CN108802741A - Mobile robot sonar data fusion method based on DSmT theories - Google Patents
Mobile robot sonar data fusion method based on DSmT theories Download PDFInfo
- Publication number
- CN108802741A CN108802741A CN201810649501.0A CN201810649501A CN108802741A CN 108802741 A CN108802741 A CN 108802741A CN 201810649501 A CN201810649501 A CN 201810649501A CN 108802741 A CN108802741 A CN 108802741A
- Authority
- CN
- China
- Prior art keywords
- sonar
- dsmt
- reliability
- mobile robot
- data
- 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
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S15/00—Systems using the reflection or reradiation of acoustic waves, e.g. sonar systems
- G01S15/88—Sonar systems specially adapted for specific applications
- G01S15/93—Sonar systems specially adapted for specific applications for anti-collision purposes
- G01S15/931—Sonar systems specially adapted for specific applications for anti-collision purposes of land vehicles
Landscapes
- Engineering & Computer Science (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Physics & Mathematics (AREA)
- Acoustics & Sound (AREA)
- Computer Networks & Wireless Communication (AREA)
- General Physics & Mathematics (AREA)
- Measurement Of Velocity Or Position Using Acoustic Or Ultrasonic Waves (AREA)
- Control Of Position, Course, Altitude, Or Attitude Of Moving Bodies (AREA)
Abstract
The present invention proposes a kind of mobile robot sonar data fusion method based on DSmT theories, solves the problems, such as the fusion of the sonar data of acquisition in a conflict situation, realizes that step is：Information is obtained by sonar sensor, establishes sonar sensor measurement model；Sonar detection sector region is divided into according to measurement model：Depletion region and region can be occupied；By the two dimensional surface environment rasterizing of mobile robot, the differentiation frame of grid is provided, establishes the reliability assignment computation model differentiated under frame；The DSmT hybrid combining rules for building the different moments sonar data under constraints, the fusion of mobile robot sonar data is completed according to the DSmT hybrid combining rules of different moments sonar data.DSmT theories have been applied in the sonar data fusion of mobile robot by the present invention, solves the problems, such as the fusion of colliding data, so that it is more accurate to the judgement of trellis states could around robot, it can be used for map building of the mobile robot in circumstances not known in practical application.
Description
Technical field
The invention belongs to field of computer technology, are related to data fusion, specifically a kind of moving machine based on DSmT theories
Device voice receives data fusion method, can be used for map building of the mobile robot in circumstances not known in practical application.
Background technology
Using intelligent mobile robot detection unknown complex environment be always domestic and international roboticist research hot spot and
Difficult point project.Map building is then a kind of form of expression of the mobile robot to environment sensing, in circumstances not known, mobile machine
For people by the sensor loaded in ontology, such as sonar, laser is infrared, and the sensors such as vision obtain the information of ambient enviroment, and
Information is recombinated and is merged, the profile of ambient enviroment is then sketched the contours of or image and mobile robot itself is determined
Position.And sonar sensor is due to its cheap price, the advantages that easy occupation mode, convenient data processing, is often used to make
For the important perceptron of mobile robot.Due to the limitation of sensor itself, the data that sensor provides generally comprise largely
Unascertained information, these information are often imperfect, inaccurate, fuzzy, sometimes even contradictory, mistake, directly
Map building is carried out using perception information and hardly results in accurate environmental model, therefore usually requires to reprocess perception information,
Accurate environmental information is obtained by the fusion of multi-sense.
Paper " the mobile robot three-dimensional environment Modeling Technique Research merged based on information " (Beijing that Zhang Qin is delivered at it
University of Post and Telecommunication, Ph.D. Dissertation, 2013.1) in using the method for D-S evidence theory to the sonar sensor of mobile robot
Data are merged, and two dimensional surface grating map in circumstances not known is established.Trellis states could is divided by this method：Unreachable region,
Region (determination has barrier or clear) and zone of ignorance are determined, by the composition rule of D-S evidence theory to sonar number
According to being merged.This method is disadvantageous in that：The composition rule of used D-S evidence theory will merge the coke that conflicts under frame
The reliability of member is evenly distributed in all proposition elements, but according to D-S compositional rules, when two data conflict completely,
It can not be synthesized with the rule, when two evidence height conflict, carry out synthesis with the rule and may result in reality often
Manage the result runed counter to.
A kind of patent " mobile robot grating map creating method of real time data fusion " (patent of Hunan University's application
Application number：CN200810143537.8, publication number：CN101413806 it is disclosed in) a kind of using neural network and Bayes's reason
The sonar data fusion method of opinion.This method extracts three sonar sensors nearest with current computation grid cell distance same
The measured value at one moment, as the input of neural network, the input of neural network is the state of grid：It is idle, occupied and not
It determines state, finally uses the state of Bayes rule update grid.Deficiency is existing for the method for the patent application publication：God
Determination through network parameter needs the support of great amount of samples data, and the precision of parameter is affected to output result, fusion knot
The reliability of fruit can also be affected.
During the establishment map of mobile robot, the prior art is for that in the fusion treatment of sonar data, cannot solve
The certainly fusion problem of colliding data, and operand does not reach greatly peak efficiency, the precision of parameter also affects output as a result, shadow
The reliability of map building result is rung.
Invention content
It is an object of the invention to overcome the shortcomings of above-mentioned prior art, it is proposed that one kind is more accurately and reliably based on
The mobile robot sonar data fusion method of DSmT theories.
The present invention is a kind of mobile robot sonar data fusion method based on DSmT theories, which is characterized in that including
There are following steps：
(1) information is obtained by sonar sensor, establishes the measurement model of sonar sensor：
Circumstances not known where mobile robot is two dimensional surface environment, passes through mobile machine in two dimensional surface environment
The sonar sensor in multiple and different directions that human body loads obtains the information of ambient enviroment, and establishes the measurement of sonar sensor
Model：The sound wave angle of departure by sonar and maximum measurement distance, constitute sonar detection sector region range, if the survey that sonar returns
Magnitude is the distance measure away from the nearest target of sonar range in the sector region；
(2) according to the measurement model of sonar sensor, sonar detection sector region range is divided into：Depletion region and can
Occupy region：
(2a) depletion region：[0, R- ε), it is 0 there are the probability of barrier in the area, wherein R refers to that sonar returns
Measurement distance, ε refers to the measurement error of sonar；
(2b) can occupy region：[R- ε, R+ ε] is in the area 0 there is no the probability of barrier；
(3) rasterizing is carried out to the two dimensional surface environment residing for mobile robot, provides the differentiation frame of trellis states could, builds
Found the computation model of the reliability assignment under the differentiation frame：
(3a) carries out rasterizing to the two dimensional surface environment residing for mobile robot, and each grid represents 80*80cm's
Space size judges the state of grid for three kinds：Sky has barrier, is unknown；
(3b) provides the differentiation frame Ω of trellis states could according to DSmT theories：Ω={ E, O }, wherein E represents sky, O generations
Table has barrier；
(3c) establishes grid around mobile robot and exists according to the measurement model of sonar sensor and the division of measured zone
Differentiate the reliability assignment computation model under frame Ω={ E, O }, according to the region where grid, reliability assignment computation model is divided into
The reliability assignment computation model of the depletion region of sonar to measure, the reliability assignment computation model for occupying region of sonar to measure and
It has been more than the computation model for occupying region of sonar to measure, DSmT theories are established on the basis of ultrapower collection, reliability assignment meter
The reliability assignment that model specifically includes each element concentrated by the ultrapower that differentiation frame is constituted is calculated, the reliability of each element is assigned
The value range of value is [0,1], and it is 0 to require ultrapower to concentrate the reliability assignment value of this element of empty set, and ultrapower is concentrated all
The sum of reliability assignment of element is 1；
(4) the DSmT hybrid combining rules for establishing the different moments sonar data under constraints, according to different moments
The DSmT hybrid combining rules of sonar data complete the fusion of sonar data.
The present invention merges the sonar sensor measurement data of different moments by DSmT theories, judges mobile machine
The state of grid around people：Clear has barrier and unknown, successfully solves the fusion of sonar data in a conflict situation
Problem.
Compared with prior art, the present invention having the following advantages that：
First, the present invention merges the sonar sensor measurement data of different moments by DSmT theories, judges to move
The state of grid around mobile robot sets clear, has barrier and unknown three kinds of situations, DSmT theories to pass through reservation
The combination brief inference of the focal element of conflict, has been given the fusion reliability of unknown state by coke member of the evidences conflict item as data fusion,
Problem is merged to reach to the effective of trellis states could so as to well solve the information occurred under high conflict situations in evidence
Identification.
Second, the present invention carries out fusion treatment by DSmT theories to sonar data, reduces the calculating in fusion process
Amount, and the reliability of fusion results is not influenced.
Description of the drawings
Fig. 1 is the implementation flow chart of the present invention；
Fig. 2 is the sonar sensor measurement model that the present invention uses；
Fig. 3 is that the measured zone for the sonar that the present invention uses divides；
Fig. 4 is the trellis states could reliability assignment computation model based on sonar sensor that the present invention uses；
Fig. 5 is the combination for multiple evidences that the present invention uses；
Fig. 6 is the angle setting of sonar sensor in front of mobile robot in emulation experiment of the present invention.
Specific implementation mode
Below in conjunction with the drawings and specific embodiments, to the detailed description of the invention.
Embodiment 1
In the map building based on sonar, since the uncertainty of sonar information is stronger, multiple sonar sensors are needed
Information is obtained in different periods, and multiple sonar information are merged.At this stage, the application of information fusion technology is main
It is in military field.Modern war has been extensively using various high-tech, it is necessary to use multisensor and multi-source information system,
More battle field informations, data processing amount and processing capacity can be obtained considerably beyond single sensor, with radar, it is infrared,
The sensor such as photoelectricity and quantity are continuously increased, and multi-sensor information fusion technology is able to extensive use.In addition, information is melted
Conjunction technology is gradually permeated to multi-field.In mobile robot field, the application range of information fusion technology is increasingly extensive,
But there are still many problems, cannot such as complete the processing to colliding data, for example, when the data of two different sensors acquisitions
Conflicting in the presence of height, when even completely opposite, it is believed that this is colliding data, during mobile robot map-building, by
In the factor of sonar sensor itself, there is colliding datas in the information that sonar sensor is got, therefore obtained map
Data precision is also to be improved, for this purpose, the present invention expands research and innovation, proposes a kind of mobile machine based on DSmT theories
Voice receives data fusion method, it may also be said to be that the sonar data based on DSmT theories is melted in a kind of mobile robot map-building
Conjunction method successfully solves the problems, such as the fusion to colliding data, and referring to Fig. 1, the present invention includes following steps：
(1) information is obtained by sonar sensor, according to the operation principle of sonar sensor, establishes the survey of sonar sensor
Measure model：
Circumstances not known where mobile robot is two dimensional surface environment, passes through mobile machine in two dimensional surface environment
The sonar sensor in multiple and different directions that human body loads obtains the information of ambient enviroment, according to the basic work of sonar sensor
Make principle, and establish the measurement model of sonar sensor, referring to Fig. 2：The sound wave angle of departure by sonar and maximum measurement distance, structure
At sonar detection sector region range, if the measured value that sonar returns be in the sector region away from the nearest target of sonar range away from
From measured value.
(2) according to the measurement model of sonar sensor, sonar detection sector region range is divided into：Depletion region and can
Occupy region.It is sensor beam width, the as angle of departure of sound wave referring to Fig. 2, ψ, ε is sensor measurement errors, and P points are barrier
It is actual distance of the target P points to sonar to hinder object target, ρ, and angle [alpha] isIt is with the angle between sonar central axes (x-axis), i.e., incident
Angle, R are measured value of the sonar sensor to P point barriers, and for the measurement distance R that sonar returns, it is sonar referring to Fig. 3, ε
Measurement error, ψ be sonar the angle of departure.Exist between the measurement data and real data that are got due to sonar sensor and misses
Difference, the present invention just consider into error when designing a model, and more fully consider actual conditions.
(2a) defines depletion region：[0, R- ε), it is 0 there are the probability of barrier in the area referring to Fig. 3.
(2b) definition can occupy region：[R- ε, R+ ε], referring to Fig. 3, in the area, the probability there is no barrier is
0。
(3) rasterizing is carried out to the two dimensional surface environment residing for mobile robot, provides the differentiation frame of trellis states could, builds
Found the computation model of the reliability assignment under the differentiation frame.Referring to Fig. 4, using sonar central axes as x-axis, sonar present position is original
Point o, ψ are the angle of departure of sonar；OP1 is the measurement distance value that sonar returns, and OP2=OP1+ ε, OP3=OP1- ε, ε are sensor
Measurement error.After carrying out rasterizing to two dimensional surface environment, improves the search efficiency of state space and maintain the essence of data
Degree.
(3a) carries out rasterizing to the two dimensional surface environment residing for mobile robot, and each grid represents 80*80cm's
Space size, according to robot sonar sensor measurement data, the present invention judges the state of grid for three kinds：It is empty, have barrier,
It is unknown, wherein dummy status is clear state.
(3b) provides the differentiation frame Ω of trellis states could according to DSmT theories：Ω={ E, O }, wherein E represents sky, O generations
Table has barrier, and trellis states could is unknown state when not only may be E but also may be O, in DSmT theories, by mutual exclusive base
The perfect set of this proposition composition is collectively referred to as differentiating frame, indicates the be possible to answer to a certain problem, but only one of which
Answer is correct.
(3c) establishes grid around mobile robot and exists according to the measurement model of sonar sensor and the division of measured zone
Differentiate the reliability assignment computation model under frame Ω={ E, O }, according to the region where grid, reliability assignment computation model is divided into
The reliability assignment computation model of the depletion region of sonar to measure, the reliability assignment computation model for occupying region of sonar to measure and
It has been more than the computation model for occupying region of sonar to measure, DSmT theories are established on the basis of ultrapower collection, reliability assignment meter
The reliability assignment that model specifically includes each element concentrated by the ultrapower that differentiation frame is constituted is calculated, the reliability of each element is assigned
The value range of value is [0,1], and it is 0 to require ultrapower to concentrate the reliability assignment value of this element of empty set, and ultrapower is concentrated all
The sum of reliability assignment of element is 1.
That is, reliability assignment computation model is fallen according to grid around mobile robot in the spacious area of sonar to measure
Domain, fall sonar to measure occupy region, be more than sonar to measure occupy three kinds of region situation depending on, reliability assignment
The value for being each element of ultrapower collection that is made of differentiation frame on [0,1] section, and it is 0 to require the value of empty set, institute
Have element value and be 1.
(4) establish the DSmT hybrid combining rules of different moments sonar data under constraints, the present invention by some
Grid portion is occupied by barrier, is partly empty situation, i.e. E ∩ O have directly been classified as barrier situation, therefore, setting E ∩ O
=Φ is constraints, and the fusion of sonar data is completed according to the DSmT hybrid combining rules of different moments sonar data.
The present invention merges the sonar sensor measurement data of different moments by DSmT theories, and DSmT theories pass through
Retain coke member of the evidences conflict item as data fusion, solve the problems, such as that the information fusion under high conflict situations occurs in evidence,
And the mixing fusion rule that DSmT theories are utilized under certain constraints, reduces the calculation amount in fusion process, and
The reliability of fusion results is not influenced.
Embodiment 2
With embodiment 1, the foundation described in step 4 exists mobile robot sonar data fusion method based on DSmT theories
The DSmT hybrid combining rules of different moments sonar data under constraints refer to, more by being loaded by mobile robot ontology
The trellis states could that the sonar sensor of a different directions is got it is empty, have the combination reliability under barrier and unknown situation common
Form new combination reliability, and by new combination reliability and non-empty characteristic functionMultiplication obtains the reliability assignment of grid.
The classical rule of combination of DSmT theories is without any constraints, but in practical evident information fusion process, greatly
Mostly it is Prescribed Properties, DSmT theory hybrid combining rules just solve the problems, such as this, its Prescribed Properties collection, more adapt to real
The needs of border fusion.Cartographic information is the set of all trellis states could information, and letter is carried out to trellis states could around mobile robot
Assignment is spent, the formulation of the construction and search strategy of map state space is simplified, can avoid searching present in conventional search algorithm
Rope multiple shot array problem need not also carry out complicated operation, several from completeness, time complexity, space complexity, optimality
It can achieve the effect that for a criterion satisfied.
Embodiment 3
Mobile robot sonar data fusion method based on DSmT theories is with embodiment 1-2, the foundation described in step 4
The DSmT hybrid combining rules of different moments sonar data under constraints, specific reliability fusion results and each combination
Rule calculation formula be respectively：
Wherein, m1() and m2() is the basic reliability assignment from two different data sources, mμ(Ω) (C) indicates band
The reliability fusion results of Prescribed Properties, μ are constraints E ∩ O=Φ, and wherein Φ is absolute empty set, C ∈ DΩ,It is non-
Empty characteristic function, ifThenValue is ' 0 ', is otherwise ' 1 ',For constraint condition set, Φ indicates exhausted
To empty set,Indicate opposite empty set condition.In above formula, S1(C) it is the DSmT rules of combination without any constraints, S2
(C) it is the item that the combination reliability assignment of all absolute empty sets and opposite empty set is allocated to total unknown collection and relatively unknown collection
Rule of combination under part, total unknown collection refers to the union for differentiating all propositions in frame, S3(C) it is to believe the combination of the focal element of conflict
Spend the rule of combination under conditions of being assigned in the union of the focal element of conflict.They three collectively constituted DSmT hybrid combinings rule
Then, reliability fusion results m has been obtainedμ(Ω)(C)。
The present invention DSmT hybrid combinings rule by the basic reliability assignment from different data sources, under constraints into
Row repeatedly fusion, has obtained the authentic and valid result for being more in line with actual conditions.
Embodiment 4
Mobile robot sonar data fusion method based on DSmT theories is with embodiment 1-3, the basis described in step 4
The DSmT hybrid combining rules of different moments sonar data complete the fusion of sonar data, and DSmT theories establish the base in ultrapower collection
On plinth, including have the following steps：
(4a) trellis states could differentiates the ultrapower collection D under frame ΩΩFor：DΩ={ Φ, E, O, E ∪ O, E ∩ O }, setting E ∩ O
=Φ is constraints, and wherein Φ is absolute empty set, DΩPass through union to be added by the proposition element in differentiation frame Ω：∪
It is calculated with shipping：The set for the combination of sentences that ∩ is formed.
(4b) at constraints E ∩ O=Φ, using DSmT hybrid combinings rule by preceding t-1 moment trellis states could reliability
Fusion results, merged with the grid reliability assignment calculated according to current t moment sonar to measure data, complete mobile machine
The fusion of the sonar data for the ambient condition information that people is obtained in the two dimensional surface of circumstances not known by sonar sensor.
Obtained robot sonar sensor measurement data is walked by each emulation, calculates the robot of each emulation step
The reliability assignment of surrounding trellis states could, and the reliability assignment of different emulation steps is merged, have to trellis states could to reach
Effect identification.
Embodiment 5
Mobile robot sonar data fusion method based on DSmT theories is with embodiment 1-4, the use described in (4b)
DSmT hybrid combinings rule is by the fusion results of preceding t-1 moment trellis states could reliability, and according to current t moment sonar to measure data
The grid reliability assignment of calculating is merged, and has been specifically included：
Wherein, mij(Ω) (C) indicates the i-th row, the reliability fusion results of jth row grid, C ∈ DΩ,For non-empty feature
Function, C are the state of grid, and the constraints of fusion is：E ∩ O=Φ；C values are respectively ultrapower collection DΩIn Φ, O, E and E
4 kinds of state reliabilities of grid when ∪ O；S1(C) be free DSmT models rule of combination, free DSmT models refer to directly by
Ultrapower collection space DΩThe Fusion Model constituted, without any constraints；S2(C) by all absolute empty sets and sky relatively
The combination reliability assignment of collection is allocated to total unknown collection and relatively unknown collection, and total unknown collection refers to differentiating all propositions in frame
Union, such as：Differentiate that frame Ω={ E, O }, total unknown collection are：E ∪ O, relatively unknown collection are：Differentiate that frame mid portion divides proposition
Union；S3(C) by the combination reliability of the focal element of conflict, such as：M (E ∩ O), has been assigned in the union of the focal element of conflict：m(E∪O)；Melt
Close calculating process and result mij(Ω) (C) is as follows：
(1) C=Φ, then mij(Φ)=0.
(2) C=O has barrier state：
Non-empty characteristic functionThen：mij(O)=S1(O)+S2(O)+S3(O), wherein：
S2(O)=0, S3(O)=0.
(3) C=E, i.e. dummy status：
Non-empty characteristic functionThen：mij(E)=S1(E)+S2(E)+S3(E), wherein：
S2(E)=0, S3(E)=0.
(4) C=E ∪ O, i.e. unknown state：
Non-empty characteristic functionThen：mij(E ∪ O)=S1(E∪O)+S2(E∪O)+S3(E ∪ O), wherein：
S2(E)=0,
As can be seen that by the combination reliability of the focal element of conflict in (4), i.e.,：
It has been distributed on the fusion reliability of unknown state.
In the calculating process that 4 kinds of state reliabilities of above-mentioned grid merge,Indicate the lattice-shaped at the t-1 moment
The fusion results of state reliability,Indicate the trellis states could reliability assignment that the sonar to measure data according to t moment calculate.
The reliability assignment of trellis states could around robot to calculating each emulation step by DSmT hybrid combining rules
Process carries out exhaustive division and calculates description, covers all situations in practice.
Embodiment 6
For mobile robot sonar data fusion method based on DSmT theories with embodiment 1-5, the present invention is that one kind is based on
The mobile robot sonar data fusion method of DSmT theories includes following steps referring to Fig. 1：
(1) according to the operation principle of sonar sensor, the measurement model of sonar sensor is established, referring to Fig. 2, sonar sensing
Device is most common a kind of sensor in mobile robot.Its operation principle is exactly by transmitter transmitting ultrasonic listening letter
Number, then by receiver receive the reflected signal of barrier, according to transmitting and receive time difference come calculate sensor with
The distance between barrier.Due to cheap, easy to use, sonar is used widely in mobile robot.Sound
Sensor of receiving detects object by emitting the sound wave of a branch of taper, and receives back wave to calculate the distance of object.By
In sound wave, there are certain angles of departure, and the range of sonar detection is sector region, and the object detected be sector region in away from
The nearest object of sonar range.In Fig. 2, for coordinate system using sonar central axes as x-axis, sonar present position is origin, the measurement mould
Type includes：
● ψ is sensor beam width, the as angle of departure of sound wave；
● ε is sensor measurement errors；
● P points are obstacle target, and ρ is actual distance of the target P points to sonar；
● angle [alpha] isWith the angle between sonar central axes (x-axis), i.e. incidence angle；
● R is measured value of the sonar sensor to P point barriers.
(2) according to the measurement model of sonar sensor, the measured zone range of sonar is further divided, as shown in figure 3,
R is the measurement distance that sonar returns, and is the measurement error of sonar, and ψ is the angle of departure of sonar.The measurement range of sonar is：ψ emits
In angle, and in the maximum measurement range in sonar.And for sonar return measurement distance R, by the measurement range of sonar into
One step is divided into depletion region and can occupy region：
(2a) depletion region：[0, R- ε), it is 0 there are the probability of barrier in the area.
(2b) can occupy region：[R- ε, R+ ε] is in the area 0 there is no the probability of barrier.
(3) rasterizing is carried out to the two dimensional surface environment residing for mobile robot, provides the differentiation frame of trellis states could, builds
Found the computation model of the reliability assignment under the differentiation frame：
(3a) carries out rasterizing to the two dimensional surface environment residing for mobile robot, flat to the two dimension residing for mobile robot
Face ring border carries out rasterizing, refers to that the environment that will be described is divided into many grids with certain size size, and passes through grid
Trellis state indicates wherein to there is a possibility that barrier.The size of grid is identical under normal circumstances, but can also basis
Actual applicable cases use the grid of mutative scale, and the size of lattice dimensions just determines the precision of map.Grating map is not
Only intuitive is strong, and is easy to create and safeguard, for other opposite maps, grating map can preferably describe environment.Cause
And its in mobile robot map-building using more.Each grid represents the space size of 80*80cm, as shown in figure 4,
Using sonar central axes as x-axis, sonar present position is the angle of departure that origin o, ψ are sonar；OP1 be sonar return measurement away from
From value, OP2=OP1+ ε, OP3=OP1- ε, ε are sensor measurement errors.Judge the state of grid for three kinds：It is empty (accessible
Object), have barrier, be unknown.
(3b) provides the differentiation frame of trellis states could according to DSmT theories：Ω={ E, O }, wherein E represents sky, and O is represented
There is barrier.
(3c) establishes grid around mobile robot and exists according to the measurement model of sonar sensor and the division of measured zone
Differentiate the reliability assignment computation model under frame Ω={ E, O }, according to the region where grid, reliability assignment computation model is divided into
The reliability assignment computation model of the depletion region of sonar to measure, the reliability assignment computation model for occupying region of sonar to measure and
It has been more than the computation model for occupying region of sonar to measure, DSmT theories are established on the basis of ultrapower collection, reliability assignment meter
The reliability assignment that model specifically includes each element concentrated by the ultrapower that differentiation frame is constituted is calculated, the reliability of each element is assigned
The value range of value is [0,1], and it is 0 to require ultrapower to concentrate the reliability assignment value of this element of empty set, and ultrapower is concentrated all
The sum of reliability assignment of element is 1；Reliability assignment of the grid in the case where differentiating frame Ω={ E, O } around mobile robot is calculated,
Refer to using sonar central axes as x-axis, sonar present position is the angle of departure that origin o, ψ are sonar；OP1 is the survey that sonar returns
Distance value is measured, OP2=OP1+ ε, OP3=OP1- ε, ε are sensor measurement errors, are specifically included：
(3c1) for the grid in sonar transmission angle ψ ranges, the calculating of state reliability assignment is specifically, using d tables
Show the distance (unit cm) of certain grid centre distance sonar：
1. as d < OP1- ε, i.e., the grid falls the depletion region in sonar to measure, then the depletion region of sonar to measure
Reliability assignment computation model be：
2. as OP1- ε < d < OP1+ ε, i.e., the grid is fallen in the region that occupies of sonar to measure, then sonar to measure
The reliability assignment computation model that region can be occupied is：
3. as d > OP1+ ε, i.e., the grid has been over the region that occupies of sonar to measure, not in our discussion
In range, without calculating reliability assignment.
(3c2) represents the space size of 80*80cm due to grid, and d is the distance of grid centre distance sonar, as d >
When OP1+ ε, i.e. what the central point of the grid had been over sonar to measure occupies region.But the central point of grid can not account for
According to region, region can not be being occupied without it can be shown that grid entirely, perhaps the part of grid can occupy region.Therefore, in order to
Judge whether grid portion can occupy region, the judgement of distance is retracted 50 centimetres, further compares the pass of d-50 and OP1+ ε
System：
1. as d-50 < OP1- ε, it is in the case where occupying region of sonar to measure for grid portion, then sonar to measure
The reliability assignment computation model of depletion region be：
2. as OP1- ε < d-50 < OP1+ ε, it is in the case where occupying region of sonar to measure for grid portion, then
The reliability assignment computation model for occupying region of sonar to measure is：
3. as d-50 > OP1+ ε, grid is not on the region that occupies of sonar to measure, then has been more than sonar to measure
The computation model that region can be occupied is：
(4) rule of combination of the different moments sonar data under constraints is established：
(4a) trellis states could differentiates that the ultrapower collection under frame is：DΩ={ Φ, E, O, E ∪ O, E ∩ O } concentrates ultrapower
E ∪ O indicate unknown state, some grid portion is occupied by barrier, is partly empty situation, is i.e. E ∩ O are directly classified as
There is barrier situation, therefore, sets E ∩ O=Φ as constraints.Obtained sonar sensor is walked according to each emulation to survey
Data are measured, reliability assignment of the grid in the case where differentiating frame Ω={ E, O } around mobile robot is calculated.
(4b) at constraints E ∩ O=Φ, using DSmT hybrid combinings rule by preceding t-1 moment trellis states could reliability
Fusion results, merged with the grid reliability assignment calculated according to current t moment sonar to measure data.In DSmT theory frames
Under frame, the combination calculating of multiple evidences can be obtained with the calculating recursion of the combination of two evidences, specific as shown in Figure 5, wherein
M1, m2...mn indicate the state reliability assignment of the same grid at t1, t2 ..., tn moment respectively.
The present invention proposes the modeling method to sonar to measure under DSmT theoretical frames, and uses DSmT blending algorithms pair
The information that multiple sonars obtain on mobile robot ontology is merged, and the implementation for realizing two-dimensional environment map creates, and overcomes
The limitation of prior art, also provides new method and thinking for the map building of robot and navigation.
The technique effect of the present invention is explained again below by the experiment and result of emulation.
Embodiment 7
Mobile robot sonar data fusion method based on DSmT theories is with embodiment 1-6, and the present invention is with Pioneer 2
Mobile robot is reference, and emulation generates the measurement data of 16 sonar sensors, 8 sensings respectively before mobile robot
The measurement data of device and the below measurement data of 8 sensors.Angle configurations of the sonar sensor in mobile robot, to move
In front of mobile robot for the setting of 8 sonar sensors, as shown in fig. 6, the coordinate system (Xt Ot Zt) in Fig. 6 is moving machine
Device people's coordinate system, respectively at clockwise and anticlockwise 90 ° of mobile robot, 50 °, 30 °, 10 ° of this 8 orientation, which are provided with, to be passed
8 sensors are arranged in sensor, mobile robot front altogether, can cover all angular ranges, and 8, robot rear sonar passes
The angle configurations of sensor and front configuration are symmetrical.
Sonar sensor parameter is as shown in the table：
1 sonar sensor major parameter of table
The measurement data of 16 sonar sensors is obtained in emulation, the sample size of acquisition is big, the measurement of sonar sensor
Ranging from 0-300cm obtains the target object data within the scope of relatively large distance, and the angle of departure of sonar sensor is 150, each
Sonar sensor obtains the target object data within the scope of larger angle.
According to emulation experiment, in the sonar sensor measurement model of the present invention, measurement error 10cm, error is smaller,
Influence of the precision of parameter to output result is reduced, improves the data precision of output result, the result obtained after fusion is more
Add it is correct reliable, it is more accurate to the judgement of the circumstances not known residing for mobile robot, create the ground for being more in line with actual conditions
Figure.
In conclusion a kind of mobile robot sonar data fusion method based on DSmT theories proposed by the present invention, purport
Multiple sonar sensor measurement data that obtained robot body loads are being walked by each emulation, are calculating mobile machine
Reliability assignment of the grid in the case where differentiating frame Ω={ E, O } around people, under DSmT theoretical frames, to the reliability of different emulation steps
Assignment is merged, and solves the problems, such as the fusion of colliding data so that more accurate to the judgement of trellis states could around robot.
Realize that step is：(1) information is obtained by sonar sensor, establishes the measurement model of sonar sensor；(2) it is sensed according to sonar
Sonar detection sector region range is divided by the measurement model of device：Depletion region and region can be occupied；(3) to mobile machine
Two dimensional surface environment residing for people carries out rasterizing, provides the differentiation frame of trellis states could, establishes the reliability under the differentiation frame
The computation model of assignment；(4) the DSmT hybrid combining rules for establishing the different moments sonar data under constraints, according to not
The DSmT hybrid combining rules of sonar data complete the fusion of mobile robot sonar data in the same time.The present invention is melted with DSmT
Hop algorithm merges the information that multiple sonars obtain on mobile robot ontology, realizes the implementation wound of two-dimensional environment map
It builds, overcomes prior art limitation in data applicability and data precision, more to the judgement of trellis states could around robot
It is accurate, can be used for map building of the mobile robot in circumstances not known in practical application, have broad application prospects.
Claims (5)
1. a kind of mobile robot sonar data fusion method based on DSmT theories, which is characterized in that include following steps：
(1) information is obtained by sonar sensor, establishes the measurement model of sonar sensor：
Circumstances not known where mobile robot is two dimensional surface environment, passes through mobile robot sheet in two dimensional surface environment
The sonar sensor in multiple and different directions that body loads obtains the information of ambient enviroment, establishes the measurement model of sonar sensor：
The sound wave angle of departure by sonar and maximum measurement distance, constitute sonar detection sector region range, if the measured value that sonar returns
For the distance measure away from the nearest target of sonar range in the sector region；
(2) according to the measurement model of sonar sensor, sonar detection sector region range is divided into：It depletion region and can occupy
Region：
(2a) depletion region：[0, R- ε), it is 0 there are the probability of barrier in the area, wherein R refers to the survey that sonar returns
Span is from ε refers to the measurement error of sonar；
(2b) can occupy region：[R- ε, R+ ε] is in the area 0 there is no the probability of barrier；
(3) rasterizing is carried out to the two dimensional surface environment residing for mobile robot, provides the differentiation frame of trellis states could, establishing should
Differentiate the computation model of the reliability assignment under frame：
(3a) carries out rasterizing to the two dimensional surface environment residing for mobile robot, and each grid represents the space of 80*80cm
Size judges the state of grid for three kinds：Sky has barrier, is unknown；
(3b) provides the differentiation frame Ω of trellis states could according to DSmT theories：Ω={ E, O }, wherein E represents sky, and O representatives have
Barrier；
(3c) establishes grid around mobile robot and is differentiating according to the measurement model of sonar sensor and the division of measured zone
Reliability assignment computation model under frame Ω={ E, O }, according to the region where grid, reliability assignment computation model is divided into sonar
The reliability assignment computation model of the depletion region of measurement, the reliability assignment computation model for occupying region of sonar to measure and it is more than
The computation model for occupying region of sonar to measure, DSmT theories are established the ultrapower collection on the basis of, and reliability assignment calculates mould
Type specifically includes the reliability assignment for each element concentrated by the ultrapower that differentiation frame is constituted, the reliability assignment of each element
Value range is [0,1], and it is 0 to require ultrapower to concentrate the reliability assignment value of this element of empty set, and ultrapower concentrates all elements
The sum of reliability assignment be 1；
(4) the DSmT hybrid combining rules for establishing the different moments sonar data under constraints, according to different moments sonar
The DSmT hybrid combining rules of data complete the fusion of mobile robot sonar data.
2. the mobile robot sonar data fusion method according to claim 1 based on DSmT theories, which is characterized in that
The DSmT hybrid combining rules of different moments sonar data of the foundation under constraints described in step 4 refer to, will be by moving
The trellis states could that gets of sonar sensor in multiple and different directions that mobile robot ontology loads it is empty, have barrier and unknown
In the case of combination reliability be collectively formed new combination reliability, and by new combination reliability and non-empty characteristic functionIt is mutually multiplied
To the reliability assignment of grid.
3. the mobile robot sonar data fusion method according to claim 2 based on DSmT theories, which is characterized in that
The DSmT hybrid combining rules of different moments sonar data of the foundation under constraints described in step 4 are specific to calculate public affairs
Formula is：
Wherein, m1() and m2() is the basic reliability assignment from two different data sources, mμ(Ω) indicates band Constrained item
The reliability fusion results of part, μ are constraints E ∩ O=Φ, and wherein Φ is absolute empty set；C is the state of grid, C ∈ DΩ,For non-empty characteristic function, ifThenValue is ' 0 ', is otherwise ' 1 ',For constraints
Collection, Φ indicate absolute empty set,Indicate opposite empty set condition, S1(C) it is the DSmT rules of combination without any constraints,
S2(C) it is that the combination reliability assignment of all absolute empty sets and opposite empty set is allocated to total unknown collection and relatively unknown collection
Under the conditions of rule of combination, total unknown collection refers to the union for differentiating all propositions in frame, S3(C) it is by the combination of the focal element of conflict
Brief inference arrived in the union of the focal element of conflict under conditions of rule of combination.
4. the mobile robot sonar data fusion method according to claim 1 or 2 or 3 based on DSmT theories, special
Sign is, the fusion that sonar data is completed according to the DSmT hybrid combining rules of different moments sonar data described in step 4
Including having the following steps：
(4a) trellis states could differentiates the ultrapower collection D under frame ΩΩFor：DΩ={ Φ, E, O, E ∪ O, E ∩ O }, setting E ∩ O=Φ
For constraints, wherein Φ is absolute empty set, DΩPass through union to be added by the proposition element in differentiation frame Ω：∪ and friendship
Operation：The set for the combination of sentences that ∩ is formed；
(4b) at constraints E ∩ O=Φ, using DSmT hybrid combinings rule melting preceding t-1 moment trellis states could reliability
It closes as a result, being merged with the grid reliability assignment calculated according to current t moment sonar to measure data, completion mobile robot exists
The fusion of the sonar data of the ambient condition information obtained by sonar sensor in the two dimensional surface of circumstances not known.
5. the mobile robot sonar data fusion method according to claim 4 based on DSmT theories, which is characterized in that
Described in (4b) using DSmT hybrid combinings rule by the fusion results of preceding t-1 moment trellis states could reliability, and according to current t
The grid reliability assignment that moment sonar to measure data calculate is merged, and has been specifically included：
Wherein, mij(Ω) (C) indicates the i-th row, the reliability fusion results of jth row grid, C ∈ DΩ,For non-empty characteristic function,
The constraints of fusion is：E ∩ O=Φ；C values are respectively ultrapower collection DΩIn Φ, O, E and E ∪ O when grid 4 kinds of shapes
State reliability fusion calculation process and result mij(Ω) (C) is as follows：
(1) C=Φ, then mij(Φ)=0；
(2) C=O has barrier state：
Non-empty characteristic functionThen：mij(O)=S1(O)+S2(O)+S3(O), wherein：
S2(O)=0, S3(O)=0；
(3) C=E, i.e. dummy status：
Non-empty characteristic functionThen：mij(E)=S1(E)+S2(E)+S3(E), wherein：
S2(E)=0, S3(E)=0；
(4) C=E ∪ O, i.e. unknown state：
Non-empty characteristic functionThen：mij(E ∪ O)=S1(E∪O)+S2(E∪O)+S3(E ∪ O), wherein：
S2(E)=0,
In the calculating process that 4 kinds of state reliabilities of above-mentioned grid merge,Indicate the trellis states could reliability at the t-1 moment
Fusion results,Indicate the trellis states could reliability assignment that the sonar to measure data according to t moment calculate.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810649501.0A CN108802741B (en) | 2018-06-22 | 2018-06-22 | Mobile robot sonar data fusion method based on DSmT theory |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810649501.0A CN108802741B (en) | 2018-06-22 | 2018-06-22 | Mobile robot sonar data fusion method based on DSmT theory |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108802741A true CN108802741A (en) | 2018-11-13 |
| CN108802741B CN108802741B (en) | 2022-05-17 |
Family
ID=64084479
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810649501.0A Active CN108802741B (en) | 2018-06-22 | 2018-06-22 | Mobile robot sonar data fusion method based on DSmT theory |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108802741B (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111160447A (en) * | 2019-12-25 | 2020-05-15 | 中国汽车技术研究中心有限公司 | Multi-sensor perception fusion method of autonomous parking positioning system based on DSmT theory |
| CN111208521A (en) * | 2020-01-14 | 2020-05-29 | 武汉理工大学 | Multi-beam forward-looking sonar underwater obstacle robust detection method |
| CN111931833A (en) * | 2020-07-30 | 2020-11-13 | 上海卫星工程研究所 | Multi-source data driving-based space-based multi-dimensional information fusion method and system |
| CN112612037A (en) * | 2020-12-01 | 2021-04-06 | 珠海市一微半导体有限公司 | Fusion positioning method and mobile robot |
| CN112734878A (en) * | 2020-12-31 | 2021-04-30 | 南昌工学院 | Method, equipment and storage medium for detecting connectivity between two points of large grid map |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070088982A1 (en) * | 2005-10-18 | 2007-04-19 | Valerie Guralnik | System and method for combining diagnostic evidences for turbine engine fault detection |
| CN101413806A (en) * | 2008-11-07 | 2009-04-22 | 湖南大学 | Mobile robot grating map creating method of real-time data fusion |
| CN103778441A (en) * | 2014-02-26 | 2014-05-07 | 东南大学 | Dezert-Smaradache Theory (DSmT) and Hidden Markov Model (HMM) aircraft sequence target recognition method |
| CN107462892A (en) * | 2017-07-28 | 2017-12-12 | 深圳普思英察科技有限公司 | Mobile robot synchronous superposition method based on more sonacs |
-
2018
- 2018-06-22 CN CN201810649501.0A patent/CN108802741B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070088982A1 (en) * | 2005-10-18 | 2007-04-19 | Valerie Guralnik | System and method for combining diagnostic evidences for turbine engine fault detection |
| CN101413806A (en) * | 2008-11-07 | 2009-04-22 | 湖南大学 | Mobile robot grating map creating method of real-time data fusion |
| CN103778441A (en) * | 2014-02-26 | 2014-05-07 | 东南大学 | Dezert-Smaradache Theory (DSmT) and Hidden Markov Model (HMM) aircraft sequence target recognition method |
| CN107462892A (en) * | 2017-07-28 | 2017-12-12 | 深圳普思英察科技有限公司 | Mobile robot synchronous superposition method based on more sonacs |
Non-Patent Citations (3)
| Title |
|---|
| HUIMIN CHAI: "A NOVEL APPROACH TO EVIDENCE COMBINATION IN BATTLEFIELD SITUATION ASSESSMENT USING DEZERT-SMARANDACHE THEORY", 《PROCEEDINGS OF THE 2013 INTERNATIONAL CONFERENCE ON MACHINE LEARNING AND CYBERNETICS》 * |
| 李新德 等: "基于经典DSmT的Sonar栅格地图创建", 《计算机应用研究》 * |
| 李鹏 等: "基于混合DSm模型的移动机器人动态环境地图构建", 《机器人》 * |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111160447A (en) * | 2019-12-25 | 2020-05-15 | 中国汽车技术研究中心有限公司 | Multi-sensor perception fusion method of autonomous parking positioning system based on DSmT theory |
| CN111160447B (en) * | 2019-12-25 | 2023-11-14 | 中国汽车技术研究中心有限公司 | Multi-sensor perception fusion method of autonomous parking positioning system based on DSmT theory |
| CN111208521A (en) * | 2020-01-14 | 2020-05-29 | 武汉理工大学 | Multi-beam forward-looking sonar underwater obstacle robust detection method |
| CN111931833A (en) * | 2020-07-30 | 2020-11-13 | 上海卫星工程研究所 | Multi-source data driving-based space-based multi-dimensional information fusion method and system |
| CN111931833B (en) * | 2020-07-30 | 2022-08-12 | 上海卫星工程研究所 | Multi-source data driving-based space-based multi-dimensional information fusion method and system |
| CN112612037A (en) * | 2020-12-01 | 2021-04-06 | 珠海市一微半导体有限公司 | Fusion positioning method and mobile robot |
| WO2022116657A1 (en) * | 2020-12-01 | 2022-06-09 | 珠海一微半导体股份有限公司 | Fused positioning method and mobile robot |
| CN112612037B (en) * | 2020-12-01 | 2023-10-24 | 珠海一微半导体股份有限公司 | Fusion positioning method and mobile robot |
| CN112734878A (en) * | 2020-12-31 | 2021-04-30 | 南昌工学院 | Method, equipment and storage medium for detecting connectivity between two points of large grid map |
| CN112734878B (en) * | 2020-12-31 | 2023-06-20 | 南昌工学院 | Method, equipment and storage medium for detecting connectivity between two points of large grid map |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108802741B (en) | 2022-05-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108802741A (en) | Mobile robot sonar data fusion method based on DSmT theories | |
| CN105955258B (en) | Robot global grating map construction method based on the fusion of Kinect sensor information | |
| CN1940591B (en) | System and method of target tracking using sensor fusion | |
| JP2501010B2 (en) | Mobile robot guidance device | |
| Arras | An introduction to error propagation: derivation, meaning and examples of equation CY= FX CX FXT | |
| US20200233061A1 (en) | Method and system for creating an inverse sensor model and method for detecting obstacles | |
| CN106066154B (en) | A kind of extracting method of target and its control point suitable for quickly scanning scene | |
| CN109085838A (en) | A kind of dynamic barrier rejecting algorithm based on laser positioning | |
| CN108801268A (en) | Localization method, device and the robot of target object | |
| Park et al. | Radar localization and mapping for indoor disaster environments via multi-modal registration to prior LiDAR map | |
| CN109917788A (en) | A kind of control method and device of Robot wall walking | |
| CN109583416A (en) | Pseudo- Lane detection method and system | |
| Mueller et al. | GIS-based topological robot localization through LIDAR crossroad detection | |
| CN104966123A (en) | SLAM data association method based on fuzzy-self-adaptation | |
| CN109341688A (en) | A kind of map calling location algorithm based on construction sequence | |
| Jiang et al. | Intelligent Plant Cultivation Robot Based on Key Marker Algorithm Using Visual and Laser Sensors | |
| CN106908054A (en) | A kind of positioning path-finding method and device based on ultra-wideband signal | |
| Krotkov et al. | Adaptive control of cooperating sensors: Focus and stereo ranging with an agile camera system | |
| Kao et al. | Feature extraction from a broadband sonar sensor for mapping structured environments efficiently | |
| CN113376574B (en) | Person position matching method, device and equipment based on wireless positioning and image processing | |
| CN104619017A (en) | Map-aided indoor positioning system WiFi (wireless fidelity) access point deploying scheme | |
| CN108919236A (en) | A kind of space junk laser ranging effect emulation analysis method and device | |
| CN105445741B (en) | A kind of method, apparatus and system of target positioning | |
| Li et al. | Robot map building from sonar sensors and DSmT | |
| Li et al. | Robot map building from sonar and laser information using DSmT with discounting theory |
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 |