CN2898895Y - Two-purpose indoor and outdoor guider of automatic carrying vehicle - Google Patents

Abstract

A Two-purpose indoor and outdoor guider of automatic carrying vehicle and zigbee module execution approach belongs to a guiding device of automatic carrying vehicle, which comprises a CCD camera in the front end of the carrying vehicle, a cartwheel decoder on the wheel, a zigbee module in the fixed position of site and in the carrying vehicle, and a central processor in the carrying vehicle. The utility model can finish the works such as: the guiding signs can be identified from the ground image, consequently the information of the carrying vehicle position and driving direction can be obtained; Dead-Reckoning can be executed by combining with the vehicle position information from the decoder; the guiding signs can be positioned and tracked on real time; the interference on the device by vehicle wobbling can be eliminated; a wireless communication units between the general controlling center and each carry vehicle are established to coordinate the interaction among the carrying vehicles and provide the forecast for crash prevention; the errors resulting from time accumulation in the decoder are corrected by the wireless positioning function in zigbee module etc. The guiding device is applicable to both indoor and outdoor environment with a high guiding positioning accuracy.

Description

The dual-purpose Automatic Guided Vehicle guidance device of indoor and outdoor

One, technical field

The implementation method of dual-purpose Automatic Guided Vehicle guidance device of indoor and outdoor of the present invention and zigbee module belongs to the guidance device of Automatic Guided Vehicle.

Two, background technology

The Automatic Guided Vehicle device is called for short AGVS, is the effective means of logistics transportation in current flexible manufacturing device (FMS) and the automated warehouse storage device.The nucleus equipment of Automatic Guided Vehicle device is Automatic Guided Vehicle (AGV), and as a kind of unmanned industrial transportation vehicle, AGV has promptly obtained application in 1950's.The general accumulator of using is as power, and dead weight capacity is from several kilograms to up to a hundred tons, and the work-yard can be office, workshop, also can be harbour, harbour.Modern AGV is by computer-controlled, and microprocessor is housed on the car.Most AGVS are furnished with device integrated control ﹠ management computing machine, are used for the operation process of AGV is optimized, and send the carrying instruction, the route of member during tracking transmits and control AGV.

Because the difference of indoor and outdoor surroundings causes the guidance device of Automatic Guided Vehicle that very big difference is also arranged.The Automatic Guided Vehicle of indoor environment is subjected to the restriction body in place less, and dead weight capacity is lighter, but very high to the accuracy requirement of location and guiding, guide mode commonly used has laser type, induction, rail mounted etc.; The Automatic Guided Vehicle body of outdoor environment is bigger, and dead weight capacity is big, and requirement can all weather operations, and guide mode commonly used has GPS, gyroscope, accelerometer etc.

These guidance devices are all very expensive, require the higher device of guidance accuracy for those, and the expense of device is high especially.And some guide mode will carry out transformation by a relatively large margin to the environment of Automatic Guided Vehicle operation, and invisible has the cost that has improved device.Need carrier to shuttle back and forth in indoor and outdoor environment for some, traditional Automatic Guided Vehicle guidance device has significantly exposed certain weak point.

Three, utility model content

The purpose of this utility model is to provide a kind of guiding, bearing accuracy height, can round-the-clockly be suitable for using under the two kinds of environment in indoor and outdoor the dual-purpose Automatic Guided Vehicle guidance device of indoor and outdoor with low cost.

Realize the dual-purpose Automatic Guided Vehicle guidance device of indoor and outdoor of above-mentioned purpose, be included in Automatic Guided Vehicle car body front end the CCD camera is installed, high-precision wheel demoder is installed on the wheel of car body, on fixed position that uses the place and carrier, the zigbee module is installed respectively, carry out mutual communication, central processing unit is placed on the carrier.Zigbee module on CCD camera, wheel demoder and the carrier links to each other with central processing unit respectively, the markings of device by finishing in advance on the CCD camera identification ground, obtain the positional information and the course information of carrier, obtain the displacement information of car body again by the wheel demoder, carry out dead reckoning with course information and positional information fusion, last central processing unit is passed to the control device of carrier, the work of guiding carrier in real time with the result.In whole process, device is also set up the wireless communication of total activation center and each carrier by the zigbee module, coordinate the inter-working of each carrier, simultaneously the wireless locating function of zigbee module can also correct to a certain extent the wheel demoder in time accumulation and the error that produces makes whole device finish whole work better.

The present invention has following beneficial effect:

The guiding position precision of device is about 5CM; Guiding course precision is about 0.5 degree; The guiding bearing accuracy is in 20cm; Be transferred to the control signal of control device more than 12 times p.s..

Device cost is well below the guidance device of other above-mentioned kinds.

Can round-the-clock occasion work at indoor and outdoors.

Be swift in response about 0.1 second during the indoor and outdoor surroundings conversion.

Little to the work-yard transformation, only get final product with a spot of guiding sign of picture.

Four, description of drawings

Fig. 1 is that the dual-purpose Automatic Guided Vehicle guidance device of indoor and outdoor is formed synoptic diagram.

Fig. 2 is video acquisition and processing module software configuration synoptic diagram.

Fig. 3 is lines identifications---a Hough conversion synoptic diagram.

Fig. 4 is the former figure of Hough conversion.

Fig. 5 is the result schematic diagram of Hough conversion.

Fig. 6 is the distribution schematic diagram of lines in the image.

Fig. 7 is the coordinate conversion synoptic diagram of image lines.

The first width of cloth logos of Fig. 8 right and wrong line following schematic diagram.

Fig. 9 is fuzzy its parametric line synoptic diagram, the maximum identification point number M of colour code ax curve, identification straight line number L curve, modified value t curve with it, Hough transformation parameter d (%) curve.

Figure 10 is a fuzzy device inference rule synoptic diagram.

Figure 11 is a micropower wireless network location synoptic diagram.

Label title among Fig. 1: 1 and 6, the wheel demoder, the 2.zigbee module, 3. central authorities handle, 4. video sign guide line, 5. carrier.

Five, embodiment

As shown in Figure 1, the dual-purpose Automatic Guided Vehicle guidance device of indoor and outdoor of the present invention comprises: the CCD camera, and the zigbee module, wheel demoder and central processing unit, wherein:

Zigbee module: mainly be placed on the carrier and plurality of fixed position, ground.The Zigbee modular power is little, and volume has a credit card-sized.

CCD camera: be placed in the dead ahead of carrier, the square road surface of 4M, field range the place ahead.The resolution about 1,000,000 of CCD.

Wheel demoder: be placed on the wheel in the place ahead of carrier.Price view apparatus permissible accuracy and deciding is very cheap.

Central processing unit: be placed on the carrier, realize with industrial computer at present, available DSP etc. makes the central processing unit of control panel after the commercialization, reduces cost.

Device is laid the CCD camera in Automatic Guided Vehicle body the place ahead, obtains car body the place ahead ground image in real time.High-precision wheel demoder is installed on wheel, is obtained the displacement situation of wheel in real time.The zigbee module is installed on some fixed positions in place and Automatic Guided Vehicle, is used for doing positioning correcting and communicates by letter.The communication data of view data, displacement data and zigbee imports the central processing system on the car body in real time into, and central processing system is finished following function:

Identification guiding sign (track lines, turnout selection marker) from the image on ground;

Locate the track homing sign in real time, and make the parameters of handling calculating reach best parameter setting apace by the method for fuzzy self-adaption;

Elimination is because of the interference of the shake of reasons such as Uneven road, engine generation;

The sign lines obtain course information from image, and and the displacement information of wheel demoder carry out behavior and calculate (DR), determine the position of car body in real time;

Detecting car body the place ahead from image has clear, and anti-collision early warning is provided;

Zigbee module automated communications on each carrier provide anti-collision early warning;

Zigbee on the carrier communicates by letter the calculation correction locating information with ground location zigbee;

Eliminate the accumulated time error of wheel DECODER information with the positioning correction information of zigbee;

Zigbee on the carrier communicates by letter with the total activation chamber, finishes the overall scheduling to carrier;

Control information after handling is passed in real time the control device of locomotive.

General DR device is made up of mileage gauge and gyroscope or magnetic compass, and not only cost is higher but use gyroscope, and the course information of gained can produce cumulative errors in time, can not guarantee very high positioning accuracy request in the special military occasion; And some complicated magnetic environment does not allow to use magnetic compass yet.This algorithm adopts cheap video frequency pick-up head to get the course information of auto levelizer, and the higher mileage gauge of service precision obtains the displacement information of vehicle simultaneously, finishes dead reckoning.The most important thing is: the machine vision itself that camera produces just can provide the route information of vehicle ', travels accurately with the route of guiding vehicle along regulation; Can be by route information and course information that machine vision produces along with the accumulated time error.So just fundamentally improved the accuracy of device.The information that vision and DR obtain utilizes serial ports to be transferred to controller of vehicle, just can finish the guiding to vehicle.

Shown in Figure 2 is video acquisition and processing module software architecture diagram.

The flow process of whole software compiles realization under the environment of VC++6.0, will describe one by one below.

(1) rim detection---enhancement mode Sobel operator.

The edge is the essential characteristic of image.Comprising valuable object boundary information in the edge, these information can be used for graphical analysis, Target Recognition and image filtering.Because edge and noise all in high-frequency range, are difficult to distinguish with frequency band in image, the task of rim detection seeks to suppress noise and the best compromise that improves marginal sharpness often.

In rim detection, a kind of algorithm commonly used is the sobel operator.Two templates are arranged, and one is the detection level edge. one is to detect vertical flat edge.As shown in the table:

1	2	1
			0	0	0
-1	-2	-1

-1	0	1
			-2	0	2
-1	0	1

B ₁ B ₂

For the better effects if that makes rim detection and increase the judgement of algorithm to straight line on the tilted direction, we have increased by two templates again, shown in the following form:

2	1	0
			1	0	-1
0	-1	-2

0	1	2
			-1	0	1
-2	-1	0

B ₃ B ₄

Specific algorithm is as follows:

1. the image that collects is placed in one 2 dimension group, and sets up a target array with original image array sky of a size;

2. each point in the traversal entire image array, when traversing some of image, the point on this point and 8 neighborhoods around it is formed a 3*3 matrix;

3. this matrix is dealt with processing mode with these four templates one by one: the absolute value of summation after two numbers on the relative position multiply each other;

4. compare this four number, the value of maximum is filled in the correspondence position of target image array;

5. the target array is gained.

Below part be the alternative scheme of algorithm, be applicable to that image will distinguish the markings and the shade of white, thereby get rid of the interference of shade markings identification.

In the link of edge extracting,, be increased to 4 templates the directivity edge of image is detected, as shown in Table 1 by improvement to the Sobel operator.Help better to extract the edge of markings like this.In addition when extracting rim value, unlike usual Sobel algorithm, ask the maximal value of all template result of calculation absolute values, but the absolute value of abandoning seeking template remains the maximal value and the minimum value of all templates respectively, obtains two width of cloth edge detection graph.Because markings are white in image, pixel value is bigger; The road surface is darker, and pixel value is less.In the calculating of rim detection, template calculated value＞0 of the left hand edge of markings, template calculated value＜0 of right hand edge has just comprised the left and right edges of markings respectively in resulting like this two width of cloth edge detection graph.As shown in Table 2.

Form 1

-1	-2	-1
			0	0	0
1	2	1

-1	0	1
			-2	0	2
-1	0	1

-2	-1	0
			-1	0	1
0	1	2

0	1	2
			-1	0	1
-2	-1	0

B ₁ B ₂ B ₃ B ₄

Form 2

Specific algorithm is as follows:

1. the image that collects is placed in two 2 dimension groups, and sets up two target arrays with original image array sky of a size;

2. each point in the traversal entire image array, when traversing some of image, the point on this point and 8 neighborhoods around it is formed a 3*3 matrix;

3. this matrix is dealt with processing mode with these four templates one by one: summation after two numbers on the relative position multiply each other;

4. this four number relatively, the correspondence position that the value and the minimum value (negative) of maximum filled in two target image arrays respectively;

5. the target array is gained.

(2) image segmentation---optimal threshold is cut apart.

Though what the resulting image of rim detection showed is the information at the edge of all things, still is gray level image.If directly discern calculating, calculated amount is still very big, so we will transform into the bianry image that has only black and white with gray level image by cutting apart of image.

At first, we want the grey level histogram of statistical picture.Set up an array.

The histogram of image with the method that the probability density function of two or more normal distributions is similar to, has been represented a kind of method that is referred to as optimal thresholdization.Threshold value is taken as from corresponding to the nearest gray-scale value in minimum probability place between two or more normal distribution maximal values, and consequently minimal error cuts apart.The step of this algorithm is as follows:

1. obtain the minimum and maximum gray-scale value Z in the image _lAnd Z _k, make T °=(Z of threshold value initial value _l+ Z _k)/2;

2. according to threshold value T ^kImage segmentation is become two parts of target and background, obtain the average gray value Z of two parts _OAnd Z _B:

$Z_O=\frac{\underset{z(x,y)<T^k}{\mathrm{\&Sigma;}}z(x,y)\&times;N(x,y)}{\underset{z(x,y)<T^k}{\mathrm{\&Sigma;}}N(x,y)}$

$Z_B=\frac{\underset{z(x,y)>T^k}{\mathrm{\&Sigma;}}z(x,y)\&times;N(x,y)}{\underset{z(x,y)>T^k}{\mathrm{\&Sigma;}}N(x,y)}$

In the formula z (x, y) be on the image (x, gray-scale value y), N (x, y) be (x, weight coefficient y), General N (x, y)=1.0;

3. obtain new threshold value: T ^K+1=(Z _O+ Z _B)/2;

4. if T ^K=T ^K+1, then finish, otherwise K ← K+1 changes step 2;

5. after obtaining threshold value, original image is transformed into bianry image according to threshold value.Pixel value greater than threshold value is 255; Pixel value less than threshold value is 0.

(3) identification of lines---Hough conversion

The Hough conversion is used for searching straight line in image.Its principle is very simple: suppose to have one to be s with the initial point distance, deflection is the straight line of θ, as shown in Figure 3.Every bit on the straight line all satisfies equation

s＝xcosθ+ysinθ

Utilize this fact, we can find out certain bar straight line.To provide one section program below, be used for finding out straight line (see figure 4) the longest in the image.Find two end points of straight line, between them, connect the straight line of a redness.In order to see effect clearly, the result is retouched into thick line, as shown in Figure 5.

What find as can be seen, is the longest straight line really.Method is, opens a two-dimensional array as counter, and first dimension is an angle, and second dimension is a distance.The ultimate range that elder generation's calculating may occur is Be used for determining the size of array second dimension.For each black color dots, the variation range of angle is (in order to reduce storage space and computing time, angle increases by 1 ° at every turn) from 0 ° to 179 °, obtains corresponding coming apart from s by equation, and corresponding array element [s] [θ] adds 1.Open an array Line simultaneously, calculate two end points up and down of every straight line.After all pixels have all been calculated, finding in the array element maximumly, is exactly that the longest straight line.The end points of straight line can find in Line.Though what be noted that we handle is binary map, is actually 256 grades of gray-scale maps, but only used 0 and 255 two kind of color.

Specific algorithm is as follows:

1. transform domain of initialization, the array in [s] [θ] space, the pixel count on the quantification number image diagonal direction on the s direction, the quantification number on the θ direction is 180;

2. all marginal points in the sequential search image to each marginal point, add 1 on the each point of the correspondence of transform domain;

3. obtain maximal value and record in the transform domain;

4. this peaked d% of set-up and calculated according to parameter is worth Y as reference;

5. the each point on the ergodic transformation territory, if this value greater than Y just with its taking-up, other point does not deal with;

6. obtain utmost point footpath and polar angle (s, θ) on its transform domain according to the position of point of record;

7. try to achieve its parameter under rectangular coordinate system according to (s, θ).

Below part be the alternative scheme of algorithm, be applicable to that image will distinguish the markings and the shade of white, thereby get rid of the interference of shade markings identification.

The original image of cutting apart through image is processed into two width of cloth bianry images, includes the left and right marginal information of markings respectively.Device comes the left and right edges information of distinguishing mark line by the Hough conversion.The midpoint of setting on the image base is polar limit, and bianry image is converted into polar coordinate system from rectangular coordinate system.Point in so original image on certain bar straight line changes into lines behind the polar coordinates will pass through all that identical a bit (S, θ), S and θ represent utmost point footpath and the polar angle of this straight line under polar coordinate system respectively.If a statistics is done by each some institute in the polar coordinates through the number of times of lines, so by the maximum point (S of process number of times _Max, θ _Max) utmost point footpath and the polar angle polar coordinates characteristic that is exactly straight line the longest in the image, this straight line is exactly the left and right edge of markings.Specific algorithm is as follows:

1. transform domain of initialization, set up polar coordinates two-dimensional space array, span on the S direction is done the utmost point maximal value directly that can show from 0 to image, i.e. distance from image base mid point to summit, the right, the step-length that quantizes is a p pixel, on the θ direction scope of value from 0 spend to 180 the degree, its quantization step is the q degree;

2. all marginal points in the sequential search image to each marginal point, add 1 on the each point of the correspondence of transform domain;

3. obtain maximal value and record in the transform domain;

4. this peaked d% of set-up and calculated according to parameter is worth Y as reference;

5. the each point on the ergodic transformation territory, if this value greater than Y just with its taking-up, other point does not deal with;

According to the position of point of record obtain utmost point footpath on its transform domain and polar angle (S, θ);

7. according to (S θ) tries to achieve its parameter under rectangular coordinate system.

Attention: p, q, d can carry out real-time adjustment according to the quality of image recognition, and details sees below.Initial value: p=2; Q=1; D=70.

Through the Hough conversion, device can extract several lines (the d value is relevant for the number of lines and the precision of parameter and p, q) respectively from two width of cloth bianry images.Its parameter is averaged respectively and can be obtained the accurate parameters on border, the markings left and right sides, and these precise parameters are averaged and can be obtained the accurate parameters of markings center line.

Like this, the polar angle of the polar angle of lateral marker line and vertical markings is vertical, is two line segments about 90 degree if the similar polar angle difference of sharpness appears in device when following the tracks of.This lateral marker line appearred in key diagram picture constantly.

Because the gray-scale value of markings is with respect to the ground height, the gray-scale value of shade is low with respect to ground.If then the left hand edge image of device analysis is positioned at the left side of right hand edge image, illustrate that this edge is the edge of markings; Otherwise, be the shade edge.

(4) tracking of image is handled.

All line informations of cutting apart taking-up of image have a lot of interference.These lines mainly contain following a few class, as shown in Figure 6.The rectangular coordinate plane of image pixel point value and their position process is converted to the polar coordinates plane.As shown in Figure 7, pole location is in the base midpoint, i.e. the position at car place.

Need during first width of cloth Flame Image Process all to handle, from figure, find out straight line in all lines, and the parameter information of straight line is noted, be stored in the public variable group.The information of depositing in the public variable group is the S (being accurate to 1 pixel) of every straight line in the image, θ (being accurate to 1 °), and D (being accurate to 1 pixel), the implication of its expression is as shown in Figure 7.

After first width of cloth Flame Image Process was finished, subsequent images just can be carried out line tracking.Concrete processing procedure: from the public variable group, take out the position of going up the piece image cathetus, this position in this width of cloth image and on every side among a small circle in search for straight line, again the result is passed in the public variable group.As shown in Figure 8.

Do like this and both reduced calculated amount, make the real-time of device be improved, and reduced the interference of the unnecessary barrier of image boundary part.

(5) image separation threshold adaptive feedback adjusting.

Though device can access position and drift angle, the identification quality of markings and the instability of comparatively accurate markings by aforementioned algorithm.The drift angle of straight line still is more stable, and error remains in 0.5 degree.But the precision of the center position of markings is swung in the scope of 1-10 pixel; When carrying out the markings tracking, every width of cloth treatment of picture time also swings in 0.06 second-0.12 second scope.According to one's analysis, causing the unsettled reason of device identification quality is because CCD obtains the instability of picture quality.Because in some military environment, vehicle carries out all weather operations.The shade of the reflective and various shelters on the sunlight on daytime, the light in evening, rainy day ground etc. all can make the brightness of ground image and sharpness be affected, thereby directly causes the instability of markings identification quality.Find that through after a large amount of experiments parameter T in the algorithm and the value of d directly affect the computing time of every width of cloth image and the quality of identification.Therefore, this algorithm design a data negative-feedback process module, assess the quality of every width of cloth image recognition quality in real time, and information feedback returned, adjust the value of T and d, make device in next width of cloth image recognition, reach effect preferably.Circulate down with this, make device reach best at short notice.The module of this data feedback realizes with the mode of fuzzy device: input parameter is maximum identification point number M ax, identification straight line number L; Output parameter is the modified value t of Hough transformation parameter d and threshold value.Membership function be respectively trapezoidal membership function ftrap (x, a, b, c, d) and Triangleshape grade of membership function ftri (x, a, b, c), concrete representation is as follows, each parametric line of fuzzy device and fuzzy inference rule are shown below:

$\mathrm{ftrap}(x,a,b,c,d)=\left\{\begin{array}{cc}0& x\&le;a\\ (x-a)/(b-a)& a\&le;x\&le;b\\ 1& b\&le;x\&le;c\\ (d-x)/(d-c)& c\&le;x\&le;d\\ 0& x\&GreaterEqual;d\end{array}\right.---\left(1\right)$

$\mathrm{ftri}(x,a,b,c)=\left\{\begin{array}{cc}0& x\&le;a\\ (x-a)/(b-a)& a\&le;x\&le;b\\ (c-x)/(c-b)& b\&le;x\&le;c\\ 0& x\&GreaterEqual;c\end{array}\right.---\left(2\right)$

Each parametric line of fuzzy device as shown in Figure 9, fuzzy inference rule is as shown in figure 10.

Like this, the threshold value T of the threshold value T+ threshold value modified value t=Fig. 8 width of cloth image in this width of cloth image of Fig. 7 cooperates Hough transformation parameter d again, and the parameter in the correction image algorithm that just can be real-time makes the quality of device identification keep stable.Simultaneously, because threshold value modified value t arranged, can every width of cloth image all use the method calculated threshold of optimal threshold, only be used in first width of cloth image and calculate once, after this t that utilizes this fuzzy device to produce in the image calculation revises, and has further improved the speed of calculating.

The elimination of shake

Through the processing of previous step, can obtain the precise parameters information of markings in the image.But said here precise parameters information is at the image that obtains, and that is to say, if Flame Image Process is resulting when AGV is actionless.But when AGV moves, owing to can cause the slight shake of car body under the situation of locomotive engine or Uneven road, owing to the CCD camera is connected on the car body, so the interference that the image that CCD obtained is also inevitably shaken.The function of this module is exactly to eliminate the interference that identification caused that lines are given in shake.

So, the shake of CCD only can cause some interference to the relevant horizontal information of lines parameter, though little, asking accurate angle to consider still it to be eliminated better.In addition, the parameter final purpose of identification lines is to use for the control device for AGV, if the center point of distinguishing mark line always repeatedly about beat, brought inconvenience also for the control of locomotive, because what the control of locomotive needed is level and smooth stable control, do not allow continuous sudden change.

The specific algorithm of eliminating shake is as follows:

1. the value of setting a parameter D is adjustable;

2. the absolute value of the difference of the center point coordinate of the markings of the center point coordinate of the markings of this identification and last time identification is C;

3. if C＞D, the parameter of this identification is effective;

4. if C＜D, the parameter of this identification is invalid, but keeps a record;

5. in the image processing process next time, if the point of identification and the invalid point of last identification are moving in the same way, then this identification effectively;

6. in the image processing process next time, be reverse moving if the point of identification and last time are discerned invalid point, and the distance＞D that moves, then this identification is effective;

7. in the image processing process next time, be reverse moving if the point of identification and last time are discerned invalid point, and the distance＜D that moves, then this identification is invalid.

(6) novel dead reckoning (DR) method.

Dead reckoning (DR) is a kind of autonomous land vehicle localization method.It utilizes range information and angle information to calculate and determines the current position of vehicle with the initial point of ground point as local coordinate system.The motion of surface car can be regarded the two dimensional surface motion as, according to kinematic principle, knows the starting point and the angle, initial heading of vehicle, by the operating range of real-time measuring vehicle and the variation of course angle, just can extrapolate the positional information of vehicle.Suppose that under local horizontal coordinates the initial position of vehicle is (x ₀, y ₀), angle, initial heading θ ₀, collect operating range S every regular hour T _iWith course angle θ _i, then vehicle position at any time can be provided by following formula:

$\left\{\begin{array}{c}{x}_k=x_0+\underset{i=1}{\overset{k}{\mathrm{\&Sigma;}}}{S}_i\&CenterDot;\cos {\mathrm{\&theta;}}_{i-1}\\ y_k=y_0+\underset{i=1}{\overset{k}{\mathrm{\&Sigma;}}}{S}_i\&CenterDot;\sin {\mathrm{\&theta;}}_{i-1}\end{array}\right.$

Present applied DR device all is to be heading sensor with gyroscope or magnetic compass, is displacement transducer with mileage gauge or wheel demoder, and operating range S is provided respectively in real time _iWith course angle θ _iGyroscope not only costs an arm and a leg, and error must just can obtain course angle information by integration along with accumulated time, and calculated amount is bigger, is unsuitable for all weather operations.In addition, there are many magnetic environments in some special military occasion, also should not be with magnetic compass as heading sensor; Also inapplicable GPS assists the DR algorithm: the one, because military occasion can not rely on GPS, and the 2nd, when in cavern and tunnel, carrying for a long time because of vehicle, the GPS no signal.

This algorithm adopts cheap CCD camera as heading sensor, and high precision wheel demoder well solves these problems as displacement transducer.By aforementioned algorithm, device can obtain the needed course information of DR algorithm in real time, and precision is 0.5 degree, and frequency is 12Hz.Device can guarantee that vehicle advances along set traffic route, is not subjected to the influence of environment.The illumination that only relies on headlight obtains with regard to the information that is enough to finish vision, guaranteed that vehicle is at night with do not have in the tunnel of illumination and do not work normally.

(7) the concrete implementation step of zigbee module:

A. in the needs locating area, setting up one by control center and some FFD (full-function device), the wireless network communication apparatus of composition with fixed position. each FFD is exactly a network node.On locating area figure, each location of network nodes is labelled.As shown in figure 11:

Full-function device (Full function device is hereinafter to be referred as FFD) can be supported any topological structure, can be used as network negotiate person and common negotiator, and can communicate with any equipment.

B. the self-navigation vehicle technology AGV to the needs location identifies, and goes up installation RFD (subduing function device) at self-navigation vehicle technology AGV (hereinafter to be referred as AGV);

Cut down function device (Reduced function device is hereinafter to be referred as RFD), only support hub-and-spoke configuration, can not become any negotiator, can communicate, realize simple with network negotiate person.RFD can only communicate by letter with FFD, can not communicate by letter with other RFD, but the circuit of their inside lacks than FFD, has only seldom or do not have catabiotic internal memory.

The installation of FFD must guarantee carrier when certain fixing point of process sign track, and the RFD on the carrier can receive at least 3 FFD signals (communication range between FFD and the RFD is 10m), and the interference of barrier can not be arranged.Like this, just can position by the method for signal attenuation location.

C. set up between the RFD on network node FFD and the AGV, and writing to each other between network node FFD and the control center;

D. whether receive the positioning signal that certain AGV goes up RFD according to network node FFD, and the intensity size of the positioning signal that receives, determine the position of this moving target from the position of interdependent node, and will send RFD on the AGV to by the FFD around the AGV the control signal of vehicle;

The zigbee equipment of setting up such network should have some following characteristics.

RFD on each AGV can both send the signal of the same intensity of regulation.The content of this signal should be the ID sign indicating number of this AGV, and this is the most basic.Can also add in the time of subsequent development on the coordinate, car of the destination that vehicle will be advanced whether information such as container are arranged;

Described control center computing machine stores the address of each FFD, the identity identification information of each moving target and the network management software;

Described network node FFD should have positioning signal intensity Presentation Function (RSSI);

FFD in the described network should have the function of automatic transfer from the information of other FFD and RFD.

Claims (1)

1, the dual-purpose Automatic Guided Vehicle guidance device of a kind of indoor and outdoor, it is characterized in that: the CCD camera is installed at Automatic Guided Vehicle car body front end, high-precision wheel demoder is installed on the wheel of car body, on fixed position that uses the place and carrier, the zigbee module is installed respectively, carry out mutual communication, central processing unit is placed on the carrier; Zigbee module on CCD camera, wheel demoder and the carrier all links to each other with central processing unit.

Images