent decision-making method for coordinated development of coal and coalbed methane |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20170426 Termination date: 20181203 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |