JPH01260509A - Travel controller - Google Patents

Abstract

PURPOSE:To perform travel following an orbit even when it is the one whose dynamic characteristic is unknown by using a quantity of dislocation as an ambiguous value, and setting the experimental knowledge of the human as a control rule. CONSTITUTION:The title device is provided with an image processing device 9 which digitizes information obtained from a camera 10 which photographs a targeted path and calculates the quantity of dislocation from the path, and the device infers a controlled variable from the quantity of dislocation by a CPU1, a ROM2, and a RAM3, and outputs it to an actuator 6 via an I/O interface 5. Here, visual information obtained from the camera 10 is constituted so that it can decide the manipulated variable of the steering angle of a traveling vehicle by a fuzzy inference. In other words, the quantity of dislocation is handled as an ambiguous quantity, and the experimental knowledge of the human is used as the control rule. In such a way, it is possible to perform the travel following the orbit even when it is the one whose dynamic characteristic is unknown.

Description

DETAILED DESCRIPTION OF THE INVENTION [Object of the Invention] (Industrial Application Field) The present invention relates to a travel control device for a robot used for inspection and repair work in nuclear power plants and the like.

(Prior Art) Robots are used in nuclear power plants and the like for inspection and repair work. In order to operate such robots safely and efficiently, image information from cameras has been used. That is, a camera is used to take pictures of marks indicating the target path, and the steering is controlled so that the vehicle travels along the marks.

However, such a travel control device has been operated using a control method based on the following equation (O).

In other words, y is the positional deviation from the target path, θ is the direction deviation from the target path, and east is the steering angle, and the steering angle ψ is as follows (
1) It is expressed by the formula.

-2=1V=-(k,y+k,0)
(],)Tap,k,,constant,then y,0 is estimated by Kalman filter, 1-noted ○
] formula, the steering angle 'IJ+ was determined, and the vehicle was controlled to travel in accordance with this steering angle ψ.

In such a travel control system, the constants of , , , , etc. are determined by repeating preliminary travel.

(Problem to be solved by the invention) In the conventional driving control method described above, constants , , and 2 have to be determined while evaluating the characteristics of the traveling vehicle, which takes a lot of time. .

The present invention was made to solve these drawbacks, and its purpose is to use the amount of deviation as an ambiguous value, and to use human experiential knowledge as a control rule, so that it can be applied to objects whose dynamic characteristics are unknown. It can also run while following the trajectory,
Furthermore, it is an object of the present invention to provide a travel control device that can smoothly converge to the Sakura trajectory without the need for gain adjustment.

[Structure of the invention]

(Means for Solving the Problems) In order to achieve the above object, the present invention includes a camera that images the target passage, and an image processing device that digitizes information obtained from the camera and divides the amount of deviation from the passage. , CI-' Infer the control amount from the displacement plate by U, ROM, RA, and M,
A driving control device that outputs output to an actuator via an I10 interface is characterized in that it is configured to determine the manipulated amount of the steering angle of a traveling vehicle by fuzzy inference based on visual information obtained from a camera.

(Function) According to the travel control device of the present invention, since the amount of deviation is treated as an ambiguous amount and human experiential knowledge is used as a control rule, it is possible to perform control that matches human senses without evaluating dynamic characteristics. It is possible.

In addition, the input membership function is distributed in a 4-uniform manner,
By distributing the output membership function so that it is closer to zero, the target position can be approached smoothly and gain adjustment is not necessary.

(Example) Embodiments of the present invention will be described with reference to the drawings.

FIG. 1 is a configuration diagram of a travel control device according to an embodiment of the present invention. As shown in the figure, CPU for fuzzy inference
1, an input/output membership function used for fuzzy inference, an R○M 2 for storing control rules, and a RAM 3 for performing division using the input/output membership functions and control rules are connected to the multipath 4, respectively. In addition, a camera 10 that photographs the path in front of the vehicle and an image processing device 9 that digitizes the information obtained by the camera 10 and divides the amount of deviation of the vehicle from the path by the image calculate the amount of deviation. It is connected to the multipath 1 via a communication interface 8 for outputting and a communication interface 7 for inputting the amount of wear to the computer side. Further, an I/O interface 5 is connected to the multipath 4 for transmitting the operation amount obtained by fuzzy inference from the amount of deviation to the actuator 6 (a motor that changes the direction of the axle of the vehicle). A console 13 for human operation is configured to send signals to an input communication interface 7 via a wireless transmitter 2 and a receiver 11.

Next, the operation of this embodiment will be explained.

The overall flow of control is as follows: First, the camera 10 photographs the path along which the vehicle is traveling. The image processing device 9 calculates the amount of deviation from the image data, and the communication interface 7 calculates the amount of deviation.
．． 8 to the reduction unit 14.

The calculation unit 14 performs fuzzy inference using the membership functions and control rules stored in the ROM 2 to determine the steering angle and outputs it to the actuator 6 via the interface 5. The knob commands for Star 1- and S1 are given by a human using the console 13.

Next, a method for determining the amount of deviation from the image of the passage taken by the camera will be explained in detail with reference to FIG. 2.

FIG. 2 is an explanatory diagram when calculating the amount of deviation by image processing from an image taken by camera 10 of a path along which a vehicle travels.

The passage is indicated by two parallel white lines, and when this is photographed with the camera 10, it appears as shown in line]9 in Fig. 2. This image is sent to the image processing device 9 and subjected to the following image processing. First, the number of meters in front of the vehicle's gaze point is determined, and the mask 18 is set at a location on the screen corresponding to that point. The image data in the mask 18 is subjected to averaging filtering processing, differential processing, binarization using a certain threshold, and by extracting only those with a certain area, the two white line parts 20 can be extracted. . Distance a11 between each extracted white line 20 and the center line 17 of the screen
Find a2 and use this difference to find out how far the vehicle is moving.

Next, a method of calculating the steering angle by performing fuzzy inference from the amount of deviation in the H] calculation section will be described.

The input variables are the amount of deviation e and the amount of change Δey, and the output variable is the steering angle Δ■. etsu, △etsu.

In order to represent △V as an ambiguous value, membership functions are set for each of them. This is e8゜△eX+ △V
The value of NB (negatively large) NM (negatively medium) N
S (negatively small) Z○ (habo 0) PS (positively small)
It is a function that classifies into seven sets of PM (just medium) and PB (just large) and shows how much it fits into which set. Here, the membership functions for e8°△e are uniformly distributed as shown in FIG. 3, and the membership functions for eight ■ are distributed such that the closer to O, the wider the width as shown in FIG.

Control rules for inferring outputs from inputs are written using human experiential knowledge, for example, if eX is PB and △llll, and L is PB, then △V should be PS. Using this rule, multi-adjustment inference is performed to find the steering angle.

In order to output the determined steering angle to the actuator, the steering angle is converted into a voltage according to the relationship between the angle and the voltage set in advance as shown in FIG. 5, and is output to the actuator.

''As mentioned above, according to this embodiment, by setting the deviation amount to an ambiguous value and using human knowledge as the control rule,
It is possible to travel by following the path without evaluating the dynamic characteristics of the controlled object. Further, by distributing the output membership function so that the closer it is to 0, the wider the width, the convergence to the target path is faster and smoother than when the output membership function is uniformly distributed, and no regain adjustment is required.

〔Effect of the invention〕

As explained above, according to the travel control device of the present invention, travel can be performed while following the trajectory using only human experiential knowledge, even if the dynamic characteristics are not known. Furthermore, since the target trajectory can be smoothly reached without the need for gain adjustment, the time taken for gain adjustment can be saved.

[Brief explanation of the drawing]

Fig. 1 is a configuration diagram of an embodiment of the present invention, Fig. 2 is a diagram for explaining a method of photographing the path on which the vehicle is traveling with a camera and calculating the amount of deviation by image processing the image. A membership function that represents a fuzzy table set of input variables, which is used when calculating a manipulated variable from a quantity by fuzzy inference,
Figure 4 shows the membership function representing a fuzzy set of output variables, and Figure 5 shows the motor used to convert the manipulated variable expressed in angle into voltage to be output to the motor that changes the direction of the car. It is a graph showing the characteristics of. 1... CPU 2- ROM3
RAM 4...Multipath 5-I1
0 interface 6 - Actuator 7 Human power communication interface 8 - Output communication interface 9... Image processing device 10... Camera 11 Receiver 12... Transmitter 13-- Operation console
14... Calculation section 15... Traveling vehicle mounting section 16.
・Screen 17 Center of screen 18...Mask 19
Line indicating passage 20... Extracted passage agent Patent attorney Yoshiaki Inomata (and one other person) Figure 1 Figure 5

Claims (1)

[Claims] (1) A camera that extracts information about the path traveled, an image processing device that digitizes the information obtained by the camera and calculates the amount of deviation from the path, a communication interface that outputs the amount of deviation, and the amount of deviation. a communication interface that inputs the information into the travel control computer, a CPU that infers a control amount from the amount of deviation, a ROM that stores input/output membership functions and control rules used for the inference, and a RAM that uses it to perform calculations. In the travel control device, which is composed of an I/O interface for driving the actuator, a console for external commands, and a radio for transmitting operation data to the mobile device, the visual information obtained from the camera is transmitted. A travel control device characterized by being configured to determine a steering angle operation amount using fuzzy inference.