JP2012117833A - Image recognition device, mounting type robot, and image recognition program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To automatically detect the height of a camera for image recognition.SOLUTION: An imaging camera 5 for an image recognition device, which is turned to a forward and obliquely downward side is installed at a camera position 22. At a light source position 21 above the camera position 22, a light source device 4 for projecting a projection image 26 to the inside of a screen frame 31 of the imaging camera 5 is installed. An image recognition device mounted on a mounting type robot measures a projection image offset x2 which is a distance between a lower end of the screen frame 31 and a lower end of the projection image 26 by image data, and obtains camera height y2 by a corresponding data map in which the projection image offset x2 and the camera height y2 are made to correspond, or calculation. The camera height y2 obtained in such a way, is memorized in a memory device as a parameter, and used for recognizing level difference or the like in the image recognition device.

Description

  The present invention relates to an image recognition apparatus, a wearable robot, and an image recognition program, and for example, relates to an apparatus for setting a reference height of a camera.

In recent years, a subject is photographed with a camera, image analysis is performed, and machine control is performed using the result.
An example of such a technique is “Robot” of Patent Document 1.
In this technique, an annular pattern light is projected around a robot to pick up an image, and obstacles (steps, etc.) are detected based on the picked up image.
By the way, when the height of surrounding obstacles is detected by image analysis, the height of the camera is required as a parameter.
Therefore, conventionally, the height of the camera is measured in advance and stored in a storage medium.

However, for applications where the camera is used in a fixed state, the camera height may be measured and stored in advance.For example, when the camera position is changed or a new camera is attached. However, each time, it is necessary to manually measure the position of the camera and input it as a parameter, and there is a problem that the operation is not efficient.
For example, when installing a camera on a wearable robot and detecting a step, it is efficient to manually measure and input the camera height each time the wearer changes because there is a difference in height between wearers. There wasn't.

JP 2005-324297 A

  An object of the present invention is to automatically detect the height of a camera for image recognition.

(1) According to the first aspect of the present invention, a projection unit that irradiates a predetermined pattern of light from a light source in a predetermined irradiation direction and projects an image of the light onto a reference plane, and a camera directed to the predetermined imaging direction The height of the camera with respect to the reference plane is determined using an imaging means for capturing the projected image, the positional relationship between the light source and the camera, and the position of the image in the imaging region captured by the camera. There is provided an image recognition apparatus characterized by comprising a camera height acquisition means for acquiring.
(2) In the invention described in claim 2, the camera height acquisition means acquires the height of the camera using a distance from an end of the imaging region to the image. The image recognition apparatus described in 1. is provided.
(3) In the invention according to claim 3, when the end portion is not in the imaging region, at least one of the irradiation direction and the imaging direction is changed so that the end portion is positioned in the imaging region. The image recognition apparatus according to claim 1, further comprising a direction changing unit.
(4) In invention of Claim 4, the image recognition apparatus of Claim 1, Claim 2, or Claim 3, and the walk control means which performs walk control using the recognition result by the said image recognition apparatus, A wearable robot characterized by comprising:
(5) In the invention according to claim 5, by a projection function that irradiates a predetermined pattern of light from a light source in a predetermined irradiation direction and projects an image of the light on a reference plane, and a camera directed in a predetermined imaging direction The height of the camera with respect to the reference plane is determined using an imaging function for capturing the projected image, the positional relationship between the light source and the camera, and the position of the image in the imaging region captured by the camera. An image recognition program for realizing a camera height acquisition function to be acquired by a computer is provided.

  According to the present invention, the height of the image recognition camera can be automatically detected by using the position of the light image of the predetermined pattern in the imaging region.

It is a figure for demonstrating a mounting | wearing type robot. It is a figure for demonstrating the method to detect the height of an imaging camera. It is a flowchart for demonstrating the procedure which calculates | requires camera height y2. It is a figure for demonstrating the method to detect the height of an imaging camera in the modification 1. FIG. 10 is a flowchart for explaining a procedure for obtaining a camera height y2 in Modification 1; It is a figure for demonstrating the modifications 2 and 3. FIG.

(1) Outline of Embodiment As shown in FIG. 2A, an imaging camera 5 for an image recognition apparatus is installed at a camera position 22 obliquely downward and forward, and a light source position 21 above it. The light source device 4 for projecting the projection image 26 is installed inside the screen frame 31 of the imaging camera 5.
The image recognition apparatus mounted on the wearable robot 1 (FIG. 1A) measures the projection image offset x2 that is the distance between the lower end of the screen frame 31 and the lower end of the projection image 26 using image data, and the projection image offset. The camera height y2 is acquired by a correspondence data map in which x2 and the camera height y2 are associated or by calculation.
The camera height y2 obtained in this way is stored in the storage device as a parameter, and is used for recognizing a step or the like by the image recognition device.

(2) Details of Embodiment In this embodiment, as an example, a case where an image recognition apparatus is installed in a wearable robot will be described.
FIG. 1A is a diagram showing a mounting state of the mounting robot 1.
The wearable robot 1 is worn on the user's lower back and lower limbs to assist (assist) the user's walking. In addition, for example, it may be attached to the upper body and the lower body to assist the movement of the whole body.

The wearable robot 1 includes a waist mounting part 7, a walking assist part 2, a connecting part 8, a three-axis sensor 3, a three-axis actuator 6, an imaging camera 5, a light source device 4, and an imaging that holds the imaging camera 5 and the light source device 4. It consists of unit 9 and the like.
The waist mounting portion 7 is a fixing device that fixes the wearable robot 1 to the user's waist. The waist mounting part 7 moves together with the user's waist.
Moreover, the waist | hip | lumbar part mounting | wearing part 7 is provided with the walking actuator which is not shown in figure, and drives the connection part 8 to the front-back direction etc. according to a user's walking motion.

The connecting part 8 connects the waist mounting part 7 and the walking assist part 2.
The walking assist unit 2 is attached to a user's lower limb and is driven back and forth by a walking actuator to support the user's walking motion.
The walking support by the waist mounting portion 7, the connecting portion 8, and the walking assist unit 2 is an example, and various forms such as walking support by a multi-joint drive mechanism are possible.

The triaxial sensor 3 is installed in the waist mounting portion 7 and detects the posture of the waist mounting portion 7 and the like. The triaxial sensor 3 includes, for example, a triaxial angular velocity detection function and a triaxial angular acceleration detection function using a three-dimensional gyro, and the x-axis direction (forward direction), the y-direction (vertical direction), and the z-direction (body side direction). , Rotation angle, angular velocity, angular acceleration, and the like can be detected.
The angle around the x axis is the roll angle, the angle around the y axis is the yaw angle, and the angle around the z axis is the pitch angle.

The triaxial actuator 6 is configured by, for example, a spherical motor, and changes the roll angle, yaw angle, and pitch angle of the imaging unit 9 in which the imaging camera 5 and the light source device 4 are installed.
The light source device 4 and the imaging camera 5 are fixed to the imaging unit 9, and when the triaxial actuator 6 is driven, the irradiation direction of the light source device 4 (the direction of the optical axis of the light source device 4) and the imaging direction of the imaging camera 5. (The direction of the optical axis of the imaging camera 5) changes the roll angle, the yaw angle, and the pitch angle while maintaining the relative angle.

  As will be described later, when the height of the imaging camera 5 is detected, the imaging unit 9 needs to be directed to a walking reference plane (a surface on which the user walks, such as the ground or floor) at a predetermined angle. When the type robot 1 is mounted, the imaging unit 9 tilts depending on the mounted state, and this is corrected by the triaxial actuator 6.

  The light source device 4 irradiates light such as laser, infrared light, and visible light in a predetermined shape pattern, for example. In the present embodiment, the light source device 4 irradiates light with a shape pattern that is circular with respect to a surface perpendicular to the irradiation direction, but various shapes such as a rectangular shape, a cross, a dot, and a cross are possible. is there.

The imaging camera 5 is configured by an infrared light camera, a visible light camera, or the like that includes an optical system for forming an image of a subject and a CCD (Charge-Coupled Device) that converts the imaged subject to an electrical signal. The light source device 4 captures (shoots) a projection image irradiated on the walking reference plane.
The projected image by the light emitted by the light source device 4 in a predetermined shape pattern has a shape deformed from a circle due to the angle formed by the irradiation direction and the walking reference plane, and obstacles (such as steps) present on the walking reference plane. By analyzing this shape, the state of the walking reference plane and obstacles can be detected.

In the imaging unit 9, the light source device 4 and the imaging camera 5 are fixed, and the roll angle, the yaw angle, and the pitch angle with respect to the waist mounting portion 7 are changed by the triaxial actuator 6.
That is, by driving the imaging unit 9 with the triaxial actuator 6, the roll angle of the light source device 4 and the triaxial actuator 6 with respect to the waist mounting portion 7 while the relative positions of the light source device 4 and the imaging camera 5 are fixed, The yaw angle and pitch angle can be changed.

FIG. 1B is a diagram for explaining the mounting robot system 15 installed in the mounting robot 1.
The wearing robot system 15 is an electronic control system that controls the wearing robot 1 so as to exhibit a walking support function.
In this figure, the part used for detecting the height of the imaging camera 5 is shown, and the part used for performing walking support is omitted.

  The ECU (Electronic Control Unit) 16 is an electronic control unit including a CPU (Central Processing Unit), a ROM (Read Only Memory), a RAM (Random Access Memory), a storage device, various interfaces, and the like (not shown). Each part of 1 is electronically controlled.

The CPU executes various computer programs stored in the storage medium, calculates the height of the imaging camera 5, recognizes the steps of the stairs and the escalator, and drives the walking actuator to provide walking assistance.
The CPU is connected to the light source device 4, the imaging camera 5, the 3-axis actuator 6, and the 3-axis sensor 3 through an interface, and turns on / off irradiation from the light source device 4 and acquires imaging data from the imaging camera 5. Or the triaxial actuator 6 is driven, or the detection value is acquired from the triaxial sensor 3.

The ROM is a read-only memory and stores basic programs and parameters used by the CPU.
The RAM is a readable / writable memory, and provides a working memory when the CPU performs arithmetic processing and the like. In the present embodiment, a working memory for storing the imaging data of the imaging camera 5 and the detection value of the three-axis sensor 3 and for calculating the height of the imaging camera 5 is provided.

  The storage device includes a large-capacity storage medium configured by, for example, a hard disk or an EEPROM (Electrically Erasable Programmable ROM), and a program for detecting the height of the imaging camera 5 or a projection image of the light source device 4 Various programs such as programs for recognizing steps such as stairs and escalators, programs for assisting walking, parameters such as the height of the imaging camera 5 used for image recognition for step recognition, etc. I remember it.

FIG. 2A is a diagram for explaining a method of detecting the height of the imaging camera 5.
In this figure, with the user standing upright, the position of the user's feet is the origin, the walking reference plane is the x axis (the user's forward direction is positive), and the user's overhead direction is the y axis.
The reference position of the light source device 4 is set as the light source position 21, the reference position of the imaging camera 5 is set as the camera position 22, and the height of the camera position 22 from the walking reference plane is set as y2.
The light source position 21 is vertically above the camera position 22, and the entire projection image 26 (projected object) irradiated from the light source position 21 onto the walking reference plane is present in the screen frame 31 of the imaging camera 5. Yes.
It is assumed that the optical axis 28 of the light source device 4 and the optical axis 29 of the imaging camera 5 are both in the xy plane.

The distance from the camera position 22 to the lower end of the irradiation light is set as an irradiation offset x1, and the distance from the lower end of the screen frame 31 of the camera to the lower end of the projection image 26 is set as a projection image offset x2. The side closer to the user is the lower end, the side far from the user is the upper end, and so on.
The distance from the intersection 25 between the extension line at the lower end of the irradiated light and the extension line at the lower end of the screen frame 31 to the camera position 22 is defined as an intersection position y1.
In other words, the intersection point 25 is a point where the lower end of the irradiation light coincides with the lower end of the screen frame 31 when there is no walking reference plane.

FIG. 2B is a diagram illustrating an example of the projection image 26 captured by the imaging camera 5.
The light source device 4 emits light having a circular cross section, but is irradiated obliquely forward with respect to the walking reference plane, so that the projection image 26 has an elliptical shape with the forward direction of the user as the major axis.
The screen frame 31 of the imaging camera 5 has a rectangular shape in which the user's forward direction is the short side and the user's body side direction is the long side, and the distance from the lower end to the projection image 26 is the projection image offset x2.

  As described above, when the light source device 4, the imaging camera 5, and the walking reference plane are arranged, a proportional expression according to the following expression (1) is established.

  y1: (y1-y2) = x1: x2 (1)

  From this, equation (2) holds.

  y2 = y1 (x1-x2) / x1 (2)

Here, the irradiation offset x1 is known in advance as a fixed value, and x2 can be obtained from the number of captured image pixels. Further, since the irradiation direction of the light source device 4 and the imaging direction of the imaging camera 5 are fixed, x2 and y1 have a one-to-one correspondence, and y1 is a function of x2. Therefore, y2 is a function of x2.
Therefore, in this embodiment, a correspondence data map of y2 and x2 is created in advance and stored in the storage device, and y2 is obtained from x2 using this map.

FIG. 3 is a flowchart for explaining a procedure for the wearable robot 1 to obtain the camera height y2.
The following processing is performed by the CPU of the ECU 16 according to a predetermined program.
First, the user mounts the wearable robot 1 and stands upright, and operates the wearable robot 1 to start measuring the camera height y2.

Then, the CPU acquires the yaw angle, pitch angle, and roll angle of the waist mounting portion 7 from the triaxial sensor 3 and stores them in the RAM (step 5).
Next, the CPU calculates the inclination of the waist mounting portion 7 with respect to the walking reference plane from these angles stored in the RAM, and the directions of the light source device 4 and the imaging camera 5 with respect to the walking reference plane are as shown in FIG. The imaging unit 9 is driven by the triaxial actuator 6 so as to be in the prescribed direction, and the irradiation direction and the imaging direction are corrected (step 10).

Next, the CPU turns on the light source device 4 to emit light of a predetermined shape pattern (step 15).
Next, the CPU drives the imaging camera 5 to capture the projected image 26 projected onto the walking reference plane, and stores the image data in the RAM (step 20).
Next, the CPU reads the image data from the RAM, obtains the projection image offset x2 by counting the number of pixels from the lower end of the screen frame 31 to the lower end of the projection image 26, and stores it in the RAM (step 25).

Next, the CPU searches for the projection image offset x2 stored in the RAM using the corresponding data map stored in the storage device, and obtains the camera height y2 corresponding to the searched x2 (step 30).
Next, the CPU stores the camera height y2 as a parameter in the storage device.

In the above, the camera height y2 is obtained from the corresponding data map, but it can also be obtained by calculation as shown in Modification 1 described later.
In the present embodiment, the camera height y2 is measured while the user is standing upright. However, by continuously performing the camera height y2 in real time, the user can follow the vertical movement of the user's waist by walking. Thus, the camera height y2 can be corrected.

According to the embodiment described above, the following effects can be obtained.
(1) The installation height of the imaging camera 5 for image recognition can be automatically detected.
(2) When a plurality of users wear the wearable robot 1, the height of the imaging camera 5 can be automatically corrected according to the body shape of each user.
(3) Even if the attachment position of the waist attachment portion 7 differs depending on the user, it is not necessary to manually calibrate each time it is attached.
(4) In a system that is worn by the user and detects the step by imaging the shape of the projected image 26 with the imaging camera 5, the posture changes each time by measuring the positional correlation between the screen frame 31 and the projected image 26. In addition, the light source device 4 and the imaging camera 5 can be calibrated using a posture sensor such as a three-dimensional gyroscope or a three-axis motor.

(Modification 1)
In the embodiment described above, since the angles of the light source device 4 and the imaging camera 5 are fixed, for example, when a tall user wears the wearable robot 1, the intersection 25 (FIG. 2 (a)). May be above the walking reference plane and the projection image offset x2 may not be formed (that is, the lower end of the projection image 26 protrudes from the lower end of the screen frame 31).
For this reason, in the first modification, when the imaging direction of the imaging camera 5 is variable and the lower end of the projection image 26 protrudes from the screen frame 31, the imaging direction of the imaging camera 5 is corrected and the projection image 26 becomes the screen frame 31. Adjust so that it is located inside.

FIG. 4A is a diagram for explaining a method of detecting the height of the imaging camera 5 in the first modification, and FIG. 4B shows a projection image 26 captured by the imaging camera 5. FIG.
As shown in FIGS. 4A and 4B, the distance between the camera position 22 and the light source position 21 is the distance between the light source cameras y3 (y3 is known), and the angle between the lower end of the irradiated light and the y axis is the light source. An angle α2 and an angle formed by a line forming the lower end of the screen frame 31 and the y axis are α1. Further, θ1 = 90 ° −α1 and θ2 = 90 ° −α2.
With this setting, the following equation holds.

y1 = tan θ1 · y3 / (tan θ2−tan θ1) (3)
x1 = y3 / tan θ2 (4)

  x2 can be obtained from the image data, and since Expression (2) holds, y2 can be expressed by the following Expression (5) or Expression (6), and if θ1 and θ2 are known, The camera height y2 can be calculated by the equation (5) or the equation (6).

y2 = tan θ1 · y3 / (tan θ2-tan θ1) × (y3 / tan θ2-x2) / (y3 / tan θ2)
= Y1 * (x1-x2) (5)
= Tan θ1 · (y3−x2 tan θ2) / (tan θ2−tan θ1) (6)

FIG. 5 is a flowchart for explaining a procedure for the wearable robot 1 to obtain the camera height y2 in this modification.
In Modification 1, it is assumed that the imaging unit 9 includes an actuator that adjusts the pitch angle of the imaging camera 5 in the imaging direction.

First, the user mounts the wearable robot 1 and stands upright, and operates the wearable robot 1 to start measuring the camera height y2.
Then, the CPU acquires the yaw angle, pitch angle, and roll angle of the waist mounting portion 7 from the triaxial sensor 3 and stores them in the RAM (step 40).
Next, the CPU corrects the irradiation direction by driving the imaging unit 9 with the triaxial actuator 6 so that the direction of the light source device 4 is the direction defined in FIG. (Step 45).

Next, the CPU turns on the light source device 4 to irradiate the walking reference plane with light having a predetermined shape pattern (step 50).
Next, the CPU drives the imaging camera 5 to capture the projected image 26 and stores the image data in the RAM (step 55).

Next, the CPU analyzes the image data stored in the RAM, and determines whether or not the upper end and the lower end of the projection image 26 exist inside the screen frame 31, that is, whether the upper end to the lower end are in the screen frame 31. It is determined whether or not (step 60).
Here, when the projection image offset x2 is equal to or larger than a predetermined value larger than 0, it may be determined that the lower end exists inside the screen frame 31.

  In order to calculate the camera height y2, it is only necessary that the lower end of the projection image 26 is in the screen frame 31, and the upper end is not required to be in the screen frame 31, but here, a step is imaged later. In preparation for the recognition, the condition is that the upper and lower ends are in the screen frame 31.

When the upper end and the lower end of the projection image 26 are not in the screen frame 31 (step 60; N), the CPU changes the pitch angle of the imaging camera 5 (step 65) and returns to step 55.
More specifically, the CPU changes the pitch angle of the imaging camera 5 by a predetermined angle in the direction between the upper end and the lower end of the projection image 26 that does not exist inside the screen frame 31.
Note that the roll angle of the imaging camera 5 is corrected by driving the imaging unit 9 with the triaxial actuator 6.

On the other hand, when the upper end and the lower end of the projected image 26 are in the screen frame 31 (step 60; Y), the CPU stores the yaw angle, pitch angle, and roll angle stored in the RAM, and further the triaxial actuator 6 The light source angle α2 is calculated from the drive angle and stored in the RAM (step 70).
Since the arrangement of the waist mounting portion 7 with respect to the walking reference plane is known from the angle of the triaxial sensor 3 and the arrangement of the light source device 4 with respect to the waist mounting portion 7 is known from the driving angle of the triaxial sensor 3, α2 is calculated from these values. be able to.

Next, the CPU obtains the drive angle of the pitch angle of the imaging camera 5, calculates the camera angle α1, and stores it in the RAM (step 75).
The arrangement of the waist mounting portion 7 with respect to the walking reference plane is known from the angle of the triaxial sensor 3, and the arrangement of the imaging unit 9 with respect to the waist mounting portion 7 is known from the driving angle of the triaxial sensor 3, and the imaging unit is determined from the driving angle of the imaging camera 5. Since the imaging direction with respect to 9 is known, the camera angle α1 can be calculated from these values.

Next, the CPU reads the camera angle α1 and the light source angle α2 stored in the RAM, subtracts them from 90 °, calculates θ1 and θ2, and stores them in the RAM (step 80).
Next, the CPU reads the light source camera-to-camera distance y3 stored in advance as a parameter from the storage device, reads θ1 and θ2 from the RAM, and substitutes these into equation (3) to calculate the intersection position y1. And stored in the RAM (step 85).

Next, the CPU reads the distance y3 between the light source cameras from the storage device, reads θ2 from the RAM, substitutes these into equation (4), calculates the irradiation offset x1, and stores it in the RAM (step 90).
Next, the CPU reads the image data from the RAM and counts the number of pixels from the lower end of the screen frame 31 to the lower end of the projected image 26 to obtain the projected image offset x2 (step 95).

Next, the CPU reads the intersection position y1, the irradiation offset x1, and the projection image offset x2 from the RAM, substitutes them into the equation (5), calculates the camera height y2, and stores it in the RAM (step 100). Or you may substitute (theta) 1, (theta) 2, y3, and x2 to Formula (6).
Then, the CPU reads the camera height y2 from the RAM and stores it as a parameter in the storage device.

In the above description, the pitch angle of the imaging camera 5 can be adjusted with respect to the imaging unit 9. However, the imaging camera 5 may be fixed to the imaging unit 9 and the pitch angle of the light source device 4 may be changed.
Or it can also comprise so that the pitch angle of both the imaging camera 5 and the light source device 4 may be changed.

According to the first modification described above, the following effects can be obtained.
(1) When the projection image 26 is not inside the screen frame 31, it can be automatically corrected so that the projection image 26 is located inside the screen frame 31.
(2) A wide range of users can be supported, from users with low waist to users with high waist.
(3) When the camera height y2 is automatically corrected following the user's waist movement due to walking, the position of the projection image 26 is set so that the projection image 26 exists inside the screen frame 31 within the range of the vertical movement. Can be set.

(Modification 2)
6A and 6B are diagrams for explaining the second modification.
In the embodiment and the first modification described above, the distance between the lower end of the screen frame 31 and the lower end of the projection image 26 is the projection image offset x2. In the second modification, the position of the camera position 22 is the position of the light source position 21. The distance between the upper end of the screen frame 31 and the upper end of the projection image 26 is defined as the projection image offset x2.
The light source height y2 can be calculated also from the projection image offset x2 set in this way, and the camera height can be calculated by adding the light source inter-camera distance y3 to y2.

(Modification 3)
Since the major axis of the projection image 26 changes depending on the camera height, the lower end position 34 on the lower end side of the projection image 26 in the embodiment, the first modification, and the second modification, as shown in FIG. It is also possible to calculate the camera height from the upper end position 33 on the upper end side.

As described above, in the embodiment and the first to third modifications, the case where the wearable robot 1 is equipped with the camera height detection function has been described. However, this is not limited to the wearable robot 1 and is not limited to the image recognition apparatus. It can be widely applied when installing.
For example, when installing an image recognition device in a vehicle, it takes time and labor to accurately attach the imaging camera 5 to a predetermined position in a predetermined posture. In particular, when recognizing a distance to an obstacle around a vehicle, a parking space, or the like, a reference height h where the image recognition device is installed is determined in advance, and the distance from the reference plane (such as a road) to the image recognition device is determined. It is necessary to attach so that the height is exactly the reference height h. It takes time and labor to attach the image recognition device to the position of the accurate reference height h.
Therefore, by using the image recognition apparatus of each embodiment described above, the height y2 of the imaging camera 5 is accurately measured from the correlation between the projection image 26 and the screen frame 31, or the angle of the imaging camera 5 is adjusted. The measured accurate height y2 can be converted to the reference height h.
As described above, according to the image recognition apparatus of the present embodiment, the accurate camera height can be automatically measured and converted into the reference height h. Therefore, the image recognition apparatus (the imaging camera 5) can be used at an approximate position and orientation. Etc.) is sufficient, and the mounting operation can be made more efficient.

The following configurations can be obtained by the embodiment and the first to third modifications described above.
The image recognition apparatus included in the wearable robot 1 projects a projection image 26 onto the walking reference plane by irradiating light of a predetermined shape pattern with the light source device 4, and captures this with the imaging camera 5. Projection means for irradiating light in a predetermined irradiation direction from a light source and projecting an image of the light onto a reference plane; and imaging means for capturing the projected image by a camera directed to a predetermined imaging direction Yes.
In addition, the image recognition apparatus provided in the wearable robot 1 measures the camera height y2 using the positional relationship between the light source position 21 and the camera position 22 and the positional relationship of the projection image 26 in the screen frame 31 (imaging region). Camera height acquisition means is provided for acquiring the height of the camera with respect to the reference plane using the positional relationship between the light source and the camera and the position of the image in the imaging area captured by the camera (first). 1 configuration).

  The image recognition apparatus provided in the wearable robot 1 measures the camera height y2 using, for example, the distance between the lower end of the screen frame 31 and the lower end of the projected image 26. Therefore, the camera height acquisition means The height of the camera is acquired using the distance from the end to the image (second configuration).

  The image recognition device provided in the wearable robot 1 of the first modification changes the direction of at least one of the imaging camera 5 or the light source device 4 and calibrates so that the projected image 26 is positioned inside the screen frame 31. When the end portion is not in the imaging region, there is provided direction changing means for changing at least one of the irradiation direction and the imaging direction so that the end portion is located in the imaging region (third) Configuration).

  The wearable robot 1 uses the camera height y2 obtained in the first to third configurations to detect an obstacle such as a step and control the user's walking. A walking control means for performing walking control using the result is provided.

  In addition, the storage device of the ECU 16 includes a projection function that irradiates a predetermined pattern of light from a light source in a predetermined irradiation direction and projects an image of the light on a reference plane, and a camera directed in a predetermined imaging direction. The height of the camera with respect to the reference plane is acquired using an imaging function for capturing a projected image, the positional relationship between the light source and the camera, and the position of the image in the imaging region captured by the camera. An image recognition program for realizing a camera height acquisition function by a computer is stored.

DESCRIPTION OF SYMBOLS 1 Wearable robot 2 Walking assist part 3 3-axis sensor 4 Light source device 5 Imaging camera 6 3-axis actuator 7 Lumbar mounting part 8 Connection part 9 Imaging unit 15 Wearing robot system 16 ECU
21 Light source position 22 Camera position 25 Intersection point 26 Projected image 28, 29 Optical axis 31 Screen frame 33 Upper end position 34 Lower end position

Claims (5)

Projection means for irradiating a predetermined pattern of light from a light source in a predetermined irradiation direction and projecting an image of the light on a reference plane;
An imaging means for capturing the projected image by a camera directed to a predetermined imaging direction;
Camera height acquisition means for acquiring the height of the camera with respect to the reference plane using the positional relationship between the light source and the camera and the position of the image in the imaging area captured by the camera;
An image recognition apparatus comprising:   The image recognition apparatus according to claim 1, wherein the camera height acquisition unit acquires the height of the camera using a distance from an end of the imaging region to the image.   When the end portion is not in the imaging region, there is provided direction changing means for changing at least one of the irradiation direction and the imaging direction so that the end portion is located in the imaging region. The image recognition apparatus according to claim 1. An image recognition device according to claim 1, claim 2, or claim 3;
Walking control means for performing walking control using the recognition result by the image recognition device;
A wearable robot characterized by comprising: A projection function for irradiating a predetermined pattern of light from a light source in a predetermined irradiation direction and projecting an image of the light on a reference plane;
An imaging function for capturing the projected image by a camera directed to a predetermined imaging direction;
A camera height acquisition function for acquiring the height of the camera with respect to the reference plane using the positional relationship between the light source and the camera, and the position of the image in the imaging area captured by the camera;
An image recognition program that realizes the above on a computer.

Images