CN108307767A - Detection of obstacles obstacle avoidance system and method suitable for full-automatic weeder - Google Patents
Detection of obstacles obstacle avoidance system and method suitable for full-automatic weeder Download PDFInfo
- Publication number
- CN108307767A CN108307767A CN201810031611.0A CN201810031611A CN108307767A CN 108307767 A CN108307767 A CN 108307767A CN 201810031611 A CN201810031611 A CN 201810031611A CN 108307767 A CN108307767 A CN 108307767A
- Authority
- CN
- China
- Prior art keywords
- signal
- ultrasonic
- detection
- barrier
- weeder
- 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
-
- A—HUMAN NECESSITIES
- A01—AGRICULTURE; FORESTRY; ANIMAL HUSBANDRY; HUNTING; TRAPPING; FISHING
- A01D—HARVESTING; MOWING
- A01D34/00—Mowers; Mowing apparatus of harvesters
- A01D34/006—Control or measuring arrangements
- A01D34/008—Control or measuring arrangements for automated or remotely controlled operation
-
- A—HUMAN NECESSITIES
- A01—AGRICULTURE; FORESTRY; ANIMAL HUSBANDRY; HUNTING; TRAPPING; FISHING
- A01D—HARVESTING; MOWING
- A01D75/00—Accessories for harvesters or mowers
- A01D75/18—Safety devices for parts of the machines
- A01D75/185—Avoiding collisions with obstacles
Abstract
The invention discloses the detection of obstacles obstacle avoidance systems and method suitable for full-automatic weeder, including：MCU module, MCU module are connected by the ultrasonic transducer for transmitting of radiating circuit and ultrasonic phase array, and the ultrasonic transducer for transmitting is located at the top of weeder；MCU module control radiating circuit driving ultrasonic transducer sends out ultrasonic signal, ultrasonic signal is reflected by front obstacle, emit signal to be received by several ultrasonic transducers for reception, several ultrasonic transducers for being used for reception form ultrasonic phase array receiving array；The ultrasonic phase array receiving array passes sequentially through signal amplification conditioning module, detection low-pass filter, A/D conversion modules and is connect with MCU processing modules；MCU processing modules are connect with host computer.Judgement is identified to front obstacle in application mode identification technology and ultrasonic phased array technology, realizes the automatic intelligent work of robot and full-automatic weeder.
Description
Technical field
The present invention relates to the detection of obstacles obstacle avoidance systems and method suitable for full-automatic weeder.
Background technology
With the improvement of people's living conditions, full-automatic weeder has been obtained for increasingly extensive application.But in many
In garden or greenery patches usually in addition to the meadow of plantation, also trees, swing, unlike material the facilities such as man-made tourist site.To weeding
The requirement of machine is only to be trimmed to lawn, avoids trees, artificial hillock etc., in addition swing (if height allows) etc. should also be passed through to set
Apply, i.e., should as far as possible for all weeders can by the lawn in space trim.Existing full-automatic weeder
Since the artificial facility such as the grass in front, shrub, swing can not be identified, thus will appear the trimming of certain plot less than, cut
Drop to situations such as trees even hinder knife.
The invention of current various full-automatic weeders has very much：
A kind of intelligent weeder that CN204682323 patents are proposed, is imaged by weed identification video camera, self-navigation
Machine and controller realize the identification of front obstacle；One kind disclosed in CN206005198 patents being based on infrared ray sensing
The automation weeder of device；A kind of weeder for evading function automatically with barrier disclosed in CN106612973 patents
Deng.Using video technique, its cost is very high, while can be influenced by environmental factors such as illumination, sleet, dust, temperature, to
It limits its scope of application.It is vulnerable to the influence of the ambient environmental factors such as sunlight using infrared sensor, in actual use
It will appear measured deviation.A kind of mobile robot all-directional ultrasonic obstacle avoidance apparatus disclosed in CN202939487 patents uses
Positioning and measurement device of the ultrasonic sensor as barrier, determines position and the distance of barrier, reaches avoidance
Purpose, but it can not provide barrier attribute information, cannot distinguish between different barriers, and be easy by environment such as sleet, dusts storm
Influence and judge by accident, cannot accomplish the intelligent recognition avoidance to front obstacle.
In conclusion the critical issue that the devices such as robot of intelligent weeder and similar applications are faced is exactly nothing
The obstacle identity in front is identified in method, is the weeds that wipe out to which which distinguish；And such as illumination,
The property of barrier is steadily identified in the high-interference environments such as sleet, dust.
Invention content
In order to solve the deficiencies in the prior art, the present invention provides the detection of obstacles avoidances suitable for full-automatic weeder
System and method has using ultrasonic sensor as detection sensor, integrated use mode identification technology and ultrasonic phase
Judgement is identified to front obstacle in control battle array technology, the equipment that can realize the similar applications such as robot and full-automatic weeder
Realize automatic intelligent work.Sonac have be not illuminated by the light, smog, sleet, temperature, the environmental factors such as inflammable and explosive
It influences, it is stable and reliable in work, it round-the-clock can use.
Suitable for the detection of obstacles obstacle avoidance system of full-automatic weeder, including：MCU module, the MCU module pass through hair
The ultrasonic transducer for transmitting of transmit-receive radio road and ultrasonic phase array connects, and the ultrasonic transducer for transmitting is located at weeding
The top of machine；MCU module control radiating circuit driving ultrasonic transducer sends out ultrasonic signal, and ultrasonic signal passes through front
Barrier reflects, and transmitting signal is received by several ultrasonic transducers for reception, several are for the super of reception
Acoustic wave transducer forms ultrasonic phase array receiving array；The ultrasonic phase array receiving array passes sequentially through signal amplification conditioning mould
Block, detection low-pass filter, A/D conversion modules are connect with MCU processing modules；The MCU processing modules are connect with host computer；
Each ultrasonic transducer of the receiving array of ultrasonic phase array both participates in reception and reflects ultrasonic signal, often
The ultrasonic signal that a ultrasonic transducer receives passes through signal amplification conditioning module and carries out preposition amplification, detection and low pass filtered
Then wave processing is converted into digital signal feeding MCU module by A/D conversion modules and is stored, MCU module passes through to each
Railway digital signal is handled, so that it is determined that the attribute of front obstacle, position and shape information.
The MCU processing modules are using ARM microcontrollers as core processor.
The signal amplifies conditioning module, including：Sequentially connected signal buffer, gain adjustable amplifier and signal solution
Adjust device；Wherein demodulator of PM signal PM is used to demodulate to come from modulated signal by the obstacle information that ultrasonic wave carries, the signal
The ultrasonic transducer of the receiving array of buffer and ultrasonic phase array connects, and the demodulator of PM signal PM connects with low-pass filter
It connects.
The ultrasonic phase array works in such a way that single-shot is received more：
First, the ultrasonic transducer for being used for transmitting emits N number of short pulse, then by remaining super in ultrasonic phased array
Acoustic wave transducer, which synchronizes, receives signal, and after amplification, conditioning and detection, low-pass filtered device is stored in after sending A/D conversion modules
It is stored in MCU module, signal is received using any road of storage and weeder is measured apart from obstacle using peak detection method
The distance D of object, and judge whether distance D is more than preset distance；
If distance D is more than preset distance, then it is assumed that front clear, before weeder is according to the route of planning
Row；
If distance D is not more than preset distance, then it is assumed that there is barrier in front；
By horizontal distance immediately ahead of the center O points of ultrasonic phase array for O ' of D centered on, to make perpendicular to the face of O O ＇
For virtual detection section, virtual detection interface is parallel plane with receiving array plane, according to default rule in detection section
Interior selection virtual focus point is detected；By calculating the distance between each virtual focus point and each reception energy converter, obtain each
The range difference of virtual focus point to the ultrasonic transducer for being respectively used for reception is di, by range difference diDivided by the speed of ultrasonic wave obtains
Corresponding time ti, to obtain the phase difference θ of signal between virtual focus point and each reception energy converteri, utilize θiTo each road signal
Carry out phase shift superposition, obtain the obstacle information near each virtual focus point, so that it is determined that go out the attribute of front obstacle, position with
Shape.
Analysis method to barrier attribute is：Signal after being superimposed to phase shift does wavelet decomposition, by the high-frequency energy of signal
The ratio of component and low frequency energy component is measured as characteristic value, the svm classifier come is trained by using pre-stored data
Device is classified, and it is grass or other barriers to distinguish front obstacle；
The step of selecting virtual focus point for：Centered on O ＇ points, weeder local width is that radius does circle, circle it is inscribed just
Rectangular lower-left, upper left, bottom right, upper right, central point, central point top, central point lower section, central point left and central point right
Nine points be virtual focus point；
To reduce the influence of interference signal, using time window technology, i.e. [t is only received after A/D conversionsi-Δ,ti+Δ]
Signal in period, to further increase signal-to-noise ratio.
Suitable for the detection of obstacles barrier-avoiding method of full-automatic weeder, step：
Step (1)：Ultrasonic transducer of the train of pulse control of the MCU module output setting length of weeder for transmitting
Ultrasonic signal is sent out, wherein the energy converter positioned at array center is transmitting transducer, remaining is reception energy converter；
Step (2)：Pass through A/D after encountering the conditioned reflected ultrasonic signal of barrier, detection and low-pass filtering
Conversion module is sent into MCU module and is stored；
Step (3)：By subject to the echo-signal all the way that is arbitrarily designated, utilize peak detection method carry out obstacle distance just
Step judgement；According to the obstacle distance D measured；Judge whether distance D is more than preset distance；If distance D is more than
Preset distance, then it is assumed that front clear, weeder move ahead according to the route of planning；If distance D is no more than pre-
The distance first set, then it is assumed that there is barrier in front；
By horizontal distance immediately ahead of the center O points of ultrasonic phase array for the O ＇ points of D centered on, to make perpendicular to the face of O O ＇
To detect section, selection virtual focus point is detected in detection section；The step of selecting virtual focus point for：In being with O ＇ points
The heart, weeder local width are that radius does circle, the lower-left of the inscribed square of circle, upper left, bottom right, upper right, central point, central point
Top, below central point, nine vertex of central point left and central point right are virtual focus point, and calculate nine virtual focus points
Coordinate；
Step (4)：Emit a ultrasonic pulse string, each ultrasonic transducer of the receiving array of ultrasonic phase array again
It both participates in reception and reflects ultrasonic signal, the ultrasonic signal that each ultrasonic transducer receives passes through signal amplification and adjusts
It manages module and carries out preposition amplification, detection and low-pass filtering treatment, being then converted into digital signal by A/D conversion modules is sent into
MCU module is simultaneously stored, and being needed before feeding MCU module storage will be other than virtual focus point setpoint distance using time window
Back echo signal is shielded；
Step (5)：Using the reception signal for being stored in MCU module, according to the position of virtual focus point, with it is specified all the way
The signal that the remaining ultrasonic transducer for reception receives is carried out phase shift superposition by signal on the basis of reception signal, thus
Several signals being concerned with reference signal are formed, and the signal after superposition is stored in MCU module according to the serial number of virtual focus point；
Step (6)：To each virtual focus point, the method that step (5) is respectively adopted carries out the pretreatment of signal, then utilizes
The roads peak detection Fa Duige signal is detected, so that it is determined that whether going out each virtual focus point position with the presence of barrier and obstacle
The azimuth information of object；
Step (7)：The digital signal for taking out the intermediate point stored in MCU module, carries out wavelet function feedback respectively, profit
Reflected signal characteristic is extracted with the multiscale space energy-distributing feature of wavelet transformation, in conjunction with svm classifier, front is hindered
The attribute of object is hindered to be identified；
Step (8)：Each virtual focus point signal after MCU module is superimposed phase shift scans and recognizes, and obtains virtual focus point
The 2D digital pictures of barrier in neighbouring setpoint distance；When weeder moves, MCU module obtains the virtual of each moment successively
The 2D digital pictures of barrier, form the 3D rendering of barrier near focal point setpoint distance, and barrier is obtained according to 3D rendering
Spatial depth on shape and size；
Step (9)：The attribute of barrier after identification, shape and size are notified into host computer.
In the step (2), ultrasonic phase array is circumferentially to place four respectively on four vertex of inscribed square
For the ultrasonic transducer of reception, centre bit puts a ultrasonic transducer for transmitting, five ultrasonic transducers
It is all made of the transceiver piezoelectric ceramic transducer of same centre frequency.The circumference is centered on O points, and weeder itself is wide
Degree is that radius does circle.
In the step (5), multiple Digital Signal Processing is carried out using the signal of a synchronous acquisition, respectively obtains obstacle
Multiple features of object；
In the step (7), the wavelet function feedback is to carry out wavelet decomposition to signal and reconstruct high fdrequency component
And low frequency component, using high fdrequency component and the energy ratio of low frequency component as signal characteristic, using trained SVM divides in advance
Class device classifies to signal characteristic, so that it is determined that going out the property of front obstacle and being identified.
The step (8), steps are as follows：
Step (81)：Sliding-model control is carried out to the section of scanning, i.e., section is divided into several discrete points, respectively institute
Discrete point is stated as virtual focus point to judge to be scanned；
Step (82)：After scanning starts, from first specified discrete point as virtual focus point, arrived according to the virtual focus point
It is each used for the differential time of flight of the ultrasonic transducer of reception in ultrasonic phase array, calculates each channel signal phase to be moved
Position simultaneously carries out Signal averaging successively；
Step (83)：Barrier attribute is distinguished using grader；
Step (84)：The comprehensive information of each virtual focus point in a detection plane judges position and the 2D letters of barrier
Breath；
Step (85)：With the movement of weeder, constantly newly-generated virtual focus point is scanned, to form front
The 3D location and shape of barrier in search coverage.
Compared with prior art, the beneficial effects of the invention are as follows：
1. introducing the signal processing method of wavelet pretreatment and Wavelet decomposing and recomposing.By external hardware to echo-signal
Filtering, improves and is converted into digital signal by A/D conversion modules amplification, filters out most of noise of signal.In order to extract back
The feature of wave signal introduces wavelet transformation and handles signal, and to treated, signal carries out wavelet function feedback, obtains
The high fdrequency component and low frequency component of signal using the ratio of its high and low frequency component as signal characteristic, and are classified using SVM,
To identify meadow from front obstacle, to realize that intelligent barrier avoiding lays the first stone.
2. present invention employs the ultrasonic phased arrays that single-shot is received more.Relative to the ultrasound phase-control array 1 system of multiple illuminators and single receiver,
The present invention need not provide the driving pulse to timing requirements harshness, can reduce the error of hardware system from source, energy disappears
The complexity of consumption and circuit；
3. in order to simplify system design, the abundant hardware resource and operation energy of ARM Cortex-M4 kernels have fully been used
Power, using a synchronous acquisition and the signal of storage, by different Digital Signal Processings distinguish acquired disturbance object away from
From, orientation and attribute information, the complexity and cost of system are reduced, system power dissipation is reduced.
4. using ultrasonic wave as detectable signal, do not influenced by environmental factors such as light, magnetic field, dust, adapts to environment
Ability is strong, can be worked normally under with rugged environments such as corrosive gas, radiation, dust.
5. after tested, realizing and only needing 3 ultrasonic transducers to the positioning of front obstacle and intelligent recognition.The present invention
4 reception energy converters are used, once there is the failure of some energy converter, system still can normally work, and improve and be
The robustness of system.
6. simple in structure, at low cost, be easily achieved, be easy to adjustment, have higher stability and all weather operations etc. excellent
The detection of obstacles obstacle avoidance algorithm and network system realization suitable for full-automatic weeder of point realize weeder or similar applications
Device to the intelligent recognition function of front obstacle.
7. being handled the echo-signal of barrier using mode identification technology, to identify the property of barrier.
Wavelet analysis is a kind of emerging branch of mathematics, have multiscale space energy-distributing feature abstraction function, including energy
It measures spectrum distribution and the size, shape and type of front reverberation is closely related, in signal processing, image procossing, voice
Processing and numerous nonlinear science fields are widely applied.The signal that receives ultrasonic receiver in this patent into
Row wavelet decomposition, and reconstruct the low frequency component and high fdrequency component of signal respectively, using low frequency and the ratio of high fdrequency component as
The identification feature of front obstacle, and utilize the intelligent recognition of svm classifier realization front obstacle.
8. carrying out pattern-recognition using echo-signal merely, it can only realize that the approximate distance of barrier measures, can not obtain
Obtain the information such as orientation, the shape of barrier.To realize that the determination of barrier more information, this patent use ultrasonic phase array skill
Art, Digital Signal Processing and mode identification technology realize the identification to the azimuth-range information of different barriers.For drop
Low cost reduces the power consumption of whole system, and this patent uses the ultrasonic phase array that single-shot is received more, and Reusability once acquires
To data by way of being verified one by one to virtual focus point, to it is N number of space different parts place receiver letter
Number phase shift processing is carried out, to realize to the detection of front specific location spatial obstacle object and obtain the azimuth-range of barrier
Information.
Signal processing technology and phased-array technique are used by comprehensive, the identification to front obstacle may be implemented and sweep
It retouches, by the fusion to multiple cross section informations, realizes the identification of barrier attribute, azimuth-range.
Description of the drawings
The accompanying drawings which form a part of this application are used for providing further understanding of the present application, and the application's shows
Meaning property embodiment and its explanation do not constitute the improper restriction to the application for explaining the application.
Fig. 1 is the system block diagram of the present invention；
Fig. 2 ultrasonic phase arrays receive schematic diagram；
Fig. 3 ultrasonic phase array plan views；
Fig. 4 is detecting visual floor map in front of the present invention；
Fig. 5 is the scanning plane schematic diagram of discretization；
Fig. 6 (a) is the transformed original signal datas of A/D；
Fig. 6 (b) is signal datas of the Fig. 6 (a) after wavelet pretreatment；
Fig. 6 (c) is the classification results figure to multigroup different front reverberations.
Specific implementation mode
It is noted that following detailed description is all illustrative, it is intended to provide further instruction to the application.Unless another
It indicates, all technical and scientific terms used herein has usual with the application person of an ordinary skill in the technical field
The identical meanings of understanding.
It should be noted that term used herein above is merely to describe specific implementation mode, and be not intended to restricted root
According to the illustrative embodiments of the application.As used herein, unless the context clearly indicates otherwise, otherwise singulative
It is also intended to include plural form, additionally, it should be understood that, when in the present specification using term "comprising" and/or " packet
Include " when, indicate existing characteristics, step, operation, device, component and/or combination thereof.
As shown in Figure 1, the detection of obstacles obstacle avoidance system suitable for full-automatic weeder, which includes a MCU master controls
Module, MCU module are using ARM Cortex-M3 as the 32-bit microprocessor of kernel, using the STM32F103 microcontrollers of ST companies
For core processor；Including a radiating circuit, sends out signal driving ultrasonic transducer by MCU and sends out ultrasonic signal, to preceding
Square barrier is detected, and ultrasonic transducer uses centre frequency for 40KHz, the piezoelectric ceramic ultrasound transducing of transceiver
Device, radiation pattern are discontinuous transmission, send out the pulse signal in 10 periods every time, ensure the accurate detection to front obstacle；
Ultrasonic transducer uses simplified array arrangement in the present embodiment according to specific spatial distribution composition phased array,
Place four ultrasonic transducers respectively on four vertex of square, centre bit puts an energy converter, refers to Fig. 3 institutes
Show；The operating mode more received using single-shot in the present embodiment, in detail as shown in Figure 2；The array is connect with signal amplifying and conditioning circuit,
Signal amplifying and conditioning circuit is filtered to receiving the original signal that receives of supersonic array, amplifies, improve after, send A/D to convert
Module；A/D conversion modules are the analog-digital converters being internally integrated by kernel processor chip STM32F104, possess very high resolution
Echo-signal accurately can be converted to digital signal and MCU module is sent to be stored by rate, the transformed original signal numbers of A/D
According to shown in such as Fig. 6 (a)；Signal processing module is by the software realization based on MCU, and specific completion is to digital signal time domain or change
Change the processing in domain.Wherein Time Domain Processing includes the phase shift overlap-add operation to each road signal, threshold test, verification virtual focus point etc.
Reason, and then determine the azimuth information of front obstacle；Transform-domain processing using wavelet pretreatment and signal wavelet decomposition with
Reconstruct, obtains the attribute information of front obstacle, final to realize the functions such as barrier feature information extraction and intelligent recognition；Pass through
All virtual focus points on the section of front are scanned successively, to verify depositing for virtual focus point on the specific range section of front
In situation, and according to the 2D digital informations of echo-signal Formation cross-section barrier；When carrying invention system platform movement,
MCU can be scanned with the continuous pair cross-section of Fixed Time Interval, you can by verifying the virtual focus point on different cross section, before formation
The 3D digital informations of side space barrier, to provide foundation for the identification of barrier form and Path selection；System is to front
Corresponding prompt message is returned to according to scanning result after the completion of spacescan, output equipment is sent to, is done with controlling transmission device
Go out to act and inform user.
In conjunction with Fig. 1, for the present invention using embedded system as core, one built using ultrasonic phase array as sensor can be to preceding
Side space implements the intelligent barrier avoiding system of scanning.The system can realize that scanning probe, orientation recognition, the attribute of barrier obtain
It the functions such as takes, be imaged, and can realize that relevant information exports.
In conjunction with Fig. 1, when the resolution requirement to spatial obstacle object is not high, indexed by the discretization of reduction pair cross-section,
The structure for simplifying phased array, can be such that whole system is simplified, and the major function realized has：
1. the transmitting of ultrasonic wave, data receiver, data processing, storage and prompt message output function；
2. the orientation characteristics identification of barrier, scan image are formed, signal amplification, graphical information exports and data store work(
Energy；
3. the emission control of ultrasonic wave, signal acquisition, docking collect mail number transform-domain processing, picture signal coordinate setting
And real time information output function；
4. can be estimated to the substantially shape and its essential attribute for detecting object by remembering acquired obstacle information
Meter；
In conjunction with Fig. 3, the phased array uses 5 centre frequencies to be pressed for the EU10AIF40H07A of 40KHz, transceiver
Electroceramics ultrasonic transducer forms simplified two-dimensional rectangle array, and that MCU module controls a ultrasonic transducer transmitting is super
Sound wave, the receives echo-signal of remaining energy converter independently, and it is converted into digital signal, it is folded by the phase shift to digital signal
Add, threshold test processing, and then verify virtual focus point and whether there is, forms the scanning work(to different virtual focus points and different section
Energy.
In conjunction with Fig. 4,5, the area of space that the present invention is scanned, which is one with a certain distance from receiving array, has certain indulge
Deep, width and height space, the present embodiment receive phase using a rectangular space, nearest scanning plane centre distance
0.6 meter of battle array center is controlled, farthest scanning plane centre distance receives 1.0 meters, following ground proximity 50mm of phased array center, top margin
Ground level is higher than weeder machine clear height, and left and right width is more than machine clear span (being dynamically adapted)；Refer to Fig. 4, Fig. 5.
In conjunction with Fig. 3,4,5, the present invention carries out sliding-model control to scanning plane first, i.e., scanning plane be divided into M × N number of from
Then scatterplot is scanned verification using various discrete point as virtual focus point respectively.The virtual coke verified needed for each scanning plane
Points can select that (virtual focus point number is more, and data processing complexity and required precision are also higher, thus cost as needed
Also higher), height × width is used in the present embodiment, and totally 36 virtual focus points, refer to Fig. 5 for 6 × 6.By calculating transmitting transducer
To virtual focus point and virtual focus point to each flight time for receiving energy converter, reasonably plans scan mode, realize section
Scanning.
Scan virtual focus point position it is constant under the premise of, can will be to the ginseng of each spacescan point to accelerate sweep speed
Number calculates complete and make table in advance, realizes that the verification of virtual focus point scans using lookup table mode.
It is the transformed original signal datas of A/D in conjunction with Fig. 6 (a), Fig. 6 (b), wherein Fig. 6 (a), Fig. 6 (b) is Fig. 6 (a)
Signal data after wavelet pretreatment, from can be seen that in Fig. 6 (a) in the signal that A/D is converted, there are more high frequencies to do
Signal is disturbed, these interference signals can impact subsequent signal processing, influence the accuracy of system, and the present invention uses small echo
The digital signal processing method of analysis pre-processes signal, obtains the signal waveform in Fig. 6 (b), it can be seen that high
Frequency interference component is obviously removed, and is convenient for later signal processing, is improved the accuracy of system.
In conjunction with Fig. 6 (c), Fig. 6 (c) is the classification results figure to multigroup different front reverberations.The present invention uses DB6 small echos
1 layer of wavelet decomposition is carried out to signal and reconstructs the low frequency component and high fdrequency component of signal, compares the pass of low frequency and high fdrequency component
System obtains the attribute of front obstacle, is extracted using the multiscale space energy-distributing feature of wavelet transformation, including energy
It measures spectrum distribution and the size, shape and type of front reverberation is closely related, to provide foundation for intelligent barrier avoiding.
The digital echo signal received by A/D conversions is stored in array x [n], then can be obtained after 1 layer of wavelet decomposition
Its approximation coefficient and detail coefficients：
Wherein cA1 is 1 layer of wavelet decomposition approximation coefficient, cD1For 1 layer of wavelet decomposition detail coefficient, g0[n] is low-pass filtering
Device, h0[n] is high-pass filter.
By approximation coefficient cA1 and detail coefficients cD1Approximate signal and detail signal can be reconstructed：
Wherein A1For the approximate signal of 1 layer of Wavelet decomposing and recomposing, D1For the detail signal of 1 layer of Wavelet decomposing and recomposing, g1[n]
=g0[- n], h1[n]=h0[-n]。
Feature is extracted to echo-signal after wavelet function feedback：
Wherein s is signal characteristic information, PLFor low frequency component, PHFor high fdrequency component.
The present invention is identified signal using the machine learning algorithm of Linear SVM.Linear SVM are one linear
The grader that can divide, some training samples marked can be defined by the classification line that inhomogeneity sample space separates by being one
This, LinearSVM algorithms export the classification line of an optimization, and following formula (3) defines the expression formula of classification line：
F (x)=β0+βTx (3)
β is called weight vectors, β in formula0It is called biasing.
Classification line can have countless expression formula, i.e., by arbitrarily scaling β and β0.It is realized by algorithm, we can be with
The optimal classification line classified, the classification of complete pair signals.
In embodiment, the present invention has carried out contrast test to different front obstacles, chooses weeds respectively, trees make
For experimental subjects, by carrying out data acquisition respectively to experimental subjects, obtaining their echo-signal and carrying out wavelet pretreatment,
Compare the relationship of its high frequency and low frequency component, chooses training and test data, classify with Linear SVM algorithms, to survey
Test system judges whether correctly to front obstacle.The detection range range that the present invention is set is institute between 0.6m~1.0m
Only to choose echo-signal, at this, the corresponding point in section is analyzed.15 groups of weed datas and 15 groups of trees have been taken respectively
The wooden data are analyzed, and isolate the low frequency component and high fdrequency component of each signal, one two-dimensional coordinate point P of structure is as the letter
Number characteristic information s, with low frequency component PLFor abscissa x, with high fdrequency component PHIt is clear in order to show for ordinate y, to original
Data have done logarithm process.
The logarithmic coordinates point of signal depicts two-dimensional coordinate figure according to embodiments of the present invention, the data of weeds and trees
Point is placed in a coordinate diagram.System uses Linear SVM algorithms and classifies to data, takes respectively in 15 groups of data
First 10 groups are used as training, and latter 5 groups are used as test, refer to Fig. 6 (c), can be obtained by figure, can be right using Linear SVM algorithms
Signal is classified, and is traversed respectively to the data point of two classifications and is sought Euclidean distance, and the mathematic(al) representation of two-dimensional space is：
Wherein (x1,y1), (x2,y2) it is respectively two-dimensional coordinate point coordinates, OρFor the Euclidean distance of two coordinate points.
Show that two groups most short Euclidean distance is respectively a1--b1 and a2--b2.The midpoint c1 of a1b1, a2b2 line is taken respectively,
C2 makes straight line L as two point coordinates, this straight line is our the optimal classification lines to be looked for, it can be deduced that in formula (3)
ParameterThen show that the equation of optimal classification line L is：
X-2y+3=0 (5)
To determine the classification results of data, we can find out data point and the relative position relation for line of classifying is sentenced
It is disconnected.
Such as to signal characteristic P0(x0,y0) classification judgement is carried out, f (x, y)=x-2y+3 is enabled, there is following estimate of situation：
f(x0,y0)=x0-2y0+ 3 ＞ 0 (6)
f(x0,y0)=x0-2y0+ 3 ＜ 0 (7)
If inequality (6) is set up, signal characteristic P is obtained0(x0,y0) in the top of classification line L, it is possible to determine that classification is
Weeds.
If inequality (7) is set up, signal characteristic P is obtained0(x0,y0) in the lower section of classification line L, it is possible to determine that classification is
Trees.
By experimental verification, signal is handled with Linear SVM algorithms, can achieve the purpose that identification classification,
And then weeds and trees can be distinguished, the attribute information of front obstacle is obtained, realizes the function of intelligent barrier avoiding.
It will be understood by those skilled in the art that each module or each step of aforementioned present invention can be filled with general computer
It sets to realize, optionally, they can be realized with the program code that computing device can perform, it is thus possible to which they are stored
Be performed by computing device in the storage device, either they are fabricated to each integrated circuit modules or by they
In multiple modules or step be fabricated to single integrated circuit module to realize.The present invention is not limited to any specific hardware and
The combination of software.
The foregoing is merely the preferred embodiments of the application, are not intended to limit this application, for the skill of this field
For art personnel, the application can have various modifications and variations.Within the spirit and principles of this application, any made by repair
Change, equivalent replacement, improvement etc., should be included within the protection domain of the application.
Claims (10)
1. the detection of obstacles obstacle avoidance system suitable for full-automatic weeder, characterized in that including：MCU module, the MCU moulds
Block is connected by the ultrasonic transducer for transmitting of radiating circuit and ultrasonic phase array, the ultrasonic transducer for transmitting
Positioned at the top of weeder；MCU module control radiating circuit driving ultrasonic transducer sends out ultrasonic signal, ultrasonic signal
It is reflected by front obstacle, transmitting signal is received by several ultrasonic transducers for reception, several are used for
The ultrasonic transducer of reception forms ultrasonic phase array receiving array；The ultrasonic phase array receiving array passes sequentially through signal and puts
Big conditioning module, detection low-pass filter, A/D conversion modules are connect with MCU processing modules；The MCU processing modules with it is upper
Machine connects；
Each ultrasonic transducer of the receiving array of ultrasonic phase array both participates in reception and reflects ultrasonic signal, Mei Gechao
The ultrasonic signal that acoustic wave transducer receives, which passes through signal amplification conditioning module, to carry out at preposition amplification, detection and low-pass filtering
Then reason is converted into digital signal feeding MCU module by A/D conversion modules and is stored, MCU module passes through to each way
Word signal is handled, so that it is determined that the attribute of front obstacle, position and shape information.
2. being suitable for the detection of obstacles obstacle avoidance system of full-automatic weeder as described in claim 1, characterized in that
The signal amplifies conditioning module, including：Sequentially connected signal buffer, gain adjustable amplifier and signal demodulation
Device；Wherein demodulator of PM signal PM is used to demodulate to come from modulated signal by the obstacle information that ultrasonic wave carries, and the signal is slow
The ultrasonic transducer connection of the receiving array of device and ultrasonic phase array is rushed, the demodulator of PM signal PM is connect with low-pass filter.
3. being suitable for the detection of obstacles obstacle avoidance system of full-automatic weeder as described in claim 1, characterized in that
The ultrasonic phase array works in such a way that single-shot is received more：
First, the ultrasonic transducer for being used for transmitting emits N number of short pulse, then by remaining ultrasonic wave in ultrasonic phased array
Energy converter, which synchronizes, receives signal, and after amplification, conditioning and detection, deposit is in MCU after low-pass filtered device send A/D conversion modules
Stored in module, using storage any road receive signal and using peak detection method measure weeder apart from barrier away from
From D, and judge whether distance D is more than preset distance；
If distance D is more than preset distance, then it is assumed that front clear, weeder move ahead according to the route of planning；
If distance D is not more than preset distance, then it is assumed that there is barrier in front；
By horizontal distance immediately ahead of the center O points of ultrasonic phase array for O ' of D centered on, using perpendicular to the face of O O ' as empty
Quasi- detection section, virtual detection interface are parallel plane with receiving array plane, are selected in detection section according to default rule
Virtual focus point is taken to be detected；By calculating the distance between each virtual focus point and each reception energy converter, obtain each virtual
The range difference of focus to the ultrasonic transducer for being respectively used for reception is di, by range difference diDivided by the speed of ultrasonic wave obtains accordingly
Time ti, to obtain the phase difference θ of signal between virtual focus point and each reception energy converteri, utilize θiEach road signal is carried out
Phase shift is superimposed, and the obstacle information near each virtual focus point is obtained, so that it is determined that going out the attribute of front obstacle, position and shape
Shape.
4. being suitable for the detection of obstacles obstacle avoidance system of full-automatic weeder as claimed in claim 3, characterized in that
Analysis method to barrier attribute is：Signal after being superimposed to phase shift does wavelet decomposition, by the high-frequency energy of signal point
Amount and low frequency energy component ratio be used as characteristic value, by using pre-stored data train SVM classifier into
Row classification, it is grass or other barriers to distinguish front obstacle.
5. being suitable for the detection of obstacles obstacle avoidance system of full-automatic weeder as claimed in claim 3, characterized in that
The step of selecting virtual focus point for：Centered on O ＇ points, weeder local width is that radius does circle, the inscribed square of circle
Lower-left, upper left, bottom right, upper right, central point, above central point, below central point, central point left and central point right nine
A point is virtual focus point.
6. being suitable for the detection of obstacles obstacle avoidance system of full-automatic weeder as claimed in claim 3, characterized in that
[t is only received after A/D conversionsi-Δ,ti+ Δ] signal in the period, to further increase signal-to-noise ratio.
7. the detection of obstacles barrier-avoiding method suitable for full-automatic weeder, characterized in that step：
Step (1)：Ultrasonic transducer of the train of pulse control of the MCU module output setting length of weeder for transmitting is sent out
Ultrasonic signal, wherein the energy converter positioned at array center is transmitting transducer, remaining is reception energy converter；
Step (2)：It is converted by A/D after encountering the conditioned reflected ultrasonic signal of barrier, detection and low-pass filtering
Module is sent into MCU module and is stored；
Step (3)：Being sentenced by subject to the echo-signal all the way that is arbitrarily designated, utilizing peak detection method to carry out the preliminary of obstacle distance
It is fixed；According to the obstacle distance D measured；Judge whether distance D is more than preset distance；If distance D is more than advance
The distance of setting, then it is assumed that front clear, weeder move ahead according to the route of planning；It is set in advance if distance D is not more than
Fixed distance, then it is assumed that there is barrier in front；
By horizontal distance immediately ahead of the center O points of ultrasonic phase array for the O ＇ points of D centered on, using perpendicular to the face of O O ＇ as spy
Section is surveyed, choosing virtual focus point in detection section is detected；The step of selecting virtual focus point for：Centered on O ＇ points, remove
Careless machine local width is that radius does circle, the lower-left of the inscribed square of circle, upper left, bottom right, upper right, central point, central point top,
Nine vertex of central point lower section, central point left and central point right are virtual focus point, and calculate the seat of nine virtual focus points
Mark；
Step (4)：Emit a ultrasonic pulse string again, each ultrasonic transducer of the receiving array of ultrasonic phase array is joined
Ultrasonic signal is reflected with reception, the ultrasonic signal that each ultrasonic transducer receives passes through signal amplification conditioning mould
Block carries out preposition amplification, detection and low-pass filtering treatment, and being then converted into digital signal by A/D conversion modules is sent into MCU moulds
Block is simultaneously stored, and needs to use time window by the obstacle other than virtual focus point setpoint distance before being sent into MCU module storage
Object echo-signal is shielded；
Step (5)：Using the reception signal for being stored in MCU module, according to the position of virtual focus point, with specified reception all the way
The signal that the remaining ultrasonic transducer for reception receives is carried out phase shift superposition, to be formed by signal on the basis of signal
Several signals being concerned with reference signal, and the signal after superposition is stored in MCU module according to the serial number of virtual focus point；
Step (6)：To each virtual focus point, the method that step (5) is respectively adopted carries out the pretreatment of signal, then utilizes peak value
The roads detection Fa Duige signal is detected, so that it is determined that whether going out each virtual focus point position with the presence of barrier and barrier
Azimuth information；
Step (7)：The digital signal for taking out the intermediate point stored in MCU module, carries out wavelet function feedback, utilization is small respectively
The multiscale space energy-distributing feature of wave conversion extracts reflected signal characteristic, in conjunction with svm classifier, to front obstacle
Attribute be identified；
Step (8)：Each virtual focus point signal after MCU module is superimposed phase shift scans and recognizes, and obtains near virtual focus point
The 2D digital pictures of barrier in setpoint distance；When weeder moves, MCU module obtains the virtual focus point at each moment successively
The 2D digital pictures of barrier, form the 3D rendering of barrier, the sky of barrier are obtained according to 3D rendering in neighbouring setpoint distance
Between shape and size in depth；
Step (9)：The attribute of barrier after identification, shape and size are notified into host computer.
8. being suitable for the detection of obstacles barrier-avoiding method of full-automatic weeder as claimed in claim 7, characterized in that the step
Suddenly in (2), ultrasonic phase array be circumferentially place respectively on four vertex of inscribed square four for reception ultrasound
Wave transducer, centre bit put a ultrasonic transducer for transmitting, and five ultrasonic transducers are all made of identical central
The transceiver piezoelectric ceramic transducer of frequency.The circumference is centered on O points, and weeder local width is that radius does circle.
9. being suitable for the detection of obstacles barrier-avoiding method of full-automatic weeder as claimed in claim 7, characterized in that
In the step (7), the wavelet function feedback is to carry out wavelet decomposition to signal and reconstruct high fdrequency component and low
Frequency component utilizes trained SVM classifier in advance using high fdrequency component and the energy ratio of low frequency component as signal characteristic
Classify to signal characteristic, so that it is determined that going out the property of front obstacle and being identified.
10. being suitable for the detection of obstacles barrier-avoiding method of full-automatic weeder as claimed in claim 7, characterized in that
The step (8), steps are as follows：
Step (81)：Sliding-model control is carried out to the section of scanning, i.e., section is divided into several discrete points, respectively it is described from
Scatterplot judges as virtual focus point to be scanned；
Step (82)：After scanning starts, from first specified discrete point as virtual focus point, according to the virtual focus point to ultrasound
It is each used for the differential time of flight of the ultrasonic transducer of reception in phased array, calculates each channel signal phase to be moved simultaneously
Signal averaging is carried out successively；
Step (83)：Barrier attribute is distinguished using grader；
Step (84)：The comprehensive information of each virtual focus point in a detection plane judges position and the 2D information of barrier；
Step (85)：With the movement of weeder, constantly newly-generated virtual focus point is scanned, to form front detection
The 3D location and shape of barrier in region.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810031611.0A CN108307767B (en) | 2018-01-12 | 2018-01-12 | Detection of obstacles obstacle avoidance system and method suitable for full-automatic weeder |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810031611.0A CN108307767B (en) | 2018-01-12 | 2018-01-12 | Detection of obstacles obstacle avoidance system and method suitable for full-automatic weeder |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108307767A true CN108307767A (en) | 2018-07-24 |
| CN108307767B CN108307767B (en) | 2019-09-24 |
Family
ID=62894299
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810031611.0A Expired - Fee Related CN108307767B (en) | 2018-01-12 | 2018-01-12 | Detection of obstacles obstacle avoidance system and method suitable for full-automatic weeder |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108307767B (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109250808A (en) * | 2018-10-25 | 2019-01-22 | 罗德斌 | A kind of sewage treatment machine people of environmental protection |
| EP3603370A1 (en) * | 2018-08-03 | 2020-02-05 | Lg Electronics Inc. | Moving robot, method for controlling moving robot, and moving robot system |
| CN110974091A (en) * | 2020-02-27 | 2020-04-10 | 深圳飞科机器人有限公司 | Cleaning robot, control method thereof, and storage medium |
| CN111179568A (en) * | 2019-12-25 | 2020-05-19 | 中电海康集团有限公司 | Mowing robot communication method and system based on boundary electromagnetic signals |
| CN111466165A (en) * | 2020-04-01 | 2020-07-31 | 信阳农林学院 | Multi-terrain weeding machine, control system and control method |
| US11178811B2 (en) | 2018-08-03 | 2021-11-23 | Lg Electronics Inc. | Lawn mower robot, system of lawn mower robot and control method of lawn mower robot system |
| CN113778084A (en) * | 2021-08-30 | 2021-12-10 | 普达迪泰（天津）智能装备科技有限公司 | Complex grassland environment intelligent obstacle avoidance system based on multispectral detection |
| EP3829831A4 (en) * | 2018-08-03 | 2022-08-31 | LG Electronics Inc. | Moving robot, moving robot system, and method for moving to charging station of moving robot |
| US11561275B2 (en) | 2018-08-03 | 2023-01-24 | Lg Electronics Inc. | Moving robot, method for controlling the same, and terminal |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5204814A (en) * | 1990-11-13 | 1993-04-20 | Mobot, Inc. | Autonomous lawn mower |
| CN101122585A (en) * | 2007-09-12 | 2008-02-13 | 天津大学 | Automatic identification method for supersonic phased array for detecting oil gas pipeline girth weld defect type |
| CN102147471A (en) * | 2011-03-01 | 2011-08-10 | 山东师范大学 | Front obstacle detecting system and method based on ultrasonic phased array |
| CN102621221A (en) * | 2012-04-16 | 2012-08-01 | 中国计量学院 | Defect classification method based on phased ultrasonic wave |
| CN102999986A (en) * | 2013-01-07 | 2013-03-27 | 山东师范大学 | Embedded invasion detection system and detection method based on ultraphonic phase array |
| US20170045612A1 (en) * | 2015-08-10 | 2017-02-16 | Deere And Company | Boundary signal detection |
-
2018
- 2018-01-12 CN CN201810031611.0A patent/CN108307767B/en not_active Expired - Fee Related
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5204814A (en) * | 1990-11-13 | 1993-04-20 | Mobot, Inc. | Autonomous lawn mower |
| CN101122585A (en) * | 2007-09-12 | 2008-02-13 | 天津大学 | Automatic identification method for supersonic phased array for detecting oil gas pipeline girth weld defect type |
| CN102147471A (en) * | 2011-03-01 | 2011-08-10 | 山东师范大学 | Front obstacle detecting system and method based on ultrasonic phased array |
| CN102621221A (en) * | 2012-04-16 | 2012-08-01 | 中国计量学院 | Defect classification method based on phased ultrasonic wave |
| CN102999986A (en) * | 2013-01-07 | 2013-03-27 | 山东师范大学 | Embedded invasion detection system and detection method based on ultraphonic phase array |
| US20170045612A1 (en) * | 2015-08-10 | 2017-02-16 | Deere And Company | Boundary signal detection |
Non-Patent Citations (2)
| Title |
|---|
| 厉彦忠: "基于STM32的超声相控阵导盲系统研究", 《现代电子技术》 * |
| 赵连睿: "超声相控阵自动测距系统的设计", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP3603370A1 (en) * | 2018-08-03 | 2020-02-05 | Lg Electronics Inc. | Moving robot, method for controlling moving robot, and moving robot system |
| US11178811B2 (en) | 2018-08-03 | 2021-11-23 | Lg Electronics Inc. | Lawn mower robot, system of lawn mower robot and control method of lawn mower robot system |
| US11266067B2 (en) | 2018-08-03 | 2022-03-08 | Lg Electronics Inc. | Moving robot, method for controlling moving robot, and moving robot system |
| EP3829831A4 (en) * | 2018-08-03 | 2022-08-31 | LG Electronics Inc. | Moving robot, moving robot system, and method for moving to charging station of moving robot |
| EP3829832A4 (en) * | 2018-08-03 | 2022-09-07 | LG Electronics Inc. | Moving robot, moving robot system, and method for moving to charging station of moving robot |
| US11561275B2 (en) | 2018-08-03 | 2023-01-24 | Lg Electronics Inc. | Moving robot, method for controlling the same, and terminal |
| US11910742B2 (en) | 2018-08-03 | 2024-02-27 | Lg Electronics Inc. | Moving robot, system of moving robot and method for moving to charging station of moving robot |
| CN109250808A (en) * | 2018-10-25 | 2019-01-22 | 罗德斌 | A kind of sewage treatment machine people of environmental protection |
| CN111179568A (en) * | 2019-12-25 | 2020-05-19 | 中电海康集团有限公司 | Mowing robot communication method and system based on boundary electromagnetic signals |
| CN110974091A (en) * | 2020-02-27 | 2020-04-10 | 深圳飞科机器人有限公司 | Cleaning robot, control method thereof, and storage medium |
| CN111466165A (en) * | 2020-04-01 | 2020-07-31 | 信阳农林学院 | Multi-terrain weeding machine, control system and control method |
| CN113778084A (en) * | 2021-08-30 | 2021-12-10 | 普达迪泰（天津）智能装备科技有限公司 | Complex grassland environment intelligent obstacle avoidance system based on multispectral detection |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108307767B (en) | 2019-09-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108307767B (en) | Detection of obstacles obstacle avoidance system and method suitable for full-automatic weeder | |
| CN109977813A (en) | A kind of crusing robot object localization method based on deep learning frame | |
| CN105184250B (en) | A kind of terrain classification method of electric power corridor airborne LiDAR point cloud data | |
| CN108828621A (en) | Obstacle detection and road surface partitioning algorithm based on three-dimensional laser radar | |
| CN111427031B (en) | Identity and gesture recognition method based on radar signals | |
| CN108375367A (en) | Combined ground laser radar and the work of oblique photograph point surveying method and system | |
| JP5507418B2 (en) | Tree position detection device, tree position detection method, and program | |
| CN101815308A (en) | WLAN indoor positioning method for neural network regional training | |
| CN109581385A (en) | Target locating set and method based on the wide bionical sonar of big ear bat ears | |
| JP2011123060A (en) | Identification-only optronic system and method for forming three-dimensional image | |
| CN101196562A (en) | Method for laser radar waveshape data decomposition based on improved EM algorithm | |
| Qin et al. | Stepwise decomposition and relative radiometric normalization for small footprint LiDAR waveform | |
| CN109407112A (en) | Top fruit sprayer volume measuring method and system based on LiDAR point cloud data | |
| McKerrow et al. | Plant acoustic density profile model of CTFM ultrasonic sensing | |
| CN104360384B (en) | Micro-seismic event localization method based on wave energy automatically scanning in length and breadth and device | |
| Debnath et al. | Applications of LiDAR in agriculture and future research directions | |
| CN109613552B (en) | Detection and analysis method for TLS multi-echo point cloud vegetation shielding target | |
| Wang et al. | A wavelet-based echo detector for waveform LiDAR data | |
| Tilly et al. | Evaluation of terrestrial laser scanning for rice growth monitoring | |
| CN110532975A (en) | TIN filter correction method based on Full wave shape airborne laser radar data | |
| CN110297248A (en) | Automatic data processing method based on multibeam sounding system | |
| Wagner et al. | Robust filtering of airborne laser scanner data for vegetation analysis | |
| CN109283491A (en) | A kind of unmanned plane positioning system based on vector probe unit | |
| McKerrow et al. | Classifying surface roughness with CTFM ultrasonic sensing | |
| McKerrow et al. | Recognising leafy plants with in‐air sonar |
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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20190924 Termination date: 20220112 |