CN102145808A - Industrial laser guidance AGV double-closed-loop control system and control method thereof - Google Patents

Abstract

The invention discloses an industrial laser guidance automatic guided vehicle (AGV) double-closed-loop control system and a control method thereof. The system consists of a detection module and a control module and is characterized in that: the detection module consists of an encoder serving as a motor position sensor and a laser scanner serving as an AGV position sensor; the control module consists of a servo driver and an industrial personal computer and is a double-position closed loop control module consisting of an inner position ring and an outer position ring; the inner position ring is a motor intersection angle position ring; the outer position ring is used for reading an AGV position signal by using the laser scanner; the path tracking error of the current position of an AGV is calculated by the AGV position signal fed back from the outer position ring; and the path tracking error is used as the input quantity of the control system. By the system and the method, the independent running function of the industrial laser guidance AGV can be well realized, and the rapidity and the stability of the path tracking of the AGV can be ensured.

Description

Double closed-loop control system and the control method of industrial laser guiding AGV

Technical field

The invention belongs to Robotics and automatic field, particularly a kind of operation is in control system and the path tracking control method thereof of the industrial laser guiding AGV of the automatic navigation transport trolley of automated workshop.

Background technology

Laser guiding AGV (AGV, Automatic Guided Vehicle) is a kind of,, can be fit to the pahtfinder hard of multiple environment owing to the accuracy of positioning height with the automatic navigation transport trolley of laser as guide mode, can change driving path fast, be used widely at present.Control system is the core content of AGV, and path tracking control method is the key point that guarantees the AGV control accuracy.

The control system of present most of laser guiding AGV manufacturer all is that the NDC company from Sweden introduces, and its complete control system costs an arm and a leg.

Existing disclosed experiment type AGV system is single closed loop control system, owing to used driving systems such as retarder in the industrial AGV actuating device, all there is the error problem that is caused by the gap in these driving systems, add that again AGV is a wheeled mobile robot, exist nonholonomic constraint between wheel and the ground, closed loop control system is put by the unit that historical facts or anecdotes is tested type AGV system certainly will have problems such as dynamic response difference when being used for industrial laser guiding AGV, produce very big path tracking error.

Aspect path tracking control method, at present the most frequently used is fuzzy control method.Fuzzy control has the characteristics of robustness and good stability, be applicable to the control of complicated nonlinear systems, a lot of scholars by AGV is carried out modeling and simulating or by experiment type AGV algorithm is verified the result shows that this method is one of control effect the best way in the path tracking control method.But conventional FUZZY ALGORITHMS FOR CONTROL is being applied to the slow-footed phenomenon of path trace occurred when industrial laser guiding AGV path trace is controlled, especially when AGV walked under fast state, it is more obvious that delay effect and system response show slowly.This is because practical in industry AGV system complex, weight is big, inertia is big, so compare with general experiment type AGV, it is much bigger to control difficulty, and is difficult to reach due precision.Because the time stickiness of servo motor driving system response has influenced system's governing speed, make it to fail quick adjustment in addition to stabilized conditions.When speed was higher, because the AGV walking is very fast, delay effect can make the more difficult of accurate tracking predefined paths change.

Summary of the invention

The present invention is for avoiding above-mentioned existing in prior technology weak point, double closed-loop control system and the path tracking control method thereof of a kind of industrial laser guiding AGV are provided, guide the autonomous walking function of AGV and improve the rapidity and the stability of AGV path trace in the hope of realizing industrial laser better.

Technical solution problem of the present invention adopts following technical scheme:

The two-position closed loop control system of the industrial laser guiding of the present invention AGV is made up of detection module and control module, and its constructional feature is:

Described detection module is by constituting as the coder of motor position sensor with as the laser scanner of AGV pose sensor;

Described control module is made of servo-driver and industrial computer, and it is the two-position closed loop control module that is made of interior position ring and external position ring that described control module is set;

Position ring is a motor angle position ring in described, detects motor angle position signal by coder, to servo-driver, is electric-motor drive unit with the servo-driver with described motor angle position signal feedback;

Described external position ring is to read AGV pose signal with laser scanner, and described AGV pose signal feedback is to industrial computer, and industrial computer outputs signal to the servo-driver signal input part, and motor rotates output and drives AGV walking steering hardware by retarder; The path trace error of the AGV pose calculated signals AGV current location of feeding back in the position ring in addition is with the input of described path trace error as control system.

The characteristics of the path tracking control method of the industrial laser guiding of the present invention AGV control system are to adopt fuzzy prediction control algorithm, in the rectangular coordinate system on AGV motion plane, place, the pose of AGV is expressed as (x, y, α), wherein (x y) is the position coordinate of AGV in described rectangular coordinate system, α is the azimuth of described AGV, and described azimuth is the central axis of AGV and the angle of x axle forward;

The object pose that described AGV is set is (x ', y ', α '), and detecting the starting point pose that obtains AGV by laser scanner is (x ₀, y ₀, α ₀); Cook up from described starting point pose (x according to the associated pathway planning algorithm ₀, y ₀, α ₀) to the theoretical path L of object pose (x ', y ', α ');

Error between definition attained pose and the object pose is a position and attitude error, and described position and attitude error comprises normal direction positional error e _PnWith azimuth error e _α

Described path following method is finished by following steps:

A, the attained pose that detects AGV by laser scanner are (x _A, y _A, α _A), the azimuth error e of AGV then _αWith normal direction positional error e _PnBe respectively:

e _α＝α _A-α′ (1)

$e_{\mathrm{pn}}=k\·\frac{|x_A\tan {\mathrm{\α}}^{\′}-x^{\′}\tan {\mathrm{\α}}^{\′}+y^{\′}-y_A|}{\sqrt{1+{\left(\tan {\mathrm{\α}}^{\′}\right)}^2}}---\left(2\right)$

Set: when the AGV attained pose is positioned at the right side of theoretical path L, k=1; When the AGV attained pose is positioned at the left side of theoretical path L, k=-1;

The normal direction positional error e of b, employing differential algorithm correction AGV _PnWith azimuth error e _α, obtain revised law to positional error e _Pn0With the corrected azimuth error e _{α 0}Be respectively:

${\mathrm{<figure-callout\; id="0"\; label="e\; pn"\; filenames="HDA0000045610990000021.png,HDA0000045610990000031.png"\; state="\{\{state\}\}">e</figure-callout>}}_{\mathrm{pn}}=e_{\mathrm{pn}}+\frac{{\mathrm{de}}_{\mathrm{pn}}\left(t\right)T_d}{\mathrm{dt}}---\left(3\right)$

$=e_{\mathrm{pn}}+\frac{(e_{\mathrm{pn}}\left(i\right)-e_{\mathrm{pn}}(i-1))T_d}{T}$

$e_{\mathrm{\&alpha;}0}=e_{\mathrm{\&alpha;}}+\frac{{\mathrm{de}}_{\mathrm{\&alpha;}}\left(t\right)T_d}{\mathrm{dt}}$ (4)

$=e_{\mathrm{pn}}+\frac{(e_{\mathrm{\&alpha;}}\left(i\right)-e_{\mathrm{\&alpha;}}(i-1))T_d}{T}$

In the formula: T _dBe derivative time constant, T is the sampling period, e _Pn(i) be the normal direction position error signal of AGV current sampling point pose, e _α(i) be the azimuth error signal of AGV current sampling point pose, e _Pn(i-1) be the normal direction position error signal of the last sampling point pose of AGV, e _α(i-1) be the azimuth error signal of the last sampling point pose of AGV, e _PnAnd e _αThe sampling point difference DELTA of adjacent moment _PnAnd Δ _αBe respectively:

Δ _pn＝e _pn(i)-e _pn(i-1)；

Δ _α＝e _α(i)-e _α(i-1)；

C, try to achieve AGV working control amount β by FUZZY ALGORITHMS FOR CONTROL

At first with the revised law of AGV to positional error e _Pn0With the corrected azimuth error e _{α 0}Multiply by normal direction positional error fuzzy quantization factor k respectively _PWith azimuth error fuzzy quantization factor k _αTransform to universe of fuzzy sets scope separately, obtain the normal direction positional error input e ' of fuzzy controller _PnWith azimuth error input e ' _α:

e′ _pn＝e _pn0k _p

(5)

＝e _pnk _p+Δ _pnk _dp

e′ _α＝e _αk _α

(6)

＝e _αk _α+Δ _αk _dα

Wherein, k _DpAnd k _{D α}Be Prediction Parameters, and k _Dp=T _dk _p/ T, k _{D α}=T _dk _α/ T;

With normal direction positional error input e ' _PnWith azimuth error input e ' _αCarry out obfuscation according to membership function; Infer output Δ β according to Mamdani fuzzy reasoning method, fuzzy control rule and average weighted clarification method; Then working control amount β is: β=Δ β k _β(7)

Wherein, k _βFor fuzzy factor of proportionality, be the input control signal of industrial computer with working control amount β.

The characteristics of path tracking control method of the present invention also are:

Described fuzzy factor of proportionality k _β, Prediction Parameters k _DpAnd k _{D α}Be to adjust in real time according to the variation of the AGV speed of travel; Normal direction positional error fuzzy quantization factor k _PWith azimuth error fuzzy quantization factor k _αUnder any speed of travel of AGV, be fixed value, determine as follows:

A, make k _Dp=0, k _{D α}=0, determine k _P, k _αAnd k _βValue:

The normal direction positional error input e ' of fuzzy controller _PnBasic domain be [0.05,0.05], measure unit is m, its universe of fuzzy sets is [6,6]; The azimuth error input e ' of fuzzy controller _αBasic domain be that [0.1,0.1], measure unit are rad, universe of fuzzy sets is [6,6]; The basic domain of working control amount β is that [10,10], measure unit are degree, and universe of fuzzy sets is [1,1];

K then _P, k _αAnd k _βInitial value be respectively:

${\mathrm{<figure-callout\; id="0"\; label="k\; p"\; filenames="HDA0000045610990000021.png,HDA0000045610990000031.png"\; state="\{\{state\}\}">k</figure-callout>}}_p=\frac{6}{0.05}=120;$ $k_{\mathrm{\&alpha;}0}=\frac{6}{0.1}=60;$ $k_{\mathrm{\&beta;}0}=\frac{10}{1}=10;$

With described initial value k _P0, k _{α 0}And k _{β 0}Be the basis, determine k according to experiment effect _P, k _α, k _βValue:

At first make AGV with speed v=10m/min low speed walking, k is set _P=k _P0, k _α=k _{α 0}Remain unchanged, with k _βValue is at [k _{β 0}-10, k _{β 0}+ 5] change in the domain scope, finding and making the AGV path trace error minimum and the fastest value of regulating the speed is a ₀Make k then _β=a ₀Remain unchanged k _PAnd k _αOne of them keeps initial value constant, and another one changes in its domain scope, finds out to make the AGV path trace error minimum and the fastest class value of regulating the speed be respectively k _p=b ₀k _α=c ₀Obtaining AGV, to control best one group fuzzy factor of effect under lower-speed state be k _p=b ₀k _α=c ₀k _β=a ₀Wherein said k _PThe domain scope be

k _αThe domain scope be

Keep k _p=b ₀And k _α=c ₀Constant, change the speed of travel of AGV, make AGV respectively with 20,30,40,50, the speed walking of 60m/min, k is set _βValue is at [a ₀-10, a ₀+ 5] change in the domain scope, according to

Relation, find out and under each speed, make the minimum and the fastest k that regulates the speed of path trace error _βValue;

B, determine Prediction Parameters k _DpAnd k _{D α}:

Make AGV respectively with 10,20,30,40,50, the speed walking of 60m/min, the value of the fuzzy factor under each speed remains unchanged respectively, makes k _Dp=k _pD ₀k _{D α}=k _αD ₀, and set gradually d respectively ₀=1,2,3,4 ..., contrast d ₀The control effect of AGV when getting different values finds one group to make the AGV path trace error minimum and the fastest prediction of regulating the speed count k _Dp=b ₂k _{D α}=c ₂

Try to achieve k under other speed of travel of AGV by method of interpolation _β, k _Dp, k _{D α}Value.

Compared with the prior art, beneficial effect of the present invention is embodied in:

1, control module of the present invention is set to double closed-loop control system, puts closed loop control system with the unit of disclosed experiment type AGV and compares, and it is fast to have the error governing speed, the characteristics of AGV smooth running, and can be applicable to industrial product.

Ring is the motor angle position ring of semiclosed loop in 2 control modules of the present invention, because its electric part and mechanics are relatively independent, can adopt higher position gain, and system is easily adjusted, and dynamic response is fast; Outer shroud directly obtains the attained pose signal of AGV by laser scanner, can guarantee the control accuracy of higher AGV and follow speed by path following method;

3, AGV path tracking control method of the present invention is on the conventional FUZZY ALGORITHMS FOR CONTROL of AGV basis, utilize computer sampling control that the AGV control system adopts and differential algorithm to have the characteristics of stickiness and forecasting power when reducing, provide a kind of new fuzzy prediction control algorithm, solve the system delay problem preferably, improved the rapidity of path trace.

4, the present invention is according to the characteristics of AGV when the real work, and the path tracking control method of fuzzy control is improved, and the AGV speed of travel of giving chapter and verse is adjusted fuzzy factor of proportionality k in real time _β, factor of proportionality was little when speed was high, and factor of proportionality was big when speed was low, can effectively guarantee the AGV smooth running, adjusted tracking error in the short period of time, had improved the path trace precision of AGV.

5, the present invention is the control system of a kind of industrial laser guiding AGV, has economic, practical characteristics with respect to expensive NDC control system, and can obtain excellent control effect;

Description of drawings

Fig. 1 is an applied laser guiding AGV system schematic in the embodiment of the invention;

Fig. 2 is laser guiding AGV two-position closed loop control system figure of the present invention;

Fig. 3 is an AGV tracking error scheme drawing of the present invention;

Fig. 4 is laser guiding AGV fuzzy prediction control algorithm figure of the present invention;

Fig. 5 is the membership function figure of the applied fuzzy quantization factor in the embodiment of the invention;

The v=60m/min that Fig. 6 the inventive method obtains, k _Dp=480, k _DaThe error control design sketch of=720 o'clock laser guiding AGV;

Table 1 is the applied fuzzy control rule table of the embodiment of the invention;

Table 2 is the k under the applied part moving velocity of the embodiment of the invention _β, k _Dp, k _{D α}Value;

Table 3 is the applied experimental program table of the embodiment of the invention.

The specific embodiment

The laser guiding AGV that present embodiment adopts is V shape AGV behind the single wheel drive type that is formed by the fork truck transformation, wheelbase 1300mm conducts oneself with dignity 1 ton, and specified lifted load can reach 800kg, its front-wheel be drive wheel be again wheel flutter, each is by an AC Servo Motor Control.As shown in Figure 1, AGV is made up of car body, safety anticollision system, walking steering swivel system, communication system, two-position closed loop control system and other ancillary system.Wherein safety anticollision system adopts the S3000 laser-scan safety device of German SICK company, and this device is fixed on the front end of AGV car body, links to each other with industrial computer by connection.The walking steering swivel system is made up of servomotor, retarder, walking steering hardware.The WLAN that communication system is made up of wireless network card and wireless router can be realized the communication between AGV and the upper computer smoothly.

As shown in Figure 2, the two-position closed loop control system in the present embodiment is made up of detection module and control module;

Detection module is by constituting as the coder of motor position sensor with as the laser scanner of AGV pose sensor;

Control module is made of servo-driver and industrial computer, and it is the two-position closed loop control system that is made of interior position ring and external position ring that control module is set; Wherein, interior position ring is a motor angle position ring, detects motor angle position signal by coder, and motor angle position signal feedback is arrived servo-driver, is electric-motor drive unit with the servo-driver; The external position ring is to read AGV pose signal with laser scanner, and AGV pose signal feedback is to industrial computer, and industrial computer outputs signal to the servo-driver signal input part, and motor rotates output and drives AGV walking steering hardware by retarder;

Interior position ring is used for controlling the accurate operation of motor, because its electric automatic control part is relatively independent with the AGV actuating unit, can adopt higher position gain, and system is easily adjusted, and response is fast, guarantees the dynamic characteristics of system; Can make motor performance reach best by regulating each gain parameter of motor.

Outside in the position ring, checking system adopts the NAV200 laser orientation system of German SICK company, and laser scanner is installed on the AGV car body, by the laser-bounce plate in the continuous scanning surrounding environment, can directly calculate the attained pose signal of AGV.The external position ring is mainly used in the control of steady state error aspect, guarantees the followability of system.Because it has comprised transmission devices such as some non-linear factors such as retarder, so control accuracy mainly guarantees by path tracking control method.The path trace error of the AGV pose calculated signals AGV current location of feeding back in the position ring in addition is with the input of path trace error as control system.

In the rectangular coordinate system on AGV motion plane, place, the pose of AGV is expressed as (α), wherein (x y) is the position coordinate of AGV in rectangular coordinate system, and α is the azimuth of AGV, and the azimuth is the central axis of AGV and the angle of x axle forward for x, y;

As shown in Figure 3, the object pose that AGV is set is (x ', y ', α '), and detecting the starting point pose that obtains AGV by laser scanner is (x ₀, y ₀, α ₀); Cook up from starting point pose (x according to the associated pathway planning algorithm ₀, y ₀, α ₀) to the theoretical path L of object pose (x ', y ', α ');

Error between definition attained pose and the object pose is a position and attitude error, and position and attitude error comprises normal direction positional error e _PnWith azimuth error e _α, rule of thumb, can eliminate this two errors by the deflection angle β that adjusts the AGV front-wheel.

Present embodiment has adopted fuzzy prediction control algorithm as the AGV path tracking control method, and this algorithm is to merge differential algorithm on the basis of conventional FUZZY ALGORITHMS FOR CONTROL, and according to the fuzzy factor of proportionality of AGV speed of travel automatic compensation.Fig. 4 has shown the fuzzy prediction control algorithm figure of AGV, and the specific implementation step is as follows:

A, the attained pose that detects AGV by laser scanner are (x _A, y _A, α _A), the azimuth error e of AGV then _αWith normal direction positional error e _PnBe respectively:

e _α＝α _A-α′(1)

$e_{\mathrm{pn}}=k\&CenterDot;\frac{|x_A\tan {\mathrm{\&alpha;}}^{\&prime;}-x^{\&prime;}\tan {\mathrm{\&alpha;}}^{\&prime;}+y^{\&prime;}-y_A|}{\sqrt{1+{\left(\tan {\mathrm{\&alpha;}}^{\&prime;}\right)}^2}}---\left(2\right)$

Set: when the AGV attained pose is positioned at the right side of theoretical path L, k=1; When the AGV attained pose is positioned at the left side of theoretical path L, k=-1;

The normal direction positional error e of b, employing differential algorithm correction AGV _PnWith azimuth error e _α, obtain revised law to positional error e _Pn0With the corrected azimuth error e _{α 0}Be respectively:

${\mathrm{<figure-callout\; id="0"\; label="e\; pn"\; filenames="HDA0000045610990000021.png,HDA0000045610990000031.png"\; state="\{\{state\}\}">e</figure-callout>}}_{\mathrm{pn}}=e_{\mathrm{pn}}+\frac{{\mathrm{de}}_{\mathrm{pn}}\left(t\right)T_d}{\mathrm{dt}}---\left(3\right)$

$=e_{\mathrm{pn}}+\frac{(e_{\mathrm{pn}}\left(i\right)-e_{\mathrm{pn}}(i-1))T_d}{T}$

$e_{\mathrm{\&alpha;}0}=e_{\mathrm{\&alpha;}}+\frac{{\mathrm{de}}_{\mathrm{\&alpha;}}\left(t\right)T_d}{\mathrm{dt}}$ (4)

$=e_{\mathrm{pn}}+\frac{(e_{\mathrm{\&alpha;}}\left(i\right)-e_{\mathrm{\&alpha;}}(i-1))T_d}{T}$

In the formula: T _dBe derivative time constant, T is the sampling period, e _Pn(i) be the normal direction position error signal of AGV current sampling point pose, e _α(i) be the azimuth error signal of AGV current sampling point pose, e _Pn(i-1) be the normal direction position error signal of the last sampling point pose of AGV, e _α(i-1) be the azimuth error signal of the last sampling point pose of AGV, e _PnAnd e _αThe sampling point difference DELTA of adjacent moment _PnAnd Δ _αBe respectively:

Δ _pn＝e _pn(i)-e _pn(i-1)；

Δ _α＝e _α(i)-e _α(i-1)；

C. try to achieve AGV controlling quantity β by FUZZY ALGORITHMS FOR CONTROL.

At first with revised law to positional error e _Pn0With the corrected azimuth error e _{α 0}Carry out obfuscation, become corresponding fuzzy language variable, infer output Δ β according to existing Mamdani fuzzy reasoning method, fuzzy control rule table and weighted mean clarification method then.And output Δ β be multiply by corresponding fuzzy factor of proportionality k _βObtain final controlling quantity β; Fuzzy quantization factor k _P, k _αFuzzy factor of proportionality k _βWith Prediction Parameters k _Dp, k _{D α}Value all determine according to AGV experiment control effect, and under the different AGV speeds of travel, fuzzy factor of proportionality k _β, Prediction Parameters k _Dp, k _{D α}Value is to change automatically.Be embodied as:

1. the obfuscation of input.

At first with the revised law of AGV to positional error e _Pn0With the corrected azimuth error e _{α 0}Multiply by normal direction positional error fuzzy quantization factor k respectively _PWith azimuth error fuzzy quantization factor k _αTransform to universe of fuzzy sets scope separately, obtain the normal direction positional error input e ' of fuzzy controller _PnWith azimuth error input e ' _α:

e′ _pn＝e _pn0k _p

(5)

＝e _pnk _p+Δ _pnk _dp

e′ _α＝e _αk _α

(6)

＝e _αk _α+Δ _αk _dα

K wherein _DpAnd k _{D α}Be Prediction Parameters, and k _Dp=T _dk _p/ T, k _{D α}=T _dk _α/ T.

Then with normal direction positional error input e ' _PnWith azimuth error input e ' _αCarry out obfuscation according to membership function, become corresponding fuzzy language variable (NB, NM, NS, ZE, PS, PM, PB).Membership function all adopts triangle, selects the fuzzy set of low resolution for use in the bigger zone of error, selects high-resolution fuzzy set for use in the less zone of error, and membership function as shown in Figure 5.

2. the formulation of fuzzy control rule table.

Fuzzy control rule is the core of fuzzy controller, and it directly affects the quality of controller performance.Fuzzy control rule generally adopt " IF ... .THEN .... " mode describe, the former piece of conditional clause is input and state, and consequent is a control variable, and present embodiment is according to the operating experience of operating experience and people's steering vehicle, sum up the fuzzy control rule of AGV control system, as IF e _Pn=NB ande _α=NB, 49 rules such as Then β=NB., and it is as shown in table 1 to make the form of rule list.

3. fuzzy reasoning.

According to input and fuzzy control rule, can infer controlling quantity x according to existing Mamdani fuzzy reasoning method _iCooresponding with it membership function μ (x _i).

4. sharpening.

Adopt existing formulae (7) method of weighted mean to be transformed to accurate amount in the domain scope through controlling quantity that fuzzy reasoning obtains and cooresponding membership function thereof.

$\mathrm{\&Delta;\&beta;}=\frac{\underset{0}{\overset{48}{\mathrm{\&Sigma;}}}{x}_i\&CenterDot;\mathrm{\&mu;}\left(x_i\right)}{\underset{i=0}{\overset{48}{\mathrm{\&Sigma;}}}\mathrm{\&mu;}\left(x_i\right)}---\left(7\right)$

Then working control amount β is: β=Δ β k _β(8)

Wherein, k _βFor fuzzy factor of proportionality, be the input control signal of industrial computer with working control amount β;

5. fuzzy quantization factor k _Pn, k _α, fuzzy factor of proportionality k _βWith Prediction Parameters k _Dp, k _{D α}Determine.

By analysis as can be known, if quantizing factor k _p, k _αValue is too small, and the reaction of system is too slow, and stable state accuracy reduces; When but value was excessive, the governing speed of system was too fast, very easily produced concussion, influence control effect.Factor of proportionality k _βSystem can produce bigger overshoot when excessive, and concussion property also can increase; Then can make the system dynamics response time long when too small, be unfavorable for reaching fast steady state effect.Prediction Parameters k _DpAnd k _{D α}Size the effect of improving of controller is parabolic shape, can reach the optimal control effect within the specific limits, so these parameters of choose reasonable system have great influence for the controller performance that improves fuzzy controller.

Fuzzy factor of proportionality k _β, Prediction Parameters k _DpAnd k _{D α}Be to adjust in real time according to the variation of the AGV speed of travel; Normal direction positional error fuzzy quantization factor k _PWith azimuth error fuzzy quantization factor k _αUnder any speed of travel of AGV, be fixed value, determine as follows:

(1) makes k _Dp=0, k _{D α}=0, determine k _P, k _αAnd k _βValue:

Input variable of fuzzy controller e ' _Pn, e ' _αBe { NB, NM, NS, ZE, PS, PM, PB}, e ' with the fuzzy language set of output variable β _PnBasic domain be [0.05,0.05], measure unit is m, its universe of fuzzy sets is [6,6]; E ' _αBasic domain be that [0.1,0.1], measure unit are rad, universe of fuzzy sets is [6,6]; The basic domain of β is that [10,10], measure unit are degree, and universe of fuzzy sets is [1,1].

K then _P, k _αAnd k _βInitial value be respectively:

${\mathrm{<figure-callout\; id="0"\; label="k\; p"\; filenames="HDA0000045610990000021.png,HDA0000045610990000031.png"\; state="\{\{state\}\}">k</figure-callout>}}_p=\frac{6}{0.05}=120;$ $k_{\mathrm{\&alpha;}0}=\frac{6}{0.1}=60;$ $k_{\mathrm{\&beta;}0}=\frac{10}{1}=10;$

With initial value k _P0, k _{α 0}And k _{β 0}Be the basis, determine k according to experiment effect _P, k _α, k _βValue:

AGV is walked along a certain projected route with speed v=10m/min low speed, k is set _P=120, k _α=60 remain unchanged, with k _βValue changes in the domain scope of [0,15], and finding and making the AGV path trace error minimum and the fastest value of regulating the speed is k _β=10; Make k then _β=10 remain unchanged, k _PAnd k _αOne of them keeps initial value constant, and another one changes in its domain scope, finds out to make the AGV path trace error minimum and the fastest class value of regulating the speed be respectively k _p=120; k _α=180; Obtaining AGV, to control best one group fuzzy factor of effect under lower-speed state be k _p=120; k _α=180; k _β=10; K wherein _PThe domain scope be [30,600], k _αThe domain scope be [15,300].

Keep k _p=120 and k _α=180 is constant, changes the speed of travel of AGV, make AGV respectively with 20,30,40,50, the speed walking of 60m/min, k is set _βValue changes in the domain scope of [0,15], according to

Relation, find out and under each speed, make the minimum and the fastest k that regulates the speed of path trace error _βValue, as shown in table 2.

(2) determine Prediction Parameters k _DpAnd k _{D α}:

Make AGV respectively with 10,20,30,40,50, the speed walking of 60m/min, the value of the fuzzy factor under each speed remains unchanged respectively, makes k _Dp=k _pD ₀k _{D α}=k _αD ₀, and set gradually d respectively ₀=1,2,3,4 ..., contrast d ₀The control effect of AGV when getting different values finds one group to make the AGV path trace error minimum and the fastest prediction of regulating the speed count k _Dp=b ₂k _{D α}=c ₂

Walking with the speed of 60m/min with AGV is example, according to the experimental sequence that experimental program table as shown in table 3 is arranged, allows AGV follow the tracks of the straight line path of setting, initial error e from certain initial condition _PnBe set to about 0.08m angular error e _αFor about 0.025rad.According to experiment effect, can obtain its best Prediction Parameters is k _Dp=480, k _Da=720.Fig. 6 has shown error adjustment and the control diagram of curves of laser guiding AGV when walking with the speed of 60m/min, convenient for being figure, it is that the numerical value of measure unit: AGV can reach stabilized conditions in short time that present embodiment is converted into controlling quantity β with rad, the sum of errors controlling quantity all changes in a small range subsequently, but in the after control amount a big swing has appearred again, this is because AGV begins reduction of speed, in order to guarantee the path trace precision, factor of proportionality k _βIncrease along with reducing of speed v, controlling quantity β also increases thereupon.At present when AGV speed v≤60m/min, control accuracy can reach | e _Pn|≤0.01m, | e _α|≤0.01rad.

Prediction Parameters k under each speed _DpAnd k _{D α}Value see Table 2, do not have the pairing k of speed of displaying in the table _β, k _Dp, k _{D α}Value can adopt method of interpolation to try to achieve.

Table 1 fuzzy control rule table

K under the table 2 part moving velocity _β, k _Dp, k _{D α}Value

Table 3 experimental program table

Claims (3)

1. the two-position closed loop control system of an industrial laser guiding AGV is made up of detection module and control module, it is characterized in that:

Described detection module is by constituting as the coder of motor position sensor with as the laser scanner of AGV pose sensor;

Described control module is made of servo-driver and industrial computer, and it is the two-position closed loop control module that is made of interior position ring and external position ring that described control module is set;

Position ring is a motor angle position ring in described, detects motor angle position signal by coder, to servo-driver, is electric-motor drive unit with the servo-driver with described motor angle position signal feedback;

Described external position ring is to read AGV pose signal with laser scanner, and described AGV pose signal feedback is to industrial computer, and industrial computer outputs signal to the servo-driver signal input part, and motor rotates output and drives AGV walking steering hardware by retarder; The path trace error of the AGV pose calculated signals AGV current location of feeding back in the position ring in addition is with the input of described path trace error as control system.

2. the path tracking control method of the described industrial laser guiding AGV control system of a claim 1, it is characterized in that adopting fuzzy prediction control algorithm, in the rectangular coordinate system on AGV motion plane, place, the pose of AGV is expressed as (x, y, α), (x wherein, y) be the position coordinate of AGV in described rectangular coordinate system, α is the azimuth of described AGV, and described azimuth is the central axis of AGV and the angle of x axle forward;

The object pose that described AGV is set is (x ', y ', α '), and detecting the starting point pose that obtains AGV by laser scanner is (x ₀, y ₀, α ₀); Cook up from described starting point pose (x according to the associated pathway planning algorithm ₀, y ₀, α ₀) to the theoretical path L of object pose (x ', y ', α ');

Error between definition attained pose and the object pose is a position and attitude error, and described position and attitude error comprises normal direction positional error e _PnWith azimuth error e _α

Described path following method is finished by following steps:

A, the attained pose that detects AGV by laser scanner are (x _A, y _A, α _A), the azimuth error e of AGV then _αWith normal direction positional error e _PnBe respectively:

e _α＝α _A-α′ (1)

$e_{\mathrm{pn}}=k\&CenterDot;\frac{|x_A\tan {\mathrm{\&alpha;}}^{\&prime;}-x^{\&prime;}\tan {\mathrm{\&alpha;}}^{\&prime;}+y^{\&prime;}-y_A|}{\sqrt{1+{\left(\tan {\mathrm{\&alpha;}}^{\&prime;}\right)}^2}}---\left(2\right)$

Set: when the AGV attained pose is positioned at the right side of theoretical path L, k=1; When the AGV attained pose is positioned at the left side of theoretical path L, k=-1;

The normal direction positional error e of b, employing differential algorithm correction AGV _PnWith azimuth error e _α, obtain revised law to positional error e _Pn0With the corrected azimuth error e _{α 0}Be respectively:

$e_{\mathrm{pn}0}=e_{\mathrm{pn}}+\frac{{\mathrm{de}}_{\mathrm{pn}}\left(t\right)T_d}{\mathrm{dt}}---\left(3\right)$

$=e_{\mathrm{pn}}+\frac{(e_{\mathrm{pn}}\left(i\right)-e_{\mathrm{pn}}(i-1))T_d}{T}$

$e_{\mathrm{\&alpha;}0}=e_{\mathrm{\&alpha;}}+\frac{{\mathrm{de}}_{\mathrm{\&alpha;}}\left(t\right)T_d}{\mathrm{dt}}$ (4)

$=e_{\mathrm{pn}}+\frac{(e_{\mathrm{\&alpha;}}\left(i\right)-e_{\mathrm{\&alpha;}}(i-1))T_d}{T}$

In the formula: T _dBe derivative time constant, T is the sampling period, e _Pn(i) be the normal direction position error signal of AGV current sampling point pose, e _α(i) be the azimuth error signal of AGV current sampling point pose, e _Pn(i-1) be the normal direction position error signal of the last sampling point pose of AGV, e _α(i-1) be the azimuth error signal of the last sampling point pose of AGV, e _PnAnd e _αThe sampling point difference DELTA of adjacent moment _PnAnd Δ _αBe respectively:

Δ _pn＝e _pn(i)-e _pn(i-1)；Δ _α＝e _α(i)-e _α(i-1)；

C, try to achieve AGV working control amount β by FUZZY ALGORITHMS FOR CONTROL

At first with the revised law of AGV to positional error e _Pn0With the corrected azimuth error e _{α 0}Multiply by normal direction positional error fuzzy quantization factor k respectively _PWith azimuth error fuzzy quantization factor k _αTransform to universe of fuzzy sets scope separately, obtain the normal direction positional error input e ' of fuzzy controller _PnWith azimuth error input e ' _α:

e′ _pn＝e _pn0k _p

(5)

＝e _pnk _p+Δ _pnk _dp

e′ _α＝e _αk _α

(6)

＝e _αk _α+Δ _αk _dα

Wherein, k _DpAnd k _{D α}Be Prediction Parameters, and k _Dp=T _dk _p/ T, k _{D α}=T _dk _α/ T;

With normal direction positional error input e ' _PnWith azimuth error input e ' _αCarry out obfuscation according to membership function; Infer output Δ β according to Mamdani fuzzy reasoning method, fuzzy control rule and average weighted clarification method; Then working control amount β is: β=Δ β k _β(7)

Wherein, k _βFor fuzzy factor of proportionality, be the input control signal of industrial computer with working control amount β.

3. according to the described path tracking control method of claim 2, it is characterized in that described fuzzy factor of proportionality k _β, Prediction Parameters k _DpAnd k _{D α}Be to adjust in real time according to the variation of the AGV speed of travel; Normal direction positional error fuzzy quantization factor k _PWith azimuth error fuzzy quantization factor k _αUnder any speed of travel of AGV, be fixed value, determine as follows:

A, make k _Dp=0, k _{D α}=0, determine k _P, k _αAnd k _βValue:

The normal direction positional error input e ' of fuzzy controller _PnBasic domain be [0.05,0.05], measure unit is m, its universe of fuzzy sets is [6,6]; The azimuth error input e ' of fuzzy controller _αBasic domain be that [0.1,0.1], measure unit are rad, universe of fuzzy sets is [6,6]; The basic domain of working control amount β is that [10,10], measure unit are degree, and universe of fuzzy sets is [1,1];

K then _P, k _αAnd k _βInitial value be respectively:

$k_{p0}=\frac{6}{0.05}=120;$ $k_{\mathrm{\&alpha;}0}=\frac{6}{0.1}=60;$ $k_{\mathrm{\&beta;}0}=\frac{10}{1}=10;$

With described initial value k _P0, k _{α 0}And k _{β 0}Be the basis, determine k according to experiment effect _P, k _α, k _βValue:

At first make AGV with speed v=10m/min low speed walking, k is set _P=k _P0, k _α=k _{α 0}Remain unchanged, with k _βValue is at [k _{β 0}-10, k _{β 0}+ 5] change in the domain scope, finding and making the AGV path trace error minimum and the fastest value of regulating the speed is a ₀Make k then _β=a ₀Remain unchanged k _PAnd k _αOne of them keeps initial value constant, and another one changes in its domain scope, finds out to make the AGV path trace error minimum and the fastest class value of regulating the speed be respectively k _p=b ₀k _α=c ₀Obtaining AGV, to control best one group fuzzy factor of effect under lower-speed state be k _p=b ₀k _α=c ₀k _β=a ₀Wherein said k _PThe domain scope be

k _αThe domain scope be

Keep k _p=b ₀And k _α=c ₀Constant, change the speed of travel of AGV, make AGV respectively with 20,30,40,50, the speed walking of 60m/min, k is set _βValue is at [a ₀-10, a ₀+ 5] change in the domain scope, according to

Relation, find out and under each speed, make the minimum and the fastest k that regulates the speed of path trace error _βValue;

B, determine Prediction Parameters k _DpAnd k _{D α}:

Make AGV respectively with 10,20,30,40,50, the speed walking of 60m/min, the value of the fuzzy factor under each speed remains unchanged respectively, makes k _Dp=k _pD ₀k _{D α}=k _αD ₀, and set gradually d respectively ₀=1,2,3,4 ..., contrast d ₀The control effect of AGV when getting different values finds one group to make the AGV path trace error minimum and the fastest prediction of regulating the speed count k _Dp=b ₂k _{D α}=c ₂

Try to achieve k under other speed of travel of AGV by method of interpolation _β, k _Dp, k _{D α}Value.

Images