JPH04358209A - Obstacle avoiding device - Google Patents

Abstract

PURPOSE:To provide the obstacle avoiding device which can be applied even to one or plural moving objects frequently changing the directions at a sharp angle and can realize an exact collision avoiding operation by utilizing fuzzy theory without using any complicated algorithm. CONSTITUTION:A collision danger degree decision means 1 is provided to calculate the degree of collision danger in an area, where the obstacle is existent, by setting a multidimensional membership function expressing the adaptation of a fuzzy set in the said area so as to avoid the obstacle under stopping or moving and by calculating the adaptation while using the multidimensional membership function, and a control command generating means 2 is provided to generate a control command for the moving object under moving in the area to avoid the obstacle according to the value of the collision danger degree, and to output the command.

Description

[Detailed description of the invention]

0001

[Industrial Application Field] The present invention utilizes fuzzy control to determine the degree of risk of collision with an obstacle such as a stationary or moving object, and performs collision avoidance operations based on the risk. Regarding equipment.

0002

2. Description of the Related Art Determining the degree of collision risk using fuzzy control and thereby automating collision avoidance has been proposed as a collision avoidance system for ships ("Automation" Vol. 33, No. 6). This collision avoidance system calculates two quantities: the distance between the two ships at their point of closest approach (CPA) (closest approach distance DCPA), and the time it takes for the two ships to reach CPA (closest approach time TCPA). Based on this, the collision risk is determined according to fuzzy inference rules, and collision avoidance actions are automatically performed based on the collision risk.

0003

[Problems to be Solved by the Invention] However, the above-mentioned collision avoidance system is intended for objects that move in a nearly straight line, such as ships, and is intended for moving objects that move essentially differently from ships, such as Like robots and automated guided vehicles,
It is difficult to apply this method to avoid collisions with obstacles such as walls and desks for objects that change direction frequently and at steep angles, and is not practical. Furthermore, the method used in the above-mentioned collision avoidance system to avoid a collision by determining the collision risk is an algorithm that targets objects that move in a nearly straight line, such as ships, and is It cannot be applied to avoidance of moving obstacles that randomly change direction or speed within a limited range, such as when a guided vehicle avoids a person walking without noticing the automated guided vehicle.

[0004] Also, find the distance of closest approach or the time of closest approach,
In order to determine the degree of collision risk using a virtual space, such as by assigning a membership function to a platform set and performing fuzzy inference, and to determine control commands based on this, it is difficult for anyone other than those directly involved in the design and manufacturing to determine the degree of collision risk. It is difficult for third parties (users, etc.) to understand under what circumstances and what kind of judgment will be made. Therefore, it is difficult for users to change or adjust the avoidance control procedure.

On the other hand, regarding robots and automatic guided vehicles,
There are known methods that use complex algorithms to avoid collisions, but in any case, because the algorithms are complex, there are cases where a third party (such as a user) wants to change the criteria for determining the degree of collision risk. Difficult to handle.

Therefore, the object of the present invention is applicable to moving objects such as robots, automatic guided vehicles, and airplanes that often change direction more frequently and at steeper angles than ships, and also to complex objects. An object of the present invention is to provide an obstacle avoidance device that realizes an avoidance control operation that is easy for a user to understand and adjust without using any algorithm.

Another object of the present invention is to provide an obstacle avoidance device equipped with a control algorithm for avoiding moving obstacles that randomly change direction and speed within a limited range.

Another object of the present invention is to provide an obstacle avoidance device equipped with an avoidance control algorithm that realizes accurate avoidance operations even for a plurality of moving obstacles.

[0009]

[Means for Solving the Problems] The present invention provides a device that performs an operation to avoid a stationary or moving obstacle, using a multidimensional member that can represent the fitness of a fuzzy set within a region where the obstacle is located. collision risk determining means for determining the collision risk within the region by setting a ship function and calculating the degree of fitness using the multidimensional membership function; The present invention is characterized by comprising a control command generation means for generating and outputting a control command for a moving object to avoid the obstacle.

[0010] In one aspect of the present invention, in addition to the above configuration, a predicted value of the collision risk after the present time is calculated according to a predetermined algorithm based on the position, speed, and traveling direction of the obstacle and the moving object. A means of predicting the situation is provided,
The collision risk determination means includes an obstacle detection unit that detects the obstacle, a moving object measurement unit that measures the speed and direction of movement of the moving object, and changes depending on the speed and direction of movement of the moving object. a multidimensional membership function storage unit that stores a multidimensional membership function that includes variables and is used in correspondence with the region; a calculation unit that calculates the degree of fitness using a multidimensional membership function stored in the unit, and calculates a value of an equal fitness line intersecting the obstacle based on a detection signal from the obstacle detection unit; a collision risk output unit that outputs the calculated degree of compliance as a collision risk with respect to the obstacle; a storage unit, a target position detection unit that detects the target position of the moving object, a previous collision risk sent from the collision risk storage unit, and a current collision risk sent from the collision risk determination means. , based on the predicted value of the collision risk after the current time sent from the situation prediction means and the target position sent from the target position detection section,
a calculation unit that calculates the traveling direction and speed of the moving object to avoid the obstacle according to a predetermined algorithm; and a control command output unit that generates the control command from the traveling direction and speed calculated by the calculation unit. It is composed of:

[0011] In still another aspect of the present invention, the moving object is an object moving within a region represented by a two-dimensional plane, and the multidimensional membership function is configured to provide an equal fit of a curved shape within the region. A three-dimensional membership function representing a line, and the collision risk determining means determines the collision risk in the region based on a line of equal compatibility that changes depending on the speed and traveling direction of the object, and generates the control command. The means is to obtain the past collision risk level a predetermined time before the current time, the current collision risk level, and the predicted collision risk level after a predetermined time period from the current time point, and calculate a comprehensive evaluation value from these three collision risk values. , the direction of movement of the object to avoid the obstacle is determined from the overall evaluation value, the predicted collision risk, and the direction of the target position, and is output as a control command.

0012

According to the present invention, the degree of collision risk within the area into which the moving object is attempting to enter is determined by calculating the degree of compatibility using a multidimensional membership function.

[0013] Furthermore, by detecting an obstacle within a dangerous area, the value of the degree of compatibility with the obstacle is determined as the collision risk, and the speed and direction of movement of the moving object are adjusted based on this to avoid a collision. can do.

The present invention is suitable as a means for accurately determining the risk of a collision in order to avoid a collision of a moving vehicle moving in a region represented by a two-dimensional plane, but it can It can also be applied to avoid collisions between objects that move freely in space.

[0015]

Embodiment FIG. 1 is a block diagram showing the configuration of an obstacle avoidance device according to an embodiment.

The configuration of this device includes a collision risk determining means 1 for determining the collision risk within an area where an obstacle exists, and a control command for avoiding the obstacle based on the value of the collision risk determined thereby. Control command output means 2 that outputs
and a situation prediction means 3 that predicts the situation from information about the moving object and information about obstacles in the area in which the moving object is moving.

Collision risk determination means 1 detects the position, speed, and direction of movement (angle with respect to a certain target direction) of obstacles as a means for obtaining information regarding obstacles in the area in which a moving object is moving. It is equipped with an obstacle detection section 11 that converts the detection signal into a digital signal for calculation processing, and a signal conversion section 12 that converts the detection signal into a digital signal for arithmetic processing. It includes a measurement section 13 that measures the turning angle, and a signal conversion section 14 that converts the measurement signal into a signal for arithmetic processing. Further, a multidimensional membership function storage unit 15 storing a multidimensional membership function to be described later, and the moving object measurement unit 1
According to the measurement signal from 3, the degree of fitness is calculated using the multidimensional membership function stored in the multidimensional membership function storage unit 15, and the detection signal from the obstacle detection unit 11 is used to intersect with an obstacle, etc. It includes a calculation unit 16 that calculates the value of the compatibility line, and a collision risk output unit 17 that outputs the maximum compatibility calculated by the calculation unit 16 as a collision risk with respect to the obstacle.

In the above-mentioned collision risk determination means 1, the obstacle detection unit 11 is, for example, an image sensor that detects the size, position, etc. of an obstacle to a vehicle moving on a flat surface using light. Consists of. Further, the signal converter 12 is configured with an A/D converter that converts the signal related to the obstacle sent from the obstacle detector 11 into a digital signal. Furthermore, the moving object measurement unit 13 is configured, for example, with a measuring instrument such as a speedometer mounted on a vehicle that moves on a plane. The signal converter 14 is configured with an A/D converter built into those measuring instruments or externally attached.

The membership function storage unit 15 stores a multidimensional membership function that includes variables that change depending on the speed and direction of movement of the moving object, as will be described later, and is set for the area in which the obstacle is going to move. It consists of a memory that stores.

The calculation unit 16 and the collision risk output unit 17 are
It is composed of a CPU programmed to execute a collision risk determination operation, which will be described later.

Next, the control command output means 2 for the avoidance operation includes a collision risk storage section 21 that stores the collision risk output from the output section 17 of the collision risk determination means 1;
A target position detection unit 22 detects the target position of a moving object, a signal conversion unit 23 converts the detection signal into a signal for calculation processing, and a parameter ( The avoidance algorithm storage unit 24 stores an algorithm for calculating the moving speed and the angle indicating the direction of travel, and the collision risk before the current time sent from the collision risk storage unit 21 and the collision risk determination means 1. a calculation unit that calculates parameters for avoiding an obstacle according to an algorithm stored in the avoidance algorithm storage unit 24 based on the current collision risk level sent from the output unit 17 and the target position detected by the detection unit 22; 25, and a control command output unit 26 that outputs a control command for realizing an operation to avoid an obstacle based on the parameters calculated by the calculation unit 25. The control command is supplied to the drive mechanism of the moving object as a signal for changing or adjusting the moving speed and direction of movement of the moving object.

Further, the situation prediction means 3 includes a prediction algorithm storage unit 3 storing an algorithm for calculating the collision risk level (predicted value) at a time after the current time, as described later.
1, and prediction of the collision risk according to an algorithm stored in the prediction algorithm storage unit 31 based on information regarding obstacles and moving objects (detected values and measured values such as positions) sent from the collision risk determination means 1. The calculation unit 32 includes a calculation unit 32 that calculates a value. The predicted value is sent to the calculation section 16 of the collision risk determination means 1.

[0023] It should be noted that 3 components constituting the obstacle avoidance device shown in FIG.
The arithmetic units 16, 25, and 32 included in each of the two means are constituted by one CPU as hardware, and a program for executing the arithmetic operations of each arithmetic unit is installed in the CPU.

Next, the principle and method for determining the collision risk in the collision risk determining means 1 will be explained.

First, a multidimensional membership function corresponding to the area into which the moving object is to enter is set. This can be obtained as a dynamic multidimensional membership function by variously changing the following parameters that determine the shape of the multidimensional membership function to different values depending on the situation.

For example, consider a three-dimensional membership function that provides a parabola of equal fitness as shown in FIG. 2 in a three-dimensional space consisting of x-y orthogonal coordinates and fitness.

The symbols in the figure are defined as follows. For simplicity, the origin is assumed to be the reference point (the center point for describing the three-dimensional membership function most simply).

F (x, y, rx, ry) = 0: Function giving a parabolic shape of the line of equal fitness tx = f (x): Bell-shaped membership function in the x-tx plane ty = g (y) : Bell-shaped membership function on the y-ty plane Rx: Radius in the x-axis direction of a parabolic equifitability line with a fitness degree of 1 Ry: Radius in the y-axis direction of a parabolic equifitness line with a fitness degree of 1 ax: A parameter proportional to the fuzzy entropy for x ay: A parameter proportional to the fuzzy entropy for y The distance ry on the x-tx section of The point given by the membership function (x
, y), the suitability t will be described separately into tx and ty, but t, tx, and ty are practically the same coordinate axes.

[0029] In this case, the iso-fitness line including any point (x, y) is as follows.

[0030]

[Math 1]

Furthermore, the membership functions for tx and ty are respectively

[0032]

[Math 2]

[0033]

[Math 3]

It is expressed as follows. Here, the double sign - represents the part where x is on the positive side, and + represents the part where x is on the negative side.

[0035] On the x-tx cross section and the y-ty cross section, the distances between the equal fitness line of interest and the reference point are respectively
Since it becomes Rx+rx, Ry+ry,

[0036]

[Math 4]

[0037]

[Math 5]

Therefore, from these

[0039]

[Math 6]

At this time, the denominators of the first and second terms (
) must have the same shape, it cannot be transformed into an explicit function with respect to t. Therefore,

[0041]

[Math 7]

[0042] Find the appropriate value of s from the range of 0<s<sk,

[0043]

[Math. 8]

Find dx and dy that satisfy the following. That is,

[
0045

[Formula 9] dx=Rx-axs
...(9)

[0046]

[Formula 10] dy=Ry-ays
...(10) is obtained, and the function of the equal fitness line is transformed as shown in the following equation.

[0047]

[Math. 11]

Here, the double sign - represents a part where x is on the positive side, and + represents a part where x is on a negative side.

[0049]

[Math. 12]

The geometrical meaning of this equation is shown in FIG. This indicates that the shape is approximated by adding a straight line to the original parabola. From equation (12)

[0051]

[Math. 13]

[0052] However, when |x|< dx, x =
When dx, |y|<dy, calculate as y = dy.

Further, as shown in FIG. 2, when the equal fitness line of interest is outside the parabola with a fitness level of 1, the values inside { } are positive.

Similarly, exactly the same equation can be obtained when the isofitness line of interest is inside a parabola with a fit of 1. However, in this case, if any point (x, y) exists on the straight line part of the approximated line of equal fitness, |x|<dx and |y|<dy, and the values inside { } are positive. , when it is in the parabola part, the value inside { } is negative.

If the inside (or outside) of the parabola with a fitness level of 1 is uniformly in the area where the fitness level is 1, it is determined based on the above conditions and a value is given.

On the other hand, instead of using the above three-dimensional parabolic membership function, the above three
The same effect as when using a dimensional parabolic membership function can be obtained. The composition operation for this purpose will be explained below.

First, outside the parabola with goodness of fit 1, x
, the fitness of y is

[0058]

Similarly, for the inside of the parabola with goodness of fit 1,

[0060]

[0061] Therefore, a composite arithmetic expression as shown below can be obtained from these.

(i) For the tip side (y<0)

00
63]

[Math. 14]

(ii) In the case of the opposite side (y>0)

006
5]

[Math. 15]

◇ is a symbol indicating a composition operation (eg, algebraic product).

The meaning of the double sign is that + means outside the parabola with a fitness of 1, and - means inside the parabola with a fitness of 1.

[0068] However, when |x|<dx, tx = (t
x) When x=dx, |y|<dy, ty = (ty)
Calculate as y=dy. The synthesis position of interest has a fitness of 1
If it is outside the parabola, the values inside { } will be positive. In addition, when the synthesis position of interest is inside a parabola with a fitness of 1, when synthesis is performed on the straight line part of the equal fitness line,
If |x|<dx and |y|<dy, the values inside { } will be positive, and when the parabolic parts are combined, the values inside { } will be negative.

y>−Ry, |x|<Rx (or y<
-Ry, |x|>Rx), if tx,ty uniformly has a fitness of 1, it is determined based on the above conditions and a value is given. That is, even in the range of fitness 1, it is assumed that the above membership function exists, and the fitness of each is determined and combined.

[0070] Using the above composition calculation formula, it is possible to compose conventional membership functions using parabolic equifitness lines, and the same effect as when using the three-dimensional parabolic membership function described above can be obtained. . That is, this composition calculation method makes it possible to set a fuzzy set having a parabolic boundary shape, which has not been possible in the past, using a conventional calculation device that calculates membership functions.

Next, a case will be explained in which the above three-dimensional parabolic membership function and conventional membership function are extended to (n+1) dimensions.

This is the formation of a uniform fitness surface using an elliptic paraboloid as shown in Figure 4, and the following (n+1)-dimensional elliptic membership function and (n+1)-dimensional elliptic composition calculation formula are obtained. .

[0073]

[0074] However, 0<s<sk, sk=min(Ri/a
i) When di = Ri −ais |xi| < di, calculate as xi=di.

[0075] If inside { } is positive, outside the input point { }
When inside is negative and all |xi| < di, input point inside (i) tip side (xn<0)

[0076]

[Math. 16]

(ii) In the case of the opposite side (xn>0)

007
8]

[Math. 17]

The meaning of the double sign is that + means outside the paraboloid with a fitness of 1, and - means inside the paraboloid with a fitness of 1.

[0080] However, 0<s<sk, sk=min(Ri/a
i) When di = Ri-ais |xi|<di, calculate as ti=(ti)xi=di.

[0081] If inside { } is positive, outside the synthesis position {
} is negative and all |xi|

[0082]

[Math. 18]

Furthermore, the position of the center of the parabola is (A, B),
When the angle of the axis is φ, the following equation is obtained by moving the three-dimensional membership function.

[0084]

[Math. 19]

[0085] However, X = (x-A) cosφ+(y-B) sinφY
=-(x-A) sinφ+(y-B)cosφψ'=
ψ−φ As the simplest shape, Rx = Ry = 0, s
=0, φ=0, tth=1, we get

[0086]

[Math. 20]

FIG. 5 shows the shape of the bell-shaped membership function expressed by this equation.

[0088] Also, the composition calculation expression is

[0089]

For the denominator, select - when y is on the positive side, and select y
Select + when is on the negative side (tip side).

Similarly, the simplest shape in the (n+1) dimension is found as follows.

[0092]

For the denominator, select - when xn is positive, and xn
Select + when is on the negative side (tip side). Here, as an example of a moving object, consider a traveling vehicle 40 such as a transport vehicle that automatically travels on a flat surface as shown in FIG. This is configured, for example, as a tricycle-type vehicle having one front wheel 41 for changing direction and two rear wheels 42 and 43 for driving. This running car 40
The plane on which the vehicle travels is represented by an x-y coordinate plane, the origin of which is placed at the position of the front wheel 41 of the vehicle 40, and the direction in which the vehicle 40 moves forward is defined as the y-axis.

In this case, the risk of collision with an obstacle within the area (two-dimensional plane) into which the vehicle 40 attempts to enter is defined by the above-mentioned bell-shaped parabolic membership function (FIG. 5). It can be expressed as the goodness of fit. That is, (
From equation 20), the collision risk d is expressed as follows.

[0095]

[Math. 21]

In the above formula, X=+{x−f(v) sin θ}cos θ+ {
y-f(v) cos θ}sin θ Y=-{x-f(v) sin θ}sin θ+ {
y−f(v) cos θ}cos θ d: Collision risk (degree of suitability) f(v)
: Most dangerous distance (distance where the danger level becomes 1) f(v) = A
v: Element that widens the dangerous area in proportion to the speed v of the traveling vehicle f(v) = Av2: Element A that widens the dangerous area in proportion to the square of the speed v of the traveling vehicle A: Constant of proportionality (e.g. automatic transportation Parameters used when adjusting according to the type of cargo carried by the car) θ: Turning angle (front wheel steering angle) [Turning to the right is considered positive] ax, ay: Parameters for adjusting the size of the danger area Using the above equation (21), the collision risk d in the area in which the vehicle 40 is about to proceed is calculated as shown in FIGS. 7 and 8.
become that way. In these figures, by using the three-dimensional parabolic membership function as described above, lines with the same collision risk (equal fitness lines) are represented by parabolas. Furthermore, when there is an obstacle 50 that moves within the illustrated area, the collision risk against the obstacle 50 is determined by the value of the isocompatibility line that intersects with the obstacle 50 (in the case of FIG. 7, d=0.2, Figure 8
In this case, it is expressed as d=0.4).

Next, it is assumed that the degree of collision risk with respect to the moving obstacle 50 described above is detected at every predetermined sampling time dt, and a certain point in time (current time) is t=t0, and the previous sampling time is t. = t-1, FIG. 7 shows t=
It represents the collision risk d-1 at t-1, and FIG. 8 shows t=
Represents the collision risk d0 at t0. The collision risk determined at each sampling time is stored in the collision risk storage unit 21 in FIG.

Furthermore, if the sampling time one time after the current time t0 is t1, then FIGS. 9 and 10 show t1.
= represents the collision risk (predicted value) d1 at t0 +dt. A method for obtaining this predicted value d1 will be explained below.

The above-mentioned moving obstacle 50 randomly changes its traveling direction and speed within a limited range, but in this case, the traveling direction at the present time t=t0 is the same as the traveling direction at the same time t=t0.
Assuming uniform linear motion with a speed of 0, t1
The position of the moving obstacle 50 at =t0+dt is determined. Next, from t0 to t1, centering on that position
Find an elliptical fuzzy region with a size proportional to the moving distance between The meaning of this elliptical fuzzy region is "t=t1
= t0 + dt, the moving obstacle 50 will come around here.'' This represents an ambiguous area. At this time, the elliptical fuzzy region is given by the three-dimensional membership function of the following equation.

[0100]

[Math. 22]

In the above formula, X=+ (x-xp) cosψ+ (y-yp) s
inψY=- (x-xp) sinψ+ (y-yp
) cosψp: goodness of fit for the membership function of the fuzzy label "will come around here" xp: x-coordinate yp of the position at t=t0 + dt when the moving obstacle is assumed to move in a uniform straight line:
t = assuming that the moving obstacle moves in a straight line at a constant velocity
The y-coordinates apx, apy of the position at t0 + dt:
Parameter vp that determines the size of the elliptical area: t=t0
Speed ψ of the moving obstacle at t=t0: direction of movement of the moving obstacle at t=t0 (angle with respect to the x-axis) In this case, from t=t0 to t=t0 + dt, the vehicle itself (traveling vehicle 40) also moves in a uniform linear motion. It is assumed that For this uniform linear motion of the user, the speed at t=t0 and the direction (angle) obtained by correcting the traveling direction at t=t0 by a certain amount toward the goal destination are used. Then, as mentioned above, when the degree of fitness is maximum at the overlap between the three-dimensional parabolic membership function representing the dangerous area and the three-dimensional elliptic membership function giving the position of the moving obstacle, the maximum degree of fitness is determined by the collision. Predicted risk value d1
shall be.

Next, the three collision risks mentioned above, namely the collision risk d-1 at t=t-1, the collision risk d0 at t=t0, and the collision risk (predicted value) at t=t0 + dt, are calculated. d1, the comprehensive evaluation value D is determined by the following formula.

[0103]

[Formula 23] D=d1 + (d0・d1)2 + (
d-1 ・d0・d1)3 ...(23)
The second and third terms in the above equation are meaningfully large values only when d-1, d0, and d1 are close to 1, respectively, but since the dangerous region is expressed by a parabolic membership function, As a result, if the obstacle continues to exist in a position close to directly in front of the user, the second and third terms become values that cannot be ignored.

Next, a procedure for obtaining a control command for collision avoidance will be explained. There are two ways to avoid this.

One is as shown in FIG. 9, t=t0 +
dt, the point P (position of a moving obstacle) that gives the predicted value d1 of the collision risk does not pass in front of you (the traveling vehicle 40), and the other case is when the point P gives the predicted value d1 of the collision risk.
This is the case when . In the former example (Figure 9), you turn to the left and increase your speed to pass in front of the obstacle; in the latter example (Figure 10), you turn to the right. Take the method of moving around the obstacle by changing your speed and slowing down.

Note that the examples shown in FIGS. 7 and 8 are cases where a parabolic membership function is applied to the area into which a moving object attempts to enter, but the application is not limited to this, and iso-fitness lines of other shapes can also be applied. You may also use a membership function with . As an example, when an elliptic membership function is used, the result is as follows.

In this case, the method for determining the elliptic membership function is basically the same as that for the parabola described above, so the explanation will be omitted and the results will be shown.

As mentioned above, the plane on which the traveling vehicle 40 travels is represented by an xy coordinate plane, and the origin thereof is defined by the front wheels 41 of the traveling vehicle.
, and the direction in which the vehicle 40 moves forward is set as the y-axis, the area (2
The collision risk within the dimensional plane) is shown in Figures 11 and 12.
It can be expressed as the degree of fitness defined by a triangular elliptic membership function as shown in . That is, the collision risk level d is expressed as follows.

[0109]

[Math. 24]

[0110]X=+x cosθ+y sinθY=-
x sin θ+y cos θ According to the above equation (24), the collision risk in the dangerous area can be expressed by an elliptical isofitability line.

Furthermore, as shown in FIG. 13, the degree of danger in the dangerous area in front of oneself (driving vehicle 40) is evaluated using a parabolic membership function, while the degree of danger behind oneself (driving vehicle 40) is evaluated using an elliptic membership function. Can be done.

Next, a case where there are a plurality (n) of moving obstacles will be explained.

[0113] When the i-th (i = 1, 2, ..., n) moving obstacle is targeted, the correction angle (the front wheel angle Correction of angle θ) θ
iN is given by the following formula.

[0114]

[Formula 25] θiN=di1・(θ±θA)+ (1−
di1)・θ0 ... (25) In the above formula, θA = RCT・Di (RCT is a proportionality constant) θA:
Avoidance angle θ0: Direction angle di1 of target position: Time t1 with respect to the i-th moving obstacle = t0
+ Collision risk level (predicted value) at dt Di: Comprehensive evaluation value of collision risk level for the i-th moving obstacle In the sign ±, + indicates a case of avoiding to the right, and - indicates a case of avoiding to the left.

[0115] When the target is n moving obstacles, among the moving obstacles that the user must avoid to the right, the one with the maximum overall evaluation value D (referred to as the j-th moving obstacle) The correction angle θjN and the modified angle θkN for the moving obstacle that the user must avoid to the left that has the maximum overall evaluation value D (assumed to be the k-th moving obstacle), regarding D. Try to take a weighted average. That is, the corrected meandering angle θN for avoiding n moving obstacles is determined by the following equation.

[0116]

[Formula 26] θN = (Dj・θjN+Dk・θkN)
/(Dj +Dk) (26) Regarding speed correction, deceleration and acceleration are similarly performed in proportion to the comprehensive evaluation value D. When targeting n obstacles, a weighted average may be taken using the overall evaluation value D. In practice, the speed should not change much from a certain speed, and avoidance can be achieved mainly by the above-mentioned meandering angle adjustment.

FIGS. 14 to 17 show the case where n=3, that is, there are three
When there are three moving obstacles A, B, and C, the figure shows temporal changes in avoidance operations for these obstacles. In these figures, the sequentially arranged small circles represent the number of running vehicles 4 at each sampling time.
The positions of 0 and 3 moving obstacles A, B, and C are shown. Furthermore, black circles indicate positions at the same time.

Further, FIGS. 18 and 19 each show an example of an avoidance operation when there is one moving obstacle (indicated by A').

From these simulation examples, it can be seen that according to the above avoidance method, the vehicle 40 can reach the target position α while successfully avoiding one or more moving obstacles A, B, and C. Proven.

The illustrated embodiment has been described above, but
The present invention is not limited to this. For example, the membership function is three-dimensional when the moving object travels on a two-dimensional plane as in the example, but when moving in three dimensions like an airplane, the membership function is four-dimensional. used. Furthermore, membership functions of arbitrary dimensions and shapes can be used as necessary. Moreover, the obstacle may be either a stationary object or a moving object.

[0121]

[Effects of the Invention] The present invention uses a multidimensional membership function as described above to easily determine the degree of collision risk. The algorithm is not complicated because collision avoidance is achieved for moving objects that have a large number of collisions, and the collision risk is determined by the degree of fitness expressed by a multidimensional membership function. Further, by generating a control command based on the collision risk level and its rate of increase (for example, a control command based on fuzzy inference), a collision avoidance operation that appropriately corresponds to the characteristics of the moving object and the region in which it moves can be realized.

[0122] By simply changing the parameters of the membership function as appropriate, it is possible to easily recognize the degree of collision risk depending on the area in which the moving object is moving. A good collision avoidance system can be built.

Furthermore, by using the above-mentioned three-dimensional membership function, the risk of collision can be expressed graphically, making it possible to display information that suits human senses. Therefore, a collision avoidance device having a visual display function that is easily understood by humans is realized.

[0124] If the present invention is used in an automated moving device such as an automatic guided vehicle or a traveling robot used in a factory,
Engineers (end users) in the factory can easily adjust the level of avoidance control, and since predictive control is possible, reliable interpersonal avoidance can be achieved in the factory, preventing accidents resulting in injury or death or collisions. can do.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing the configuration of an obstacle avoidance control device according to an embodiment of the present invention.

FIG. 2 is a diagram showing a three-dimensional parabolic membership function used in the present invention.

FIG. 3 is an explanatory diagram of a case in which a straight line portion is added to the three-dimensional membership function of FIG. 2;

FIG. 4 is a diagram illustrating an elliptic paraboloid forming an equifitness surface of a multidimensional membership function.

FIG. 5 is a diagram showing the shape of a bell-shaped three-dimensional parabolic membership function used in the example.

FIG. 6 is a diagram showing the positional relationship between a traveling vehicle as a moving object and a coordinate plane in the embodiment.

FIG. 7 is a diagram showing the collision risk at the previous sampling time in the region in which the vehicle in FIG. 6 is moving;

FIG. 8 is a diagram showing the current collision risk in the area in which the vehicle in FIG. 6 is moving;

FIG. 9 is a diagram showing predicted values of collision risk at the time of one sampling in the region in which the vehicle in FIG. 6 is moving;

FIG. 10 is a diagram showing predicted values of collision risk when an obstacle passes.

FIG. 11 is a diagram showing an example of a three-dimensional elliptic membership function used in the present invention.

FIG. 12 is a diagram showing the shape of a triangular three-dimensional elliptic membership function.

FIG. 13 is a diagram showing the degree of collision risk using an elliptic membership function in a region behind a traveling vehicle.

FIG. 14 is a diagram showing an example of an avoidance operation for three moving obstacles.

15 is a diagram showing a continuation of the avoidance operation shown in FIG. 14. FIG.

16 is a diagram showing a continuation of the avoidance operation shown in FIG. 15. FIG.

17 is a diagram showing a continuation of the avoidance operation shown in FIG. 16. FIG.

FIG. 18 is a diagram showing an example of an avoidance operation for one moving obstacle.

FIG. 19 is a diagram showing another example of an avoidance operation for one moving obstacle.

[Explanation of symbols]

1... Collision risk determination means, 2... Control command output means, 3...
Situation prediction means, 11... Obstacle detection section, 12... Signal conversion section, 13... Moving object measurement section, 14... Signal conversion section, 15... Multidimensional membership function storage section, 16... Calculation section, 17... Collision risk level Output section, 21... Collision risk storage section, 22... Target position detection section, 23... Signal conversion section, 24... Avoidance algorithm storage section, 25... Calculation section, 26... Control command output section, 31...
Prediction algorithm storage unit, 32... Calculation unit, 40... Traveling vehicle, 41... Front wheel, 42, 43... Rear wheel, 50... Obstacle.

Claims (6)

[Claims] Claim 1: A device that performs an operation to avoid a stationary or moving obstacle, in which a multidimensional membership function that can represent the fitness of a fuzzy set is set within a region where the obstacle is located, and Collision risk determining means calculates a degree of collision risk within the region by calculating a degree of conformity using a dimensional membership function; and a collision risk determining means that determines the degree of collision risk within the region; An obstacle avoidance device comprising: control command generation means for generating and outputting a control command for avoiding obstacles. 2. The obstacle avoidance device according to claim 1,
Situation prediction means is provided for calculating a predicted value of a future collision risk according to a predetermined algorithm based on the position, speed, and traveling direction of the obstacle and the moving object, and the collision risk determination means an obstacle detection unit that detects an object, a moving object measurement unit that measures the speed and direction of movement of the moving object, and a variable that changes depending on the speed and direction of movement of the moving object and that corresponds to the area. a multidimensional membership function storage unit storing a multidimensional membership function to be used; and a multidimensional membership function stored in the multidimensional membership function storage unit in accordance with a measurement signal from the moving object measurement unit. a calculation section that calculates the degree of compatibility using the above-mentioned obstacle detection section, and calculates the value of a line of equal compatibility that intersects with the obstacle based on the detection signal from the obstacle detection section; and a collision risk level output unit that outputs a level of risk, and the control command generating means includes:
a collision risk storage unit that stores the collision risk output from the collision risk determination means; a target position detection unit that detects the target position of the moving object; based on the current collision risk sent from the collision risk determination means, the predicted value of the collision risk after the current time sent from the situation prediction means, and the target position sent from the target position detection section. , a calculation unit that calculates the traveling direction and speed of the moving object to avoid the obstacle according to a predetermined algorithm; and a control command output unit that generates the control command from the traveling direction and speed calculated by the calculation unit. An obstacle avoidance device comprising: 3. The obstacle avoidance device according to claim 1, wherein the moving object is an object moving within a region represented by a two-dimensional plane, and the multidimensional membership function is configured to move within the region represented by a two-dimensional plane. It is a three-dimensional membership function representing a curved line of equal compatibility, and the collision risk determining means determines the collision risk in the area based on the line of equal compatibility that changes depending on the speed and traveling direction of the object. The control command generating means obtains the past collision risk level a predetermined time before the current time, the current collision risk level, and the predicted value of the collision risk level after a predetermined time period from the current time point, and calculates these three collision risk levels. An overall evaluation value is determined from the value of , and a traveling direction of an object to avoid the obstacle is determined from the overall evaluation value, the predicted value, and a direction of the target position, and is output as the control command. Avoidance device. 4. The apparatus according to claim 3, wherein a parabolic membership function is applied to a region in front of the moving object;
An obstacle avoidance device characterized in that an elliptical membership function is applied to a region behind the moving object. 5. The obstacle avoidance device according to claim 2, wherein the predicted value of the collision risk is determined based on the movement of the obstacle for a predetermined period of time, centered on the predicted position of the obstacle after a predetermined period of time. Find a fuzzy region given by a three-dimensional elliptic membership function whose size is proportional to the distance, and calculate the degree of fitness that is maximized at the overlap between the three-dimensional elliptic membership function and the three-dimensional membership function representing the region. An obstacle avoidance device characterized by: 6. The obstacle avoidance device according to claim 3, wherein the object is a plurality of moving obstacles, and the plurality of moving obstacles that the moving object must avoid to the right are The total of the traveling direction correction angle for an obstacle that gives a maximum evaluation value and the traveling direction correction angle for an obstacle that gives a maximum comprehensive evaluation value among moving obstacles that the moving object must avoid to the left. An obstacle avoidance device characterized in that a correction angle for avoiding the plurality of moving obstacles is determined by taking a weighted average of the evaluation values.