JPS60217413A - Guiding device - Google Patents

Abstract

PURPOSE:To realize the inducing of a mobile conductor in complicated environment at a high speed by extracting an observed picture of a TV camera into plural characteristic patterns at the same time at a picture processor and using the pattern recognized best among them as the guiding index. CONSTITUTION:A mobile body position detecting means 3 detects the position in a path of a mobile body 1 based on the information from a gyro scope 1B and a detector 1C. A characteristic estimation pattern under the environment estimated based on the position from the means 3 is stored in a storage means 4. A picture processing means 5 processes the observed picture of a camer into plural characteristic patterns at the same time and extracts the characteristic pattern effective to the collation. A collation means 6 checks the coincidence between the characteristic pattern from the means 5 with the forecast characteristic pattern from the means 4. A visual control means 7 controls the direction of the camera 2 through the result of comparison of the means 6 and extracts the real picture by a storage means 9. An guidance control means 8 compares the setting value of the camera 2 and the control value, outputs an inducing signal to the mobile body 1 so as to guide the mobile body 1.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Application of the Invention] The present invention relates to a guiding device for a moving body of a ye moving corobot.

[Background of the invention]

Conventionally, devices that control the movement of mobile objects such as mobile robots are
It is composed of environment recognition means and moving body movement mechanism control means. An example of this movement control device is the document published by Carnegie Mellon University in the United States on February 24, 1983.
and TheCMU ROVeJ.

In a control device for a moving object equipped with this type of environment recognition means, it is impossible for the moving object to begin any action or work until recognition of the environment is completed. For this reason, the mobile object has to interrupt its intended operation each time it recognizes the environment, making it difficult to move continuously and perform tasks.

[Purpose of the invention]

The purpose of the present invention is to move while recognizing complex environments.

To provide a guiding device that can be operated. .

[Summary of the invention]

The present invention is characterized in that a device for guiding a moving object along a predetermined route includes a visual means for observing the real environment around the moving object, a means for detecting the route movement of the moving object, and a device for guiding a moving object along a predetermined route. means for storing a predicted pattern related to body position information; an image processing means for extracting an observed feature pattern from an actual image from a visual means; and a predicted pattern obtained from a storage device by the route movement detecting means and an image processing means. visual control means for controlling the direction of the visual means by comparing observed characteristic patterns from the observation means; and guidance control means for comparing the set value and control value of the direction of the visual means and outputting a guidance signal to the moving body. The system predicts the observation results of the outside world based on accumulated prior knowledge and experience about the environment, and by comparing this predicted data with actually obtained observation data, it is possible to predict the direction of moving objects in complex environments. This makes it possible to achieve high-speed guidance.

[Embodiments of the invention]

Hereinafter, embodiments of the present invention will be described with reference to the drawings. First, prior to a detailed explanation of the present invention, a detailed explanation of the present invention will be given. The guidance device of the present invention determines the position of a moving object based on an image taken by a visual means such as a camera mounted on the moving object, on the premise that the environment model and information regarding the movement route are already given. and calculates the direction control signal for the visual means and the guidance signal for the moving object.

To describe the algorithm for obtaining this guidance signal in more detail, as shown in FIG. 1, the prior estimate of the current position of the moving object −', then compare the predicted image f predicted by the prediction means with the observed image fm based on the corresponding predicted image fpt=environmental model information, and find the image error Et. Image error′
Ef is converted into a position error Ex by the exchange means M, and the previously estimated value X0 is corrected based on this position error E to obtain the position estimate X. Based on this estimated position value X, the visual means changes to the desired direction.

be done. Information for matching the current position of the aforementioned visual means to a preset route is always obtained from the sampling rate of the television camera as the position estimation residual Ex.

In the present invention, when comparing the predicted plane #fp and the observed image fvn for position error conversion in the calculation flow for calculating the direction control signal of the television camera described above, a dynamic model is introduced between the patterns of these images. The system uses a dynamic position estimation method obtained by

This position estimating means, as shown in FIG.
A pattern A and a pattern B are arranged within the interior, and each of the patterns A and B is provided with its own operating mechanism M^r Mn
pattern A by operating one or both of the operating mechanisms MA r Mn.

Pattern matching is performed dynamically by operating the pattern so as to reduce the interaction ψAB acting between the patterns.

Next, the interaction ΦAn acting between the patterns described above will be described.

Below, in order to simplify the explanation, as shown in Figure 2,
Let space S be a screen, and patterns A and B be binary images.

Here, pattern A is a pattern based on an observed image, and pattern B is a pattern based on a predicted image generated by an environmental model K. Now, when pattern A is given, it is necessary to find an adaptive rule that brings the position of pattern B closer to pattern A. This adaptation law is obtained by introducing a potential φA generated by pattern A onto space S. This potential φA can be determined by the following equation (1).

αφA+XA=0 (1) Operator. However, σX and σy are parameters representing the spatial spread of the potential φA.

In addition, in equation (1), XA converts pattern A into space S
This is the defined function when considered as a partial result of .

Next, the potential φA of pattern A on space S mentioned above
The effect F'B on pattern B due to the field is as follows (2
) can be obtained from the formula.For convenience of explanation, here, 2
Only movement in the X-axis and y-axis directions within the dimensional plane is targeted.

Fs=fXB-DφAds ・・・・・・・・・・・・
(2) 7'ffL,XB is a defining function when pattern B is identified with a subset of space S, and represents the vI'i gradient operator.

By the above equations (1) and (2), pattern BfK
: It is possible to obtain 1n information for deforming pattern A, that is, direction information of the visual means. However, since equation (1) in this calculation is an infinite-dimensional simultaneous equation, it is preferable to use the following approximate equation (8).

Using the solution of the potential φA of the above equation (8), the output UB of the operating mechanism MB for moving the pattern Bk to the pattern A can be determined by the following equation (4).

UB=-/Xs-DφAd, ・・・・・・・・・
(4) The pattern matching mechanism composed of the above equations (8) and (4) can be implemented in hardware using an image memory and a similar area calculation device.

Based on the principle of the present invention described above, an embodiment of the guiding device of the present invention will be described with reference to FIG. In this figure, a moving body 1 guided by the guidance device of the present invention is equipped with a running section IA such as wheels, endless tracks, etc., and can be moved in a desired direction by a steering device (not shown). The moving object 1 is equipped with a gyroscope IB for detecting the orientation and a detector ICi for detecting the amount of rotation of the moving part IA, which provides information necessary for predicting its own position.

Furthermore, the moving object 1 is equipped with a television camera 2 that captures the situation in front of it as image information. This TV camera 2
The guidance device of the present invention, which controls the direction of the television camera 2 and guides the moving object 1, is equipped with a direction changing means 2Ai. A moving object position detection means 3 detects the position on the route, and an environment feature estimation pattern estimated from the position based on the position of the moving object from this moving object position detection means 3, for example, the parallelism of stairs in a building. A storage means 4 that stores linear patterns such as line groups, boundaries between corridors and walls, windows, etc., and building route information, and observation images from the television camera 2 are simultaneously processed into a plurality of characteristic patterns, an image processing means 5 for extracting an effective feature pattern for matching; and a matching means 6 for examining the match between the observed feature pattern effective from the image processing means 5 and the predicted feature pattern from the storage means 4.
By comparing the observed feature pattern obtained by the matching means 6 with the predicted feature pattern, a signal for controlling the direction of the television camera 2 is calculated, and a signal for controlling the direction of the television camera 2 is calculated.
The vehicle 1 is provided with a visual control means 7 for outputting an output to the vehicle 1, and a guidance control means 8 for the moving body 1. The direction control signal from the visual control means 7 described above causes the television camera 2 to move along the target path as described above. As described above, the visual control means 7 calculates the case of the potential φA acting between the predicted pattern and the observed pattern using the sulhotential field calculation means 7A, and the acting force FB of the observed pattern on the predicted pattern obtained by the field of this potential φA. and an acting force calculation means 7B for calculating the gradient vector of.

The calculation contents of these calculation means 7A and 7B are executed according to the principle of the present invention described above.

As shown in FIG. 4, the image processing device 5 described above includes an input image memo IJ 501 for writing input images, a pattern memory 502 for controlling write addresses to the input image memory 501, and a processing image memo IJ for storing processed images. a memory 503 , an image calculator 504 that calculates a correction value for modifying the gray value of each pixel based on the gray value of neighboring pixels for each pixel of a given image; and an input image memory 501
and a hard logic circuit 505 that executes an operation on word length data corresponding to the larger gray value of the pixel in the processed image memory 503; Switches 506 to 509 that selectively output output, and a memory controller 51 that controls the switches 506 to 508.
1, memory controller 511, pattern memory 50
2, a hard logic circuit 505, an image arithmetic unit 504, and a controller 512 that controls the switch 510i. A plurality of image processing devices 5 are installed in order to simultaneously extract a plurality of observed feature patterns from an input image, that is, an observed image.

This image processing device 5 has a pattern memory 5 as described above.
Since the write address of the input image memory 501 can be controlled by 02 and the input image can be stored in the input image memo IJ 501, there is no need for conversion calculations 1, and distortion correction, position and orientation conversion, etc. can be performed on the image input signal in real time. can be executed.

At this time, the answer in the pattern memory 502 is the controller 5
12 changes can be set. -! In addition, this image processing device 5 includes an image arithmetic unit 504 and a hard logic circuit 505 operated by a controller 512.
It can be switched and used to have the functions of the potential field calculation means 7A and the acting force calculation means 7B described above.

The visual control means 7 controls the direction of the television camera 2 as described above, and also cuts out and controls the actual image stored in the storage means 9.

Further, the guidance control means 8 compares the direction setting value of the television camera 2 with the control value and outputs the guidance signal to the moving object 1. That is, the guidance control means 8 performs the guidance operation of the moving body 1 as shown in Table 1 below with respect to the output of the visual control means, so-called moving direction of the television camera.

Table 1 Next, the operation of controlling the direction of the television camera 2 and guiding the moving object 1 using the above-described embodiment of the guiding device of the present invention will be described.

Now, when the moving object 1 shown in FIG. 3 is moving toward the stairs in front of it and all settings are being made, the television camera 2 photographs the stairs and sends the image signal f to the image processing device 5. Output. At this time, as shown in FIG. 5, the pattern memory 502 in the image processing device 5 is set to a distortion pattern memory in which aberration correction data of the lens of the television camera 2 is written, and the image calculator 504 corresponds to horizontal diffusion. The hard logic circuit 505 is further controlled to perform pixel subtraction. With this setting, when a stair image Fl with distortion is input as shown in FIG. Stored in The staircase image F2 from which distortion has been removed is processed by the hard logic circuit 505 and the image arithmetic unit 504 into a horizontal line image F3 corresponding to, for example, steps of the stairs. This horizon image F3
is stored in the processed image memory 503.

The information of this stored horizontal line image F3 is stored in the switch 5
08,510, it is output to the potential field calculation means 7A of the guidance control device 7.

As described above, this potential field calculation means 7A can be executed with the same configuration as the image processing device 5 shown in FIG. 4. Therefore, FIG. 6 is treated as the configuration of the potential field calculation means 7A. In this case, the pattern memory 5
02 is set to perform the function of enlarging a specific area of the image information from the image processing device 5 to the full screen, that is, the entire cutting function, and the image calculation unit 504 has its calculation parameter set to isotropic diffusion. , Furthermore, the hard logic circuit 505 is set to addition mode.

In this state, the information □ of the horizontal line image F3 from the image processing device 5 shown in FIG. This information F
4 is an image arithmetic unit 50 in a hard logic circuit 505.
Computed signals with isotropic spreading from 4 are added,
It is converted into potential field information F5, which is a measure of the approach to the recognized information F4. This potential field information F5 is stored in the processed image memory 503, and is outputted to the acting force calculation means 7B connected to the switch 510 through the entire switch 510.

This acting force calculation means 7B can be realized with the same configuration as the image processing device 5 shown in FIG. 4, similar to the aforementioned potential field calculation means 7A. The operation of this acting force calculating means 7B will be explained with reference to FIG. In this case, the pattern memory '502 is set as a pattern memory for cutting, the image calculation 6504 is set as a calculation parameter for calculating the gradient in the acting force, and the no-drossic circuit 505 is set in the integration mode. ing. In this state, the pattern memory 502 extracts only part of the information F6 from the potential field information F5 from the potential field calculation means 7A and stores it in the input image memory 501. This information F6 is the hard logic circuit 505
Image performance will be added to $6504 through. Here, this information F6 is calculated into information-4F7 of the gradient field representing the elements of the acting force. This gradient field information F7 shows the movement of the child with respect to the preset direction of the television camera 2. In other words, the direction in which the television camera 2 should face is ti゜ with respect to the set movement route. This gradient field information F7 is stored in the processed image memory 503, and is also output as a direction control signal for the television camera 2 to the two direction changing means of the television camera 2 shown in FIG. 3 through the switch 510. This control signal is operated based on the direction changing means 2Atl of the television camera 2 - Table 1 described above. As a result, the direction of the TV camera 2 is set to a predetermined route, and the moving body 1 is steered in accordance with the direction of movement of the TV camera 2, so that it can be guided to the destination. 1. This control signal is compared with the predicted position value of the moving object from the moving object position predicting means 3, and calculates the distance traveled due to various errors that occur when driving in the actual environment, such as slips and uneven road surfaces. error, imbalance between left and right motors,
In addition to correcting steering errors caused by gyroscope drift, position and direction errors at the start of travel are corrected.

According to the above-described embodiment, an observed image observed by a television camera is simultaneously extracted into a plurality of characteristic patterns by an image processing device, and the most well-recognized pattern among them is used as a guidance index, so that accuracy is improved. Good guidance is possible. Further, information representing the deviation between the observed characteristic pattern and the predicted stoppage pattern is detected in real time by the visual control device 7. In the above-described embodiment, the collation means 6 is used to improve guidance. Although the consistency between the observed feature pattern and the predicted feature pattern was determined by the following, the present invention can be carried out without using this matching means 6. Also, TV camera 2
A wide-angle image input means can also be used.

〔Effect of the invention〕

As described above in detail, according to the present invention, recognition of the environment and action can be performed simultaneously, so that a mobile object can move and work at high speed in a wide range of general environments.

[BRIEF DESCRIPTION OF THE DRAWINGS] FIG. 1 is a block diagram explaining the present invention in detail, FIG. 2 is a diagram explaining the operation of the principle of the present invention, and FIG. 3 is a diagram showing an embodiment of the guiding device of the present invention. FIG. 4 is a diagram showing the configuration of an example of an image processing device constituting the apparatus of the present invention, and FIG.
6 is a diagram showing the operation of the device, FIG. 6 is a diagram showing the configuration of an example of the potential field calculating means constituting the device of the present invention, and FIG.
The figure is a diagram showing the configuration of an example of the acting force calculation means included in the device fK of the present invention. DESCRIPTION OF SYMBOLS 1... Moving body, IA... Its running part, IB... Gyroscope, IC... Detector, 2... Television camera, 2 people... Direction changing means for television camera 2, 3・
...Moving body position detection means, 4...Storage means, 5...
Image processing device, 6... Verification means, 7... Visual control device, 8... Guidance control means for the moving object 1. Violet 1 mouth 2 figure'''￥3 figure

Claims (1)

[Claims] 1. A device for guiding a moving object along a predetermined route, which includes: visual means for observing the actual environment around the moving object; means for detecting the route movement of the moving object; means for storing predicted patterns related to position information; image processing means for extracting observed feature patterns from real images from the visual means;
visual control means for controlling the direction of the visual means by comparing the predicted pattern obtained from the storage device by the route movement detecting means and the observed feature pattern from the image processing means, and a set value and a control value for the direction of the visual means. 1. A guidance device comprising: a guidance control means for comparing the guidance signal with the moving object and outputting the guidance signal to the moving object. 2. The guidance device according to claim 1, wherein the visual control means determines the image cutting position by comparing the predicted pattern and the observed characteristic pattern.