CN103309351A - Maintenance robot obstacle avoidance planning method - Google Patents

Abstract

The invention belongs to the field of sensing and detection and particularly relates to a maintenance robot obstacle avoidance planning method which is used for creating a polar bar chart based on laser ranging data so as to solve the problem that a potential field method is got into a local minimum easily and chatters frequently and to overcome the defect that the real-time speed of a robot is not considered in an enhanced vector field method in threshold function construction. The method includes creating a polar bar chart, simplifying the polar bar chart through a threshold function, and selecting a maintenance robot advancing direction according to an evaluation function. The method has the advantages of small amount of computation and rapid response; an advancing direction selection range of a next operating period of the robot is narrowed; corresponding computation time and energy consumption is reduced.

Description

A kind of robot for overhauling collision-avoidance planning method

Technical field

The invention belongs to the sensor measuring field, be particularly related to and a kind ofly make up the problem that the polar coordinates histogram easily is absorbed in local minimum and frequently trembles to solve potential field method based on laser ranging data, overcome enhanced vector field method is not considered the shortcoming of robot real-time speed in the threshold function structure robot for overhauling collision-avoidance planning method.

Background technology

In the face of increasingly serious Energy situation, world government all begins the develop actively nuclear power.As everyone knows, the nuclear power difference is that the nuclear reaction meeting produces ionising radiation and radioactive waste with the maximum characteristics of thermoelectricity.Cause radiomaterial entered environment, prophylactic repair just to seem most important for avoiding having an accident.Yet the reactor core that is related to nuclear safety most is isolated because ionising radiation intensity is too high, and personnel can't enter, so nuclear industrial robot arises at the historic moment.Nuclear industrial robot is early than bringing into use the 1950's, after develop into gradually remote-controlled robot.Use nuclear industrial robot can reduce operator's suffered nuclear radiation dosage in various examination and maintenance are processed, and can improve monitoring capacity, avoid replacing generating set, cut operating costs.

The robot that is applicable to nuclear industry mainly contains following feature: possess stronger mobility, can pass through and the cut-through thing; Equip various video cameras and sensor, good reconnaissance capability is arranged; Can arrive fast the operation place, possess good maneuverability; Have reversible controller, can be fast with feedback of status to the operator; Have good force feedback, can protect operand.Along with the development of Robotics, autonomous robot will replace manually-operated gradually.

How making robot for overhauling independently finish without the problem of bumping operation and arriving safely the appointed area by carry sensors is typical collision-avoidance planning problem.Consider to be full of high temperature and high pressure steam visibility extreme difference in the nuclear power plant accident situation in the containment, the use that can't play a role of optical sensor and ultrasonic sensor, thereby select the stronger laser range finder of antijamming capability.The present invention is based on laser ranging technique, the dynamic obstacle avoidance algorithm is studied, in the hope of realizing the safe and reliable dynamic collision prevention of robot for overhauling of real-time high-efficiency.

At present, existing several different methods is used for realizing the Robotic Dynamic collision prevention both at home and abroad, and wherein Artificial Potential Field Method and vector field method be subject to paying close attention to more and more widely in recent years.(can't pass through the barrier group of tight distribution such as Artificial Potential Field Method except the defective of algorithm self, vector field method calculated amount is large and the complicated algorithm real-time that causes of computation process is poor), they do not consider the real-time speed of robot yet, or the robot that its environmental modeling mode is only applicable to be equipped with coarse sensor (for example sonac) carries out collision-avoidance planning.Therefore, these planing methods are difficult to satisfy the practical implementation demand of robot for overhauling.

The problem that exists in order to solve above collision prevention method commonly used, and further satisfy the at the scene barrier demand of keeping away in the operation of robot for overhauling, providing polar coordinates histogram method, the method has obviously reduced calculating, is more suitable for keeping away barrier for laser range finder.Simultaneously, by in threshold function structure, considering robot speed and acceleration, thereby guaranteed that robot for overhauling can be with speed safety moving faster.

Document " VPH+:An Enhanced Vector Polar Histogram Method for Mobile Robot Obstacle Avoidance " has been mentioned close technology.But the document thinks that obstacle is made of a series of barrier point usually, and the very near barrier point of mutual distance is considered as an obstacle collection.According to the geometric relationship between each obstacle collection, the obstacle collection sorted out be divided into recessed collection or convex set, recessed collection mean in this set barrier point from robot close to, so direction corresponding to recessed collection position optional direction of not advancing as robot; Barrier point in the opposite convex set from robot relatively away from, the optional direction that direction corresponding to its position advanced as robot.Although but the method has improved the efficient of screening robot direct of travel when processing, but still may there be the barrier point that can not bump in the robot traveling process in the true fovea superior collection, and according to document method, the scope that direction corresponding to this part barrier point position also put under robot can not advance, therefore definition and the non-tight of recessed collection; And the present invention is when processing, the distance of robot barrier point on each direction is compared successively with the safe distance of definition (guarantee robot to determine that the speed Shi Buhui that advances bumps with obstacle) respectively, but so just can not omit the candidate solution of robot direct of travel.Its two, the document during cost function, is chosen the angle of obstacle direction and target direction as the restriction boot entry of robot trend target in definition, sees that from analysis the restriction of choosing this angle is also not obvious; And the present invention directly chooses the angle of the current direct of travel of robot and target direction as the restriction boot entry of robot trend target when the definition cost function, and the meaning of this angle variable is clearer and more definite, acts on more outstanding.Its three, the document carry out calculating and the comparison of cost function value in the angular range of advancing, thereby calculated amount is larger when choosing best direct of travel; And the present invention at first chooses candidate's direct of travel by the intuitive analysis of two-dimensional histogram, then carries out calculating and the comparison of cost function value in candidate's direct of travel, so calculated amount is obviously less.

Summary of the invention

What the purpose of this invention is to provide that a kind of calculated amount is less, reaction velocity is faster, practicality is higher makes up the robot for overhauling collision-avoidance planning method of polar coordinates histogram based on laser ranging data.

The object of the present invention is achieved like this:

(1) set up the polar coordinates histogram:

The distance of laser range finder detecting obstacles thing and robot for overhauling, and the distance distribution information that records passed to microcomputer, with the histogram graph representation under the polar coordinates, the x axle represents the sequence number of laser beam, the y axle represents the obstacle distance on this detection direction;

(2) simplify the polar coordinates histogram by threshold function:

Set up the robot motion equation,

$t=\frac{V\left(t\right)}{a},$

$S\left(t\right)=V\left(t\right)t-\frac{1}{2}{\mathrm{at}}^2=\frac{V^2\left(t\right)}{2a},$

Wherein: t is the time, and V (t) is t real-time speed constantly, and a is acceleration, the distance that S (t) passed by for the t time.

For detection direction i,

$V(i,t)=\frac{(\stackrel{\→}{V_t}\·\stackrel{\→}{d_i})}{d_i}=\left\{\begin{array}{cc}{V}_t\·\cos (\mathrm{\θ}\left(t\right)-i);& |\mathrm{\θ}\left(t\right)-i|\≤90\\ 0;& |\mathrm{\θ}\left(t\right)-i|\≥90\end{array}\right.,$

$S(i,t)=\frac{V^2(i,t)}{2a}=\left\{\begin{array}{cc}\frac{{V_t}^2\·{\cos }^2(\mathrm{\θ}\left(t\right)-i)}{2a};& |\mathrm{\θ}\left(t\right)-i|\≤90\\ 0;& |\mathrm{\θ}\left(t\right)-i|\≥90\end{array}\right.,$

Wherein, V (i, t) is engraved in the speed on the direction i, V when being robot for overhauling t _tBe the real-time speed of robot for overhauling, d _iBe the distance between barrier on robot for overhauling and the i direction, the working direction that θ (t) changes in time for robot for overhauling, the distance that S (i, t) passed by at direction i for the robot for overhauling t time, a is the acceleration of motion of robot for overhauling.

Set up threshold function:

$W(i,t)=S(i,t)+R+D=\left\{\begin{array}{cc}\frac{{V_t}^2\·{\cos }^2(\mathrm{\θ}\left(t\right)-i)}{2a}+R+D;& |\mathrm{\θ}\left(t\right)-i|\≤90\\ R+D;& |\mathrm{\θ}\left(t\right)-i|\≥90\end{array}\right.,$

Wherein: V _tReal-time speed for robot for overhauling, the working direction that θ (t) changes in time for robot for overhauling, a is the acceleration of motion of robot for overhauling, and R is the equivalent redius of robot for overhauling self size, and D is the safe distance between from the robot for overhauling to the barrier;

The polar coordinates histogram is converted into the binary histogram:

$H\left(i\right)=\left\{\begin{array}{cc}1& d\left(i\right)\&GreaterEqual;W(i,t)\\ 0& d\left(i\right)<W(i,t)\end{array}\right.$

Wherein: 1 expression robot P Passable, 0 expression robot cannot pass through;

(3) select the robot for overhauling working direction according to evaluation function:

But set up the FOH set,

U＝{i|H(i)＝1}，

Definition evaluation function C,

C(i)＝K ₁|i-Θ|+K ₂|i-θ(t)| i∈U，

Wherein, K ₁, K ₂Be constant, Θ is the target direction of robot for overhauling, the working direction that θ (t) changes in time for robot for overhauling,

Choose best direct of travel:

C(Ω)＝minC(i),i∈U

Wherein: Ω is optimum working direction.MinC (i) but be the minimum value of the evaluation function of all directions in the FOH.

The polar coordinates histogram is converted into the binary histogram, be relatively being converted into of the range distribution in the histogram under the polar coordinates and threshold function table formula, data can the visual representation robot binary histogram of P Passable whether, the horizontal ordinate of binary histogram is 0～180 ° scan angle, the value of ordinate is illustrated on each scanning angle in 0～180 ° of scope, and whether robot P Passable.

Beneficial effect of the present invention is:

The present invention has not only solved the intrinsic defective of traditional collision prevention method, and need not to set up just direct processes sensor raw data of complicated physical model, more is conducive to the online collision-avoidance planning of robot realization, has advantages of that calculated amount is little, reaction velocity is fast; The method has taken into full account the robot real-time speed in the threshold function structure simultaneously, more meets the practical implementation demand of robot for overhauling.Use more can the visual representation robot binary histogram of P Passable whether, dwindled next work period of robot and selected the scope of direct of travel, reduced corresponding computing time and energy consumption.

Description of drawings

Fig. 1 is stadimeter detection information diagram;

Fig. 2 is robot and barrier relative position schematic diagram;

Fig. 3 is the histogram rough schematic view;

Fig. 4 is test bench system;

Fig. 5 is laser range sensor;

Fig. 6 is LMS200 equipment pie graph;

Fig. 7 is the collision prevention algorithm flow chart;

Fig. 8 is small size multi-obstacle avoidance collision-avoidance planning experiment schematic diagram;

Fig. 9 is for falling into the trap schematic diagram;

Figure 10 is for breaking away from the trap schematic diagram.

Embodiment

Below in conjunction with accompanying drawing the present invention is described further.

The input value of polar coordinates histogram algorithm (hereinafter to be referred as pole figure method) is that the obstacle distance of sensor real-time detection distributes and comes calculating robot's direct of travel by three data treatment steps:

To change for the data of the array of depositing obstacle distance the histogram under the polar coordinates into,

Set up threshold function, and by this threshold function the polar coordinates histogram be converted into the binary histogram,

Calculate by evaluation function, choose best angle of travel from the binary histogram,

1. set up the polar coordinates histogram

Robot relies on the obstacle on the laser range finder detection working direction in traveling process.In each detect cycle, stadimeter sends beam of laser and receives the light beam that is reflected by barrier, provides distance value according to its range finding orientation, along with the gradually change of laser emission angle, finishes a work period through several detect cycle sensors.At each end of term work week sensor distance distribution information is passed to microcomputer.Then with distance distribution information with the histogram graph representation under the polar coordinates out, as shown in Figure 1.Wherein the x axle represent laser beam sequence number (also characterize simultaneously its emission angle, usually setting sensor take 1 ° as interval unit, each work period detection 181 times); The y axle represents the obstacle distance on this detection direction.

2. simplify based on the polar coordinates histogram of threshold function

Known real-time speed and acceleration fully take into account the practical implementation demand of robot for overhauling, set up and consider that robot has the equation of motion of acceleration, in the situation that just can obtain time and distance that the robot braking must be reserved.Equation is as follows:

$t=\frac{V\left(t\right)}{a},$

$S\left(t\right)=V\left(t\right)t-\frac{1}{2}{\mathrm{at}}^2=\frac{V^2\left(t\right)}{2a},$

Wherein: t is the time,

V (t) is t real-time speed constantly,

A is acceleration,

The distance that S (t) passed by for the t time.

As shown in Figure 2, please note i among Fig. 2 for either direction i( ₁, i ₂The difference of approaching degree on the direction), just above system of equations can be rewritten as:

$V(i,t)=\frac{(\stackrel{\&RightArrow;}{V_t}\&CenterDot;\stackrel{\&RightArrow;}{d_i})}{d_i}=\left\{\begin{array}{cc}{V}_t\&CenterDot;\cos (\mathrm{\&theta;}\left(t\right)-i);& |\mathrm{\&theta;}\left(t\right)-i|\&le;90\\ 0;& |\mathrm{\&theta;}\left(t\right)-i|\&GreaterEqual;90\end{array}\right.,$

$S(i,t)=\frac{V^2(i,t)}{2a}=\left\{\begin{array}{cc}\frac{{V_t}^2\&CenterDot;{\cos }^2(\mathrm{\&theta;}\left(t\right)-i)}{2a};& |\mathrm{\&theta;}\left(t\right)-i|\&le;90\\ 0;& |\mathrm{\&theta;}\left(t\right)-i|\&GreaterEqual;90\end{array}\right.,$

Wherein: V (i, t) is engraved in the speed on the direction i when being robot for overhauling t,

V _tBe the real-time speed of robot for overhauling,

d _iBe the distance between barrier on robot for overhauling and the i direction,

The working direction that θ (t) changes in time for robot for overhauling,

The distance that S (i, t) passed by at direction i for the robot for overhauling t time,

A is the acceleration of motion of robot for overhauling.

Consider simultaneously the size (describing it as the circle that radius is R) of robot for overhauling self, and safe distance (the minor increment D that nothing is bumped between robot and the barrier), obtain thus final time dependent threshold function, as shown in the formula:

$W(i,t)=S(i,t)+R+D=\left\{\begin{array}{cc}\frac{{V_t}^2\&CenterDot;{\cos }^2(\mathrm{\&theta;}\left(t\right)-i)}{2a}+R+D;& |\mathrm{\&theta;}\left(t\right)-i|\&le;90\\ R+D;& |\mathrm{\&theta;}\left(t\right)-i|\&GreaterEqual;90\end{array}\right.$

Wherein: V _tBe the real-time speed of robot for overhauling,

The working direction that θ (t) changes in time for robot for overhauling,

A is the acceleration of motion of robot for overhauling,

R is the equivalent redius of robot for overhauling self size,

D is the safe distance between from the robot for overhauling to the barrier.

In the expression formula of above threshold function, V _tAnd θ (t) can Real-time Obtaining, and a and R are steady state value (for the simplification problem, supposing that it is steady state value), and D is unique undetermined coefficient, can be according to the value of specific environment appropriate change D.

In addition, because the test platform that uses among the design is driven by stepper motor, general threshold function need to be made following modification for its action characteristic.Robot for overhauling has in afore-mentioned provides acceleration to be-clamping device of a, braking procedure is followed Newton's laws of motion, and in test platform, stepper motor just shuts down immediately when microcomputer stops to send driving pulse, before this in the one-period, robot is at θ (t) the direction regular length S that passes by when mobile, with in this length substitution threshold function expression formula must:

$W(i,t)=S\left(i\right)+R+D=\left\{\begin{array}{cc}S\&CenterDot;\cos +(\mathrm{\&theta;}\left(t\right)-i)R+D;& |\mathrm{\&theta;}\left(t\right)-i|\&le;90\\ R+D;& |\mathrm{\&theta;}\left(t\right)-i|\&GreaterEqual;90\end{array}\right.$

By comparing range distribution and the threshold function in the histogram under the polar coordinates, can obtain a binary histogram, as shown in Figure 3.

Wherein, according to following rule creation binary histogram:

$H\left(i\right)=\left\{\begin{array}{cc}1& d\left(i\right)\&GreaterEqual;W(i,t)\\ 0& d\left(i\right)<W(i,t)\end{array}\right.$

Wherein: 1 expression robot P Passable, 0 expression robot cannot pass through.

According to the definition of above threshold function, just the histogram under the polar coordinates can be converted into the binary histogram, as shown in Figure 3, in the binary histogram according to this rule creation, valley represents the impassability district, but peak value represents FOH.

The polar coordinates histogram simplify be relatively being converted into of the formula such as the range distribution in the histogram under the polar coordinates and threshold function table, data can visual representation the robot binary histogram of P Passable whether.The horizontal ordinate of binary histogram represents 0～180 ° scan angle, and the value of ordinate is illustrated on each scanning angle in 0～180 ° of scope, and whether robot P Passable.When the ordinate value is 1, represent robot P Passable on this scanning angle; Otherwise when the ordinate value is 0, then represents robot and on this scanning angle, can not pass through.The map generalization of binary column, but essence is the scope that direct of travel is provided for the optimum direct of travel of Robot Selection, has namely dwindled the scope of next work period selection direct of travel of robot, has reduced corresponding computing time and energy consumption.

3. select the robot for overhauling working direction based on evaluation function

As shown in Figure 2, peak value explanation robot can have free passage in this direction, but at this moment problem has been reduced to chooses an optimum orientation in known FOH.

But the direction of all robot P Passables is classified as FOH set U:

U＝{i|H(i)＝1}

Next define a suitable evaluation function C, utilize this function to bring whole i values among the U into, can calculate optimum working direction Ω, its expression formula is as follows:

C(i)＝K ₁|i-Θ|+K ₂|i-θ(t)| i∈U

Wherein: K ₁, K ₂Be constant,

Θ is the target direction of robot for overhauling,

The working direction that θ (t) changes in time for robot for overhauling.

The K of first of this expression formula ₁| i-Θ | the degree of pressing close to of each calculated direction and target direction has been estimated in representative, is used for guided robot trend target; Second portion K ₂| i-θ (t) | then estimate the degree of pressing close to of each calculated direction and current robot working direction, avoided the robot corner excessive.

In pole figure method, suitably choose K ₁, K ₂These two undetermined coefficients seem particularly important.In order to pay the utmost attention to guided robot trend target, K ₁Must be greater than K ₂Concrete choosing value will be fit to specific working environment.

According to the definition of above evaluation function, best direct of travel must be tried one's best simultaneously near the working direction of target direction and last work period, namely brings this direction i into the evaluation function expression formula and obtains as far as possible little value.Therefore the direct of travel of finally choosing is that the expression formula of Ω is:

C(Ω)＝minC(i),i∈U

Wherein: Ω is optimum working direction,

MinC (i) but be the minimum value of the evaluation function of all directions in the FOH.

4. speed control

If the speed of robot for overhauling is adjustable, and its maximal rate is V _MaxThe traveling mode of robot should satisfy: keep maximal rate when freely advancing, and use a less speed V when detecting barrier and implementing collision prevention.This speed control expression formula is:

$V=(1-\frac{D_{\max }-d\left(\mathrm{\&Omega;}\right)}{D_{\max }-D_{\min }})V_{\max };$ D _min＝R+D

Wherein:

Ω is by formula C (Ω)=minC (i), and i ∈ U calculates,

D _MaxBe the maximum detectable range of laser range finder,

D _MinMinimum safe distance for robot.

(1) pilot system hardware forms

1) test bench system

The test-bed of this use is comprised of stand main body frame, main control computer, HIT6502 four shaft step motor control cards and stepper motor, as shown in Figure 4.

Control subject is industrial control computer, is realized and the functions such as order reception, transmission by in fact current task setting, motion algorithm.The stepper motor motion control is by the realization of Hit6502 type four shaft step motor motion control cards, and motion control card is connected with main control computer by the ISA interface.This test-bed comprises 4 motion control cards, can realize altogether the independent control (five degree of freedom of platform six-freedom degree and spatial obstacle) to 11 degree of freedom.Mobile platform carries that laser range finder need be realized in the XOY plane (surface level) two translational degree of freedom and around the rotational freedom of Z axis (vertical to) rotation among the design.The extreme position of each degree of freedom of system all is provided with limit sensors (totally 11), sensor also connects with motion control card, move to the limit pose place of certain degree of freedom when platform, sensor sends alerting signal to motion control card, and motion control card sends brake signal to motor driver immediately.

2) laser range sensor

The design adopts is the LMS200-30106 type laser range sensor system that German SICK company produces, as shown in Figure 5.This type non-contact laser measuring system can as laser radar, be surveyed its surrounding environment with two-dimensional approach.Because it belongs to the active scan system, does not need outside passive device such as catoptron or position mark.SICK has higher resolution, being widely used in two-dimensional distance detects, particularly importantly it can the perception reflex rate only be the smooth surfaces such as 10% obstacle such as glass, perhaps works in the smog environment, thereby is selected as the main sensors of nuclear power station robot for overhauling.

According to its communication programming handbook, LMS200-30106 type laser range finder Application standard RS232 serial ports and microcomputer communication, and adhere rigidly to standard serial port communication rule.

SICK LMS200-30106 type laser range sensor system technical indicator:

Sweep limit: the 80m(maximal value) 10m(10% reflectivity)

Scanning angle: maximum 180 ° (adjustable)

Angular resolution: 0.25 °/0.5 °/1 ° (adjustable)

Response time: 53ms/26ms/13ms

Resolution/systematic error: 10mm/typ ± 15mm

Data-interface: RS232/RS422

Switching value output: typ.24VDC

Lasing safety grade: the 1(eye-safe)

Operating ambient temperature: 0-± 50 ℃

Encapsulation grade: IP65

Size (W * H * D): 155 * 210 * 156mm3

LMS200 is comprised of device housings, generating laser and relay indicating light.Fuselage two-way terminals connect respectively RS-232 serial line interface and the 24V stabilized voltage supply of microcomputer.Equipment connection such as Fig. 6:

The LMS200 fuselage interior has the servomotor of driving laser emitter member rotation, thereby can be in the situation that the fixing obstacle detection that realizes in 0 to the 180 degree scope of fuselage body.At each detect cycle end, LMS200 crosses the RS-232 serial port with the obstacle distance information exchange of scanning and sends to main control microcomputer, environmental information can be showed with barrier profile form at visualization interface after unpacking through program.

(2) programmed environment brief introduction

The programmed environment that the design adopts is the Visual C++6.0 that is released by Microsoft company, and it has been one of topmost application development system under the Windows environment since being born always.Visual C++ is not only the Integrated Development Environment of C Plus Plus, and closely links to each other with Win32, utilizes it can finish from bottom software until the exploitation of the various application programs such as the direct user oriented software in upper strata.

MFC (Microsoft Foundation Classes) is the C++ class collection of writing window application, has wherein encapsulated most of commonly used Windows api function and Windows control.The visual application development instrument of height that uses MFC class libraries and Visual C++ to provide can make program development become simpler, has greatly shortened the construction cycle.The design's the main MFC of dependence of programming part storehouse class realizes.

(3) based on the collision prevention algorithm performing step of polar coordinates histogram method

The flow chart of steps of this algorithm is as shown in Figure 7:

Posture information be to read when (1) each work period begins, the current location of robot for overhauling, the position of impact point comprised, and the description of robot physical dimension.

(2) judge then whether robot arrives the target location, if so, then finish the collision prevention navigation, otherwise need to judge whether collision prevention.

(3) judge whether that needing the condition of collision prevention is the current direct of travel obstacle whether influential robot advances.Analyze fDistance[181] array (being used for depositing obstacle distance), in the safe distance obstacle appears if robot is learnt forwardly by the laser range finder detection, then enter collision prevention branch, otherwise behind the line deflection that calculates current location and target location, the order robot advances in the direction straight.Store subsequently posture information, finish this work period.

(4) after program enters collision prevention branch, according to the basic ideas of polar coordinates histogram algorithm, set up threshold function by the robot physical dimension.

(5) with threshold function and obstacle distance array fDistance[181] relatively, but determine FOH.

(6) but set up evaluation function and process FOH information, select best working direction angle.

(7) final order Robot the party stores posture information and finishes this work period to advancing straight.

(4) the dynamic collision prevention analog simulation test of robot for overhauling

1. small size multi-obstacle avoidance collision-avoidance planning experiment

The robot for overhauling working environment of the design supposition is the nuclear power station industry spot, wherein may include a large amount of dense distribution, only is several times as much as the barrier group of robot size.The purpose of this experiment be checking in the situation that the small size multi-obstacle avoidance, the actual collision prevention effect of polar coordinates vector plot method.

In this experiment, barrier is that the radius of dense distribution as shown in Figure 8 is the circular obstacle of 30 pixels, and robot platform is set to default mode (radius is the circle of 5 pixels).

Robot is smoothly by the type barrier group, and the path is comparatively level and smooth.

Analyze collision-avoidance planning result as shown in Figure 8: put in the drawings 1 position robot and detect first in its safe distance barrier is arranged, and hindering this phase takes over to the target link direction, but therefore selecting optimal path by evaluation function in FOH is a certain angle of left avertence, and this action meets the basic thought of polar coordinates histogram method.Forward angle fluctuation is more frequent near point 2 positions, and tracing it to its cause is because distribution of obstacles comparatively dense herein, and a plurality of barriers mutually superpose on the impact of robot and cause.Below with central coordinate of circle represent the circle.When justifying between (253,124) and the circle (185,210), because the impact of circle (320,203), Robot Selection is deflection to the right; Yet be subject at once the combined influence of circle (205,297), circle (320,203) and circle (298,317), robot determines again to turn to the left side at several all after dates; Advanced again several cycles, under the impact of justifying (320,203), again turned right; Here be noted that because circle (298,317) and circle (358,334) are very pressed close to, and they are defined as continuous impassability district in the collision prevention calculating process, thereby under the effect of this " continuously obstacle ", point 2 is more smooth to the path between the point 3; After near the last point of arrival 3, owing to breaking away from whole obstacles, robot moves to impact point straight.

Above process has proved that polar coordinates histogram method has preferably collision prevention effect for undersized barrier group.But at the barrier comparatively dense, in the coefficient situation of a plurality of obstacles, obvious fluctuation can appear in the path.

2. trap environment collision-avoidance planning experiment

May there be various trap area in the working environment of robot for overhauling.Here the trap of saying refers to because algorithm errors can't break away from after causing robot to enter this zone, as shown in Figure 9.

This is because robot does not possess " memory " ability in some algorithm, after finding that there is obstacle in optimal direction, stepping back and walk back a segment distance, still goes back to wrong path.Therefore in the design's algorithm for robot has added " memory " function, namely preserve robot pose data and the environmental information in total movement cycle.

This experiment arranges continuous U-shaped trap obstacle as shown in Figure 10, and purpose is whether checking adding " memory " function collision prevention algorithm has dodging ability to the trap obstacle.Generate as shown in the figure robot path by simulated program, robot breaks away from from trap smoothly, has proved the feasibility of algorithm.

Claims (2)

1. a robot for overhauling collision-avoidance planning method is characterized in that, comprises the steps:

(1) set up the polar coordinates histogram:

The distance of laser range finder detecting obstacles thing and robot for overhauling, and the distance distribution information that records passed to microcomputer, with the histogram graph representation under the polar coordinates, the x axle represents the sequence number of laser beam, the y axle represents the obstacle distance on this detection direction;

(2) simplify the polar coordinates histogram by threshold function:

Set up the robot motion equation,

$t=\frac{V\left(t\right)}{a},$

$S\left(t\right)=V\left(t\right)t-\frac{1}{2}{\mathrm{at}}^2=\frac{V^2\left(t\right)}{2a},$

Wherein: t is the time, and V (t) is t real-time speed constantly, and a is acceleration, the distance that S (t) passed by for the t time;

For detection direction i,

$V(i,t)=\frac{(\stackrel{\&RightArrow;}{V_t}\&CenterDot;\stackrel{\&RightArrow;}{d_i})}{d_i}=\left\{\begin{array}{c}{V}_t\&CenterDot;\cos (\mathrm{\&theta;}\left(t\right)-i);|\mathrm{\&theta;}\left(t\right)-i|\&le;90\\ 0;|\mathrm{\&theta;}\left(t\right)-i|\&GreaterEqual;90\end{array}\right.,$

$S(i,t)=\frac{V^2(i,t)}{2a}=\left\{\begin{array}{c}\frac{{V_t}^2\&CenterDot;{\cos }^2(\mathrm{\&theta;}\left(t\right)-i)}{2a};|\mathrm{\&theta;}\left(t\right)-i|\&le;90\\ 0;|\mathrm{\&theta;}\left(t\right)-i|\&GreaterEqual;90\end{array}\right.,$

Wherein, V (i, t) is engraved in the speed on the direction i, V when being robot for overhauling t _tBe the real-time speed of robot for overhauling, d _iBe the distance between barrier on robot for overhauling and the i direction, the working direction that θ (t) changes in time for robot for overhauling, the distance that S (i, t) passed by at direction i for the robot for overhauling t time, a is the acceleration of motion of robot for overhauling;

Set up threshold function:

$W(i,t)=S(i,t)+R+D=\left\{\begin{array}{c}\frac{{V_t}^2\&CenterDot;{\cos }^2(\mathrm{\&theta;}\left(t\right)-i)}{2a}+R+D;|\mathrm{\&theta;}\left(t\right)-i|\&le;90\\ R+D;|\mathrm{\&theta;}\left(t\right)-i|\&GreaterEqual;90\end{array}\right.,$

Wherein: V _tReal-time speed for robot for overhauling, the working direction that θ (t) changes in time for robot for overhauling, a is the acceleration of motion of robot for overhauling, and R is the equivalent redius of robot for overhauling self size, and D is the safe distance between from the robot for overhauling to the barrier;

The polar coordinates histogram is converted into the binary histogram:

$H\left(i\right)=\begin{array}{}\end{array}\left\{\begin{array}{cc}1& d\left(i\right)\&GreaterEqual;W(i,t)\\ 0& d\left(i\right)<W(i,t)\end{array}\right.$

Wherein: 1 expression robot P Passable, 0 expression robot cannot pass through;

(3) select the robot for overhauling working direction according to evaluation function:

But set up the FOH set,

U＝{i|H(i)＝1}，

Definition evaluation function C,

C(i)＝K ₁|i-Θ|+K ₂|i-θ(t)|i∈U，

Wherein, K ₁, K ₂Be constant, Θ is the target direction of robot for overhauling, the working direction that θ (t) changes in time for robot for overhauling,

Choose best direct of travel:

C(Ω)＝minC(i),i∈U

Wherein: Ω is optimum working direction, minC (i) but be the minimum value of the evaluation function of all directions in the FOH.

2. a kind of robot for overhauling collision-avoidance planning method according to claim 1, it is characterized in that: described the polar coordinates histogram is converted into the binary histogram, be relatively being converted into of the range distribution in the histogram under the polar coordinates and threshold function table formula, data can the visual representation robot binary histogram of P Passable whether, the horizontal ordinate of binary histogram is 0～180 ° scan angle, the value of ordinate is illustrated on each scanning angle in 0～180 ° of scope, and whether robot P Passable.

Images