JP6822816B2 - Construction system and construction method using mobile robots in pipes - Google Patents

Description

The present disclosure relates to a construction system in a pipe and a construction method in the pipe for performing construction on the inner surface of the pipe by a mobile robot.

Construction work may be required on the inner surface of pipes for maintenance of pipes that make up structures such as plants. As an example, when foreign matter such as rust or dust accumulates as scale on the inner surface of the piping that constitutes a plant that has been operating for a long period of time, this scale peels off during plant operation and reaches the equipment in the subsequent stage, and the equipment concerned. May cause malfunction or damage. Therefore, maintenance work inside the pipe, such as pre-removal work of the scale, is required, and such work is a mobile robot that performs construction inside the pipe while moving inside the pipe along the longitudinal direction of the pipe. It may be done using a construction system equipped with.

The following Patent Documents 1 to 3 include a laser irradiation unit for removing scale and a coating film accumulated on the inner surface of a pipe by irradiating a laser beam as a maintenance work in the pipe as described above, and move in the pipe. While disclosing a mobile robot that performs construction inside the pipeline. In particular, in Patent Document 3, the reflection mirror is moved to control the laser irradiation direction, and the surface condition in the pipe including the thickness and degree of rust accumulated in the pipe is detected by a sensor according to the surface condition in the pipe. A technique for controlling laser irradiation conditions is disclosed.

JP-A-2002-224875 Utility Model Registration No. 3022303 International Publication No. 2013/133415

However, in the inventions described in Patent Documents 1 to 3, when the mobile robot moving in the pipe removes the scale and the coating film accumulated on the inner surface of the pipe by irradiating the laser beam from the laser irradiation unit, it is as follows. Problems arise. That is, the inner surface of the pipe becomes hot due to the laser irradiation from the laser irradiation unit of the mobile robot, and the equipment and sensors included in the mobile robot moving in the pipe may cause malfunction or failure due to heat. For example, when the mobile robot is equipped with a sensor that detects the surface state of the inner surface of the pipe and a sensor that recognizes the space inside the pipe, the high heat generated on the inner surface of the pipe due to laser irradiation deteriorates the measurement accuracy of the sensor. There is a risk that a failure will occur and the construction work will not be performed correctly.

In view of the above problems, in some embodiments according to the present invention, even if high heat is generated in the pipe by irradiating the inner surface of the pipe with laser while the mobile robot moves in the pipe, the mobile robot is provided. The purpose is to obtain a construction system in the piping that is unlikely to cause deterioration of detection accuracy or failure of equipment and sensors due to the high heat.

(1) The in-pipe construction system using a mobile robot according to some embodiments of the present invention is
A surface condition sensor for detecting the state of the inner surface of the pipe by scanning the inner surface of the pipe over a preset construction target range in the longitudinal direction of the pipe.
A first mobile robot configured to move the surface condition sensor in the longitudinal direction inside the pipe in a construction target range in the longitudinal direction of the pipe.
A laser irradiation unit for irradiating the inner surface of the pipe with a laser beam,
A second mobile robot configured to move the laser irradiation unit in the longitudinal direction inside the pipe in the construction target range.
A control unit configured to control the laser irradiation unit and the second mobile robot is provided.
The control unit
Based on the detection result of the surface condition sensor for the construction target range in which the scanning of the surface condition sensor is completed, three-dimensional data indicating the position coordinates of the inner surface of the pipe in the construction target range is acquired. , A map acquisition unit for acquiring a three-dimensional map in which the irradiation area of the laser beam is specified on the three-dimensional data.
Based on the three-dimensional map, command generation for sending a command to the laser irradiation unit and the second mobile robot so that the laser beam is selectively irradiated to the irradiation area within the construction target range. It is characterized by including a part and.

According to the configuration of (1) above, the laser light irradiation area based on the detection result of the surface condition sensor is specified for the construction target range in which the scanning of the surface condition sensor is completed by moving the first mobile robot in the longitudinal direction in the pipe. Since the laser irradiation unit irradiates the laser beam using the generated three-dimensional map, the scanning timing by the surface state sensor can be shifted sufficiently before the laser beam irradiation timing. That is, according to the configuration (1) above, the first mobile robot equipped with the surface state sensor passes through the area before the laser light irradiation area becomes hot due to the laser irradiation, and the surface state sensor Detection accuracy is less likely to decrease or failure due to high heat. Therefore, the irradiation area of the laser beam can be correctly specified from the inner surface state of the pipe, and the laser beam can be selectively irradiated to the correctly specified irradiation area.

(2) In one exemplary embodiment, in the configuration of (1) above, the map acquisition unit is based on the detection result of the surface condition sensor for the construction target range in which the scanning of the surface condition sensor is completed. ,
Three-dimensional data representing the spatial shape of the inner surface of the pipe in the construction target range is calculated.
It is characterized in that one or more data representing the positions and ranges of one or more irradiation areas to be irradiated with the laser beam are superimposed on the three-dimensional data to acquire a three-dimensional map. And.

In the configuration of (2) above, in the initial stage, three-dimensional data representing the spatial shape of the inner surface of the pipe in the construction target range is calculated. Further, in the configuration of the above (2), in the second and subsequent stages following the initial stage, a plurality of data representing the positions and ranges of the plurality of irradiation areas to be irradiated with the laser beam are sequentially superimposed on the three-dimensional data. And the 3D map is generated step by step. As a result, according to the configuration of (2) above, the 3D map is efficiently synthesized by adding the geometric information about each of the plurality of irradiation areas to the 3D data by an incremental method. can do.

(3) In one exemplary embodiment, in the configuration of (1) or (2), the first mobile robot and the second mobile robot form a single mobile robot, and the mobile robot is the mobile robot. It is configured to move on a movement path including an outward path extending the construction target range in the first direction along the longitudinal direction of the pipe and a return path extending the construction target range in the direction opposite to the first direction.
The map acquisition unit is configured to acquire the three-dimensional map in which the irradiation area is specified based on the detection result of the surface state sensor while the mobile robot moves on the outbound route.
After acquiring the three-dimensional map, the command generation unit starts the movement of the return path, and while the mobile robot moves the entire construction target range on the return path, the three-dimensional map. Based on the above, a command is sent to the laser irradiation unit and the mobile robot as the second mobile robot so that the laser beam is selectively irradiated to the irradiation area within the construction target range. It is characterized by being done.

In the configuration of (3) above, the state of the inner surface of the pipe for the construction target range is detected, and the laser irradiation for the inner surface of the pipe for the construction target range is performed by one reciprocating movement of a single mobile robot. Can be executed. Therefore, according to the configuration of the above (3), in addition to the same action and effect as the configuration of the above (1) or (2), the construction work for the entire construction target range in the pipe can be completed in a short time. Further, according to the configuration of (3) above, by forming the first mobile robot and the second mobile robot as a single mobile robot, the structure of the construction system in the pipe can be made compact.

(4) In one exemplary embodiment, in the configuration of (1) or (2), the first mobile robot and the second mobile robot are formed as separate mobile robots that move the entire construction target range. Being done
The map acquisition unit is configured to acquire the three-dimensional map in which the irradiation area is specified based on the detection result of the surface state sensor while the first mobile robot moves over the entire construction target range. ,
After acquiring the three-dimensional map, the command generation unit starts moving within the construction target range, and while the second mobile robot moves the entire construction target range, Based on the three-dimensional map, it is configured to send a command to the laser irradiation unit and the second mobile robot so that the laser beam is selectively irradiated to the irradiation area within the construction target range. It is characterized by that.

In the configuration (4) above, the first mobile robot that moves the surface condition sensor to complete the scanning of the surface condition sensor over the construction target range and the laser irradiation unit that performs laser irradiation over the construction target range are moved. The second mobile robot is formed as a separate mobile robot. Therefore, in the configuration of (4) above, if the work by the first mobile robot and the work by the second mobile robot are separated temporally and / or spatially, the laser irradiation of the laser irradiation unit in the pipe is performed. It is possible to prevent the surface condition sensor from being affected by the high heat generated by the surface condition sensor. As a result, according to the configuration of (4) above, it is possible to more effectively prevent deterioration of detection accuracy and failure due to high heat of the surface state sensor.

(5) In one exemplary embodiment, in the configurations (1) to (4) above, the laser irradiation unit rotates along the circumferential direction in the pipe so that the irradiation direction of the laser beam rotates. Includes a freely configured nozzle
The control unit is the second mobile robot along the longitudinal direction of the pipe at each position in the pipe along the longitudinal direction according to the angle range occupied by the irradiation area in the circumferential direction of the pipe. It is characterized in that it is configured to control the traveling speed of.

At the axial position (position in the longitudinal direction of the pipe) where the irradiation area to which the laser irradiation should be performed in the circumferential direction of the pipe occupies a wide angular range, the laser irradiation portion on the second mobile robot is along the circumferential direction in the pipe. It is necessary to focus on a wide range of laser irradiation sweeps. On the contrary, in the axial position where the angle range occupied by the irradiation area is narrow, the laser irradiation unit on the second mobile robot can finish sweeping the laser irradiation in a short time.

Therefore, in the configuration (5) above, at each position (axial position) along the longitudinal direction in the pipe, the longitudinal direction of the pipe is determined according to the angle range occupied by the irradiation area to which the laser irradiation should be performed in the circumferential direction of the pipe. The traveling speed of the second mobile robot is controlled according to the above. Therefore, it is possible to efficiently perform the laser irradiation work while reliably performing the laser irradiation in the irradiation area within the construction target range.

(6) In one exemplary embodiment, in the configurations (1) to (5), the control unit performs the laser light irradiation condition based on the surface condition in the pipe detected by the surface condition sensor. It is characterized in that it is configured to control.

According to the configuration of (6) above, the type of construction work (removal of oxide scale, cleaning, coating film removal, etc.) by irradiating the inner surface of the pipe with a laser, and the surface condition of the inner surface of the pipe (for example, It is possible to perform the construction work by laser irradiation over the construction target range while controlling the irradiation conditions of the laser beam according to the adhesion area of the oxide scale, the thickness of the oxide scale, the floating of the coating film, the thickness of the coating film, etc.). it can. As a result, according to the configuration of (6) above, the wall thickness of the pipe is scraped off more than necessary by laser irradiation by performing laser irradiation at the required output and for the required time at each construction site in the pipe. Can be prevented.

(7) In one exemplary embodiment, in the configurations (1) to (6), the first mobile robot is configured to be movable in the pipe within the construction target range and irradiates a scanning signal wave. By scanning the three-dimensional space in the pipe, the first point group data representing the distance and direction from the irradiation source of the scanning signal wave to the plurality of points constituting the space in the pipe is obtained. Further equipped with a first 3D scanner for acquiring as spatial measurement information representing the spatial shape in the pipe.
The control unit is characterized in that it is configured to acquire the three-dimensional map generated based on the first point cloud data.

According to the configuration of (7) above, it is possible to generate a three-dimensional map in the pipe based on the point cloud data obtained by measuring the distances to each point constituting the pipe internal space with a three-dimensional scanner. , It is possible to recognize the spatial shape in the pipe including the surface shape in the pipe with high accuracy.

(8) In one exemplary embodiment, in the configurations (1) to (7) above, a self-position acquisition unit for acquiring self-position information representing the self-position of the second mobile robot in the pipe is provided. Further prepare,
The control unit determines whether or not the self-position acquired by the self-position acquisition unit is included in the irradiation area on the three-dimensional map, and when the self-position is included in the irradiation area. It is characterized in that the laser irradiation unit is controlled so as to irradiate the laser beam.

According to the configuration of (8) above, by irradiating the laser beam when the self-position is included in the irradiation area, the laser irradiation can be performed only in the place where the laser irradiation is required in the pipe. It is possible. As a result, according to the configuration of (8) above, it is possible to prevent the wall thickness of the pipe from being unnecessarily scraped off by laser irradiation in a place where laser irradiation is not required in the pipe.

(9) In one exemplary embodiment, in the configuration of (8) above, the second mobile robot is configured to be movable in the pipe within the construction target range, and the pipe is irradiated with a scanning signal wave. By scanning the three-dimensional space inside, the second point group data representing the respective distances to the plurality of points constituting the space in the pipe is acquired as the space measurement information representing the space shape in the pipe. Also equipped with a second 3D scanner
The self-position information is calculated by collating the second point cloud data acquired by the second three-dimensional scanner with the pipe shape design information obtained at the time of designing the shape of the pipe. ..

According to the configuration of (9) above, by collating the point cloud data with the pipe shape design information, not only the current axial position of the 3D scanner (that is, the moving robot in the pipe) but also the 3D scanner (that is, the 3D scanner (that is, the moving robot in the pipe)) That is, the posture around the pipe axis (circumferential direction) of the mobile robot in the pipe can also be specified. Therefore, when the mobile robot moves in the pipe, even if the posture of the mobile robot in the pipe changes, it is possible to acquire self-position information in consideration of this, and the laser beam irradiation area can be made highly accurate. Can be identified.

(10) In one exemplary embodiment, in the configuration of (8) above, the three-dimensional map includes position information of a marker formed on the pipe in the construction target range.
A marker detector for detecting the marker and
The second mobile robot is configured to be movable in the pipe within the construction target range, and the space in the pipe is configured by irradiating a scanning signal wave and scanning the three-dimensional space in the pipe. Further provided with a second three-dimensional scanner for acquiring second point cloud data representing each distance to a plurality of points as spatial measurement information representing the spatial shape in the pipe.
The self-position information is calculated by collating the detection position of the marker superimposed on the second point cloud data with the position information of the marker included in the three-dimensional map. It is a feature.

According to the configuration of (10) above, the detection position of the marker superimposed on the second point cloud data acquired by the second 3D scanner and the position information of the marker included in the 3D map are collated. By doing so, the self-position can be accurately specified in relation to the three-dimensional map with reference to the position of the marker.

(11) In one exemplary embodiment, the configuration of (8) further includes a gyro sensor and an acceleration sensor configured to be movable in the pipe within the construction target range by the second mobile robot. ,
The self-position information is calculated by using at least the measurement results of at least one of the gyro sensor and the acceleration sensor.

According to the configuration of (11) above, a second mobile robot equipped with a gyro sensor and an acceleration sensor is provided by performing simple arithmetic processing on a small amount of data obtained as a measured value from the gyro sensor and the acceleration sensor. Can identify the current self-position of. As a result, according to the configuration of (11) above, the self-position can be accurately specified in a short time only by executing a simple arithmetic process.

(12) In one exemplary embodiment, in the configuration of (8) above, a plurality of sound sources that emit sound waves in response to a ringing command from the control unit are provided.
The mobile robot, which is at least one of the first mobile robot and the second mobile robot, further includes a sound wave sensor array.
The sound wave sensor array includes two or more sound wave observation means for observing the sound wave, and is configured to output a sound wave signal from the observed sound wave.
The self-position acquisition unit that has received the timing of the ringing command from the control unit
Based on the timing of the ringing command and the sound source signal from the sound wave sensor array, the relative position of the sound wave sensor array as seen from each position of the plurality of sound sources is estimated.
The self-position information is calculated by collating the installation positions of the plurality of sound sources with the relative positions of the sound wave sensor array.
It is characterized in that it is further configured.

According to the configuration of (12) above, even when the mobile robot is contaminated by a detached object from the inside of the pipe and the field of view of the optical device or the photographing device of the mobile robot is obstructed, the robot is provided in the pipe. It is possible to detect the self-position of the mobile robot by relying on the sound waves from a plurality of sound sources. Similarly, even if the reflected wave detection sensor that detects the reflected wave of the scanning signal wave on the mobile robot side becomes dirty and it becomes difficult to scan the inside of the pipe with the scanning signal wave in three dimensions, the sound wave from the sound source is relied on. It is possible to detect the self-position of the mobile robot. Further, in the configuration (12) above, the timing of sound emission from a plurality of sound sources, the type of sound, and the like are appropriately controlled by a ringing command from the control unit, so that the self-position of the mobile robot based on the sound waves from the sound sources can be determined. The detection can be performed with high accuracy.

(13) In one exemplary embodiment, in the configuration of (12) above, the self-position acquisition unit is
The sound source signal, which is a multi-channel sound source signal, is received from the sound wave sensor array, and the sound source separation process and the sound source are performed based on the timing of the ringing command, the positional relationship between the multi-channel sound source signal and the two or more sound wave observation means. An acoustic signal space processing unit that executes localization processing,
A sound source position estimation unit that estimates the relative positional relationship between each position of the sound source and the position of the sound wave sensor array based on the execution results of the sound source separation process and the sound source localization process.
It is characterized by having.

In the configuration (13) above, the sound source separation is performed by using the multi-channel sound source signal obtained by the plurality of sound wave observation means in the sound wave sensor array in combination with the timing of the ringing command and the positional relationship between the plurality of sound wave observation means. Execute processing and sound source localization processing. As a result, acoustic spatial information (spatial information of sound) that can express the arrival direction of sound waves and the distance to the sound source, as compared with the case where only the multi-channel sound source signals obtained by a plurality of sound wave observation means are used alone. Can be calculated accurately. Moreover, by using not only the multi-channel sound source signal but also the timing of the ringing command and the positional relationship between the plurality of sound wave observation means as auxiliary information, the sound source separation process and the sound source localization process can be executed easily and efficiently. It becomes possible to do.

(14) In one exemplary embodiment, in the configurations (1) to (13) above, the surface state sensor includes an imaging device that outputs an image of the inner surface of the pipe to the control unit.
The control unit
From the image, image recognition is used to generate inner surface state information indicating the inner surface state in the irradiation area indicated by the three-dimensional map.
It is characterized in that it is configured to generate a command to the laser irradiation unit based on the inner surface state information.

Conventionally, when the construction work on the base material surface is manually performed by irradiating the laser beam, the laser beam irradiation conditions for each construction site are determined based on the surface condition of the base metal surface visually visually recognized by the operator. By making adjustments, appropriate construction was performed according to the surface condition of the base metal. Therefore, in the configuration of (12) above, an image of the inner surface of the pipe taken by an imaging device is acquired as information that can be mechanically recognized and processed corresponding to the surface state of the surface of the base material visually visually recognized, and the image is taken. The inner surface condition information indicating the inner surface condition in the irradiation area in the pipe is generated from the image recognition. As a result, according to the above (12), by generating a command to the laser irradiation unit based on the inner surface state information, the laser irradiation condition according to the surface state of the base material, which has been manually performed conventionally. Adjustment operations can be automated for construction work in the piping.

(15) In one exemplary embodiment, in the configuration of (14) above, the control unit has the inner surface state information about the irradiation area shown by the three-dimensional map from an image obtained by photographing the inner surface of the pipe. When generating
A machine learning process that improves the accuracy of the image recognition is executed according to the teacher signal input from the outside.
It is characterized in that the image recognition is executed by using the learning result obtained by executing the machine learning process.

In the configuration of (15) above, the image recognition for generating the inner surface state information indicating the inner surface state in the irradiation area in the pipe from the photographed image of the irradiation area is the learning result obtained by executing the machine learning process. Is executed using. At that time, this machine learning process is executed so as to improve the accuracy of the image recognition according to the teacher signal input from the outside. Therefore, according to the configuration of (13) above, the control unit as the image recognition execution subject is trained by the machine learning process executed according to the teacher signal, whereby the above-mentioned inner surface state can be recognized with the required recognition accuracy. The accuracy of image recognition described above can be improved.

(16) In one exemplary embodiment, in the configuration of (15) above, the control unit has the inner surface state information about the irradiation area shown by the three-dimensional map from an image obtained by photographing the inner surface of the pipe. When generating
The machine learning process is executed by performing deep learning using a neural network in which an image obtained by photographing the inner surface of the pipe is used as an initial input and the error of the neuron output is corrected by the teacher signal.
It is characterized in that it is configured to execute the image recognition by using the learning result obtained by executing the deep learning.

In the configuration of (16) above, in order to improve the accuracy of the image recognition described above, deep learning using a neural network in which the error of the neuron output is corrected by the teacher signal is performed as a machine learning process executed according to the teacher signal. Execute. Therefore, according to the configuration of (16) above, in the machine learning process for improving the accuracy of image recognition described above, from an image obtained by photographing the inner surface of the pipe according to an engineering method such as spectroscopic spectrum analysis or color information analysis. There is no need to extract features. Therefore, according to the configuration of (16) above, when generating the inner surface state information from the image obtained by capturing the irradiation area, it is necessary to perform image recognition in which it is difficult to extract the feature amount from the image based on the engineering method. Even if this is not the case, image recognition that can recognize the inner surface state with the required recognition accuracy can be easily realized.

(17) The construction system in the pipe according to some embodiments of the present invention is
A mobile robot configured to move in the pipe along the longitudinal direction of the pipe, and
A surface condition sensor mounted on the mobile robot in front of the mobile robot in the traveling direction for detecting the state of the inner surface of the pipe,
A laser irradiation unit mounted on the mobile robot behind in the traveling direction and for irradiating the laser beam toward the inner surface of the pipe.
The laser irradiation unit and the control unit configured to control the mobile robot are provided.
The control unit
Each time the mobile robot is positioned at each of one or more positions arranged along the traveling direction of the mobile robot over the construction target range, the laser irradiation unit is controlled based on the detection result of the surface state sensor. It is configured in.

In the configuration (17) above, a surface condition sensor for detecting the condition of the inner surface of the pipe is mounted in front of the mobile robot in the traveling direction, and laser irradiation for irradiating the inner surface of the pipe with laser light is performed. The unit is mounted rearward in the direction of travel of the mobile robot. Therefore, in the configuration (17) above, while the moving robot within the construction target range is moving in the traveling direction, the step of detecting the surface state in the pipe by the surface state sensor and the laser irradiation unit are the inner surface of the pipe. Even if the steps of performing laser irradiation are simultaneously executed, the surface state sensor passes through the irradiation area to be laser-irradiated before the temperature rises due to the laser irradiation. As a result, in the configuration of the above (17), a single mobile robot is used to perform a step of detecting the surface state in the pipe and a step of irradiating the inner surface of the pipe with laser without exposing the surface state sensor to high heat. By executing at the same time, the construction work in the pipe can be completed in a short time.

(18) In one exemplary embodiment, in the configuration of (17) above, the mobile robot (a mobile robot that is at least one of the first mobile robot and the second mobile robot) is in the longitudinal direction of the pipe. It has a main body that is formed in a long shape that extends along the
A plurality of the surface state sensors are arranged at different positions in the circumferential direction of the main body portion, and include a plurality of imaging units each having a wide-angle lens.
Each of the wide-angle lenses possessed by the plurality of imaging units includes each of a plurality of angular ranges partially overlapping along the circumferential direction in the pipe in the field of view, thereby covering the entire inner surface of the pipe. It is configured to be able to take images over the direction,
The control unit
The plurality of imaging units generate inner surface state information indicating the inner surface state of the pipe from images taken inside the pipe.
It is characterized in that it is configured to generate a control command to the laser irradiation unit while identifying the position and range of the area in the pipe to which the laser irradiation should be performed based on the inner surface state information.

In the configuration (18) above, while the mobile robot is moving in the pipe, a plurality of wide-angle lenses arranged in front of the moving robot in the traveling direction are used to cover the entire inner surface of the pipe from the immediate front to the immediate rear. It is possible to take an image that covers the range of. Further, the laser irradiation can be performed from the laser irradiation unit arranged behind the moving direction of the mobile robot in parallel with the photographing in the pipe by these wide-angle lenses. Therefore, according to the configuration of (18) above, the position and range of the area in the pipe to be laser-irradiated can be determined without the need to generate a three-dimensional map in the pipe by scanning the inside of the pipe with the surface state sensor. It becomes possible to identify. Further, according to the configuration (18) above, it is possible to identify the position and range of the area in the pipe to be irradiated with the laser without having to detect the self-position of the mobile robot.

(19) In one exemplary embodiment, in the configuration of (18) above, the control unit is moving in the pipe in the traveling direction of the mobile robot in the construction target range.
Based on the detection result of the surface condition sensor, the irradiation area on the inner surface of the pipe is specified.
It is characterized in that the laser irradiation unit is controlled so that the laser beam is selectively irradiated to the irradiation area.

According to the configuration of (19) above, by controlling the laser irradiation unit so that the laser beam is selectively irradiated to the irradiation area, the laser irradiation is performed only in the place where the laser irradiation is required in the pipe. It is possible to. As a result, according to the configuration of the above (19), in addition to being able to finish the construction work in the pipe in a short time by using a single mobile robot by the same principle as the above (18), the inside of the pipe It is possible to prevent the wall thickness of the pipe from being unnecessarily scraped off by laser irradiation in a place where laser irradiation is not required.

(20) In one exemplary embodiment, in the configurations (1) to (19), the mobile robot (a mobile robot that is at least one of the first mobile robot and the second mobile robot) is
A main body formed in a long shape extending along the longitudinal direction of the pipe, and
A plurality of crawler-type traveling bodies arranged at different positions in the circumferential direction of the main body portion and configured to be able to travel on the inner surface of the pipe.
A plurality of telescopic mechanisms are arranged at different positions in the circumferential direction of the main body portion, and each has a telescopic mechanism configured to support the crawler type traveling body with respect to the main body portion so as to be expandable and contractible along the radial direction of the pipe. Arms and
It is characterized in that it is composed including.

According to the configuration of (20) above, a plurality of crawler-type traveling bodies installed at a plurality of positions along the circumferential direction of the main body of the mobile robot are provided along the radial direction by a telescopic mechanism provided by the plurality of arms. The following technical advantages can be obtained by being able to scale the robot.

For example, consider a case where a mobile robot comes into contact with the inner surface of a pipe or sits down at a bent portion of the pipe and gets stuck in the pipe, making the mobile robot inoperable. In this case, since a plurality of arms supporting each of the plurality of crawler-type traveling bodies can be contracted, the arms protruding from the main body and the crawler-type traveling body interfere with the recovery of the mobile robot from the pipe. Can be prevented from becoming. As a result, according to the configuration (20) above, even if the mobile robot becomes inoperable in the pipe due to being stuck in the pipe, the mobile robot can be smoothly collected.

(21) In one exemplary embodiment, in the configuration of (20) above, the telescopic mechanism of each of the plurality of arms is
A spring member that expands and contracts so as to press the crawler type traveling body against the inner surface of the pipe by a spring urged along the radial direction of the pipe.
A cylinder device configured to change the amount of expansion and contraction of the arm portion with respect to the main body portion by the expansion and contraction operation along the radial direction of the pipe.
An expansion / contraction amount adjusting unit configured to control the expansion / contraction amount of the arm portion by driving the cylinder device in response to an expansion / contraction instruction signal received from the control unit.
It is characterized by having.

According to this embodiment, when the user detects that the mobile robot is seated in the pipe by a sensor or the like mounted on the mobile robot, the control unit can transmit an expansion / contraction instruction signal to the expansion / contraction amount adjusting unit. .. At that time, the control unit instructs the expansion / contraction amount adjusting unit to contract the plurality of arms that support the plurality of crawler-type traveling bodies inward in the radial direction, thereby causing the plurality of arms to contract. Can be contracted. Therefore, when the mobile robot's seat is detected, the arm protruding from the main body is contracted by remote control from the control unit, and when the mobile robot is recovered from the pipe, the arm and the crawler type traveling body interfere with each other. Can be prevented from becoming.

(22) In one exemplary embodiment, in the configuration of (21) above, the control unit adaptively sends an expansion / contraction instruction signal given to the expansion / contraction amount adjusting unit according to the shape of the inner surface of the pipe. By selecting, it is further configured to control the cylinder device so as to secure a ground contact force required for traveling of the crawler type traveling body with respect to the inner surface of the pipe.

For example, when the shape of the pipe is a T-shape having a bent portion bent at a right angle, the bent portion has a peculiar inner wall surface shape different from that of the straight pipe portion in the pipe. Therefore, in order for the mobile robot to smoothly pass through the bent portion, it is necessary to appropriately expand and contract the arm portion according to the shape of the inner wall surface. Therefore, in the configuration of (22) above, the ground contact force required for traveling of the crawler type traveling body is secured with respect to the inner surface of the pipe by the action of the spring member and the cylinder device that respectively constitute the expansion / contraction mechanism of the plurality of arms. I try to do it. Specifically, when the mobile robot passes through the bent portion in the pipe, the necessary ground contact force is secured as follows by the action of the spring member and the cylinder device. That is, by pressing the crawler type traveling body against the inner surface of the pipe according to the peculiar inner wall surface shape at the bent portion, the ground contact force required for the running of the crawler type traveling body is maintained against the inner surface of the pipe. Is possible.

(23) The in-pipe construction method according to some embodiments of the present invention is
A step of detecting the state of the inner surface of the pipe while moving the surface condition sensor in the longitudinal direction inside the pipe in the construction target range in the longitudinal direction of the pipe.
Based on the detection result of the surface condition sensor for the construction target range in which the scanning of the surface condition sensor is completed, three-dimensional data indicating the position coordinates of the inner surface of the pipe in the construction target range is acquired. Then, the step of acquiring the three-dimensional map in which the irradiation area of the laser beam is specified on the three-dimensional data, and
By controlling the laser irradiation unit at each of one or more positions in the pipe along the longitudinal direction over the construction target range, the irradiation of the laser light toward the inner surface of the pipe is controlled. Steps to do and
With
In the step of controlling the irradiation of the laser beam, the laser irradiation unit is controlled so that the laser beam is selectively irradiated to the irradiation area within the construction target range based on the three-dimensional map. It is characterized by that.

According to the configuration of (23) above, the laser light irradiation area based on the detection result of the surface condition sensor is specified for the construction target range in which the scanning of the surface condition sensor is completed by moving the first mobile robot in the longitudinal direction in the pipe. Since the laser irradiation unit irradiates the laser beam using the three-dimensional map, the scanning timing by the surface state sensor can be shifted sufficiently before the laser beam irradiation timing. That is, according to the configuration of (23) above, the first mobile robot equipped with the surface state sensor passes through the area before the laser light irradiation area becomes hot due to the laser irradiation, and the surface state sensor Detection accuracy is less likely to decrease or failure due to high heat. Therefore, the irradiation area of the laser beam can be correctly specified from the inner surface state of the pipe, and the laser beam can be selectively irradiated to the correctly specified irradiation area.

(24) The in-pipe construction method according to some embodiments of the present invention is
A mobile robot, a surface state sensor mounted on the mobile robot in front of the mobile robot in the traveling direction, and a laser irradiation unit mounted on the mobile robot behind the surface state sensor in the traveling direction. Steps to put the construction robot including, into the pipe,
A step of moving the construction robot in the traveling direction in the piping,
A step of detecting the state of the inner surface of the pipe by the surface state sensor while the construction robot is moving in the traveling direction in the construction target range in the longitudinal direction of the pipe.
By controlling the laser irradiation unit based on the detection result of the surface state sensor each time the construction robot is positioned at each of one or more positions arranged along the traveling direction over the construction target range. A step of controlling the irradiation of laser light from the laser irradiation unit toward the inner surface of the pipe, and
It is characterized by having.

In the configuration (24) above, a surface condition sensor for detecting the condition of the inner surface of the pipe is mounted in front of the mobile robot in the traveling direction, and laser irradiation for irradiating the inner surface of the pipe with laser light is performed. The unit is mounted rearward in the direction of travel of the mobile robot. Therefore, in the configuration of (24) above, while the moving robot within the construction target range is moving in the traveling direction, the step of detecting the surface state in the pipe by the surface state sensor and the laser irradiation unit are the inner surface of the pipe. Even if the steps of performing laser irradiation are simultaneously executed, the surface state sensor passes through the irradiation area to be laser-irradiated before the temperature rises due to the laser irradiation. As a result, in the configuration of the above (24), a single mobile robot is used to perform a step of detecting the surface state in the pipe and a step of irradiating the inner surface of the pipe with laser without exposing the surface state sensor to high heat. By executing at the same time, the construction work in the pipe can be completed in a short time.

From the above, according to some embodiments of the present invention, even if high heat is generated in the pipe by irradiating the inner surface of the pipe with laser while the mobile robot moves in the pipe, the device included in the mobile robot It is possible to obtain a construction system in a pipe in which the detection accuracy of the type and the sensors does not deteriorate due to the high heat.

It is a figure which shows the whole structure of the construction system in a pipe which concerns on some embodiments. It is a figure which shows the structure of the construction robot among the construction system in a pipe which concerns on some embodiments. It is a figure which shows the structure of the control system among the construction system in a pipe which concerns on some embodiments. It is a figure explaining the concept of a 3D map. It is a figure which shows the perspective view of the mobile robot and the front view of the mobile robot moving in a pipe according to some embodiments. It is a figure which shows the structure which supports the crawler type traveling body from the main body part so that it can expand and contract along the radial direction. It is a figure which shows the space recognition sensor which consists of a 3D scanner and an image pickup apparatus. It is a figure explaining the operation of acquiring the point cloud data representing the entire internal space of a pipe by a three-dimensional scanner. It is a figure which shows the scenario which detects the self-position while moving in a pipe by a construction robot. It is a figure which shows the scenario which detects the self-position while moving in a pipe by a construction robot. It is a figure which shows the scenario which detects the self-position while moving in a pipe by a construction robot. It is a conceptual diagram of a microphone array used for self-position detection of a mobile robot. It is a flowchart which shows the flow of the processing operation executed by the construction system in a pipe. It is a figure which shows the structure of the mobile robot of the in-pipe construction system which concerns on still another embodiment. It is a figure which shows the state of performing the construction work in a pipe using two mobile robots according to still another embodiment. It is a figure which shows the structure of the mobile robot of the in-pipe construction system which concerns on still another embodiment. It is a figure which shows the structure which photographes the surface state in a pipe using a plurality of wide-angle lenses arranged in the circumferential direction of a mobile robot. It is a figure which shows the structure of the learning mechanism which improves the accuracy of image recognition according to the machine learning process based on deep learning.

Hereinafter, some embodiments of the present invention will be described with reference to the accompanying drawings. However, the dimensions, materials, shapes, relative arrangements, etc. of the components described as embodiments or shown in the drawings are not intended to limit the scope of the present invention to this, but are merely explanatory examples. Absent.
For example, expressions such as "same", "equal", and "homogeneous" that indicate that things are in the same state not only represent exactly the same state, but also have tolerances or differences to the extent that the same function can be obtained. It shall also represent the state of existence. On the other hand, the expressions "equipped", "equipped", "equipped", "included", or "have" one component are not exclusive expressions that exclude the existence of other components.

Hereinafter, first, the configuration of the construction system in the pipe according to some embodiments will be described with reference to FIGS. 1 to 3. Subsequently, the three-dimensional map used by the construction system in the pipe according to some embodiments will be described with reference to FIG. Subsequently, the configurations of the surface state sensor and the mobile robot constituting the construction system in the pipe according to some embodiments will be described with reference to FIGS. 5 and 7. Subsequently, the operation of the construction system in the pipe according to some embodiments will be described with reference to FIGS. 5 to 9. Subsequently, the configuration and operation of the construction system in the pipe according to still another embodiment will be described with reference to FIGS. 10 and 13. Finally, a learning mechanism for enabling the construction system in the pipe according to some embodiments to correctly recognize the state of the inner surface of the pipe to be constructed will be described with reference to FIG.

The construction system 1 in the pipe shown in FIG. 1 is composed of a construction robot 10 and a control system 3 that perform construction on the inner surface of the pipe 2 while moving in the pipe 2. The construction robot 10 investigates the spatial shape and the state of the inner surface in the pipe 2 while moving in the longitudinal direction in the pipe 2 over the area to be constructed in the pipe 2, and according to the result of the investigation, the inner surface of the pipe 2 is investigated. The construction is carried out by irradiating the space with a laser. The control system 3 may be a computer system that controls the operation of the construction robot 10 while transmitting and receiving control signals to and from the construction robot 10 via a communication cable.

Here, the area of the pipe 2 that is the target of the construction work using the construction robot 10 is called the construction target range R1 and is defined along the longitudinal direction in the pipe 2. The construction target range R1 is set in advance by the method described later below prior to the construction work using the construction system 1 in the pipe. Further, the construction target range R1 is set to cover the entire range (from one end to the other end of the pipe 2) or only a part thereof along the longitudinal direction in the pipe 2. You may.

For example, in one example, the construction target range R1 may be defined as follows. When a part of the pipe 2 is made of a material containing stainless steel, the inner surface of the part of the pipe 2 has high erosion resistance, so that it is necessary to frequently perform the construction work using the construction robot 10 in the part. Absent. Therefore, in this case, the portion of the pipe 2 made of a material that does not contain stainless steel may be set as the construction target range R1. Further, as another example, when the pipe shape of the pipe 2 is a topology in which one-stroke writing is impossible (such as a T-shaped pipe) and the inner surface of the pipe 2 is irradiated with a laser, the construction target is The range R1 may be set as follows. That is, in this case, since it is necessary to move the construction robot 10 a plurality of times along a plurality of different movement paths in the pipe 2, each of the plurality of different movement paths constituting the entire pipe shape of the pipe 2 is used. It may be set as the construction target range R1 of.

Hereinafter, the configuration of the construction robot 10 will be described in detail with reference to FIG. 2, and the configuration of the control system 3 will also be described in detail with reference to FIG.

FIG. 2 shows the configuration of the construction robot 10 among the construction systems 1 in the pipe according to some embodiments. The construction robot 10 irradiates the surface condition sensor 14 for detecting the state of the inner surface of the pipe 2 and the laser beam toward the inner surface of the pipe over the construction target range R1 preset in the longitudinal direction of the pipe 2. The laser irradiation unit 15 is provided. Further, the construction robot 10 includes a first mobile robot 11a and a second mobile robot 11b as moving means for moving the surface state sensor 14 and the laser irradiation unit 15 in the pipe 2. Specifically, the first mobile robot 11a moves the surface state sensor 14 in the longitudinal direction (d1, d2) inside the pipe 2 in the construction target range R1 in the longitudinal direction (d1, d2) of the pipe 2. It is configured as follows. Further, the second mobile robot 11b is configured to move the laser irradiation unit 15 in the longitudinal direction (d1, d2) inside the pipe 2 in the construction target range R1.

In the construction robot 10 shown in FIG. 2, the first mobile robot 11a and the second mobile robot 11b form a single mobile robot 11. Further, the mobile robot 11 is installed between the power cable C1 for supplying the driving power and the power for generating the laser beam from the external power source to the construction robot 10 and the control system 3 described later with reference to FIG. A communication cable C2 for transmitting and receiving a control signal for controlling the robot 10 is connected. The first mobile robot 11a and the second mobile robot 11b can be realized as separate mobile robots. However, in the following embodiments described later with reference to FIGS. 1 to 13, for convenience of explanation, the first mobile robot 11a and the second mobile robot 11b are described as forming a single mobile robot 11. I do.

Next, a control system 3 that controls the construction robot 10 while transmitting and receiving control signals to and from the construction robot 10 will be described with reference to FIG. As shown in FIG. 3A, the control system 3 may be configured by connecting the control terminal 31, the three-dimensional CAD system 32, the arithmetic processing server 33, and the hub 35 via the communication network 34. .. A communication cable C2 from the mobile robot 11 shown in FIG. 2 is connected to the hub 35 to relay signal transmission / reception between the control terminal 31, the three-dimensional CAD system 32, the arithmetic processing server 33, and the construction robot 10. There is.

Further, as shown in FIG. 3B, the control terminal 31 includes a control unit 310a, a self-position acquisition unit 310b, and a marker detection unit 310c. In one embodiment, the control unit 310a, the self-position acquisition unit 310b, and the marker detection unit 310c may be implemented as software programs executed on the control terminal 31. The control unit 310a is configured to control the operation of each component device constituting the construction robot 10 by sending a control command to each component device constituting the construction robot 10, as will be described in detail below. The self-position acquisition unit 310b is configured to detect the self-position in the pipe 2 of the moving robot 11 moving in the pipe 2 as the second mobile robot 11b and output the self-position to the control unit 310a. The marker detection unit 310c will be described later. Further, the control unit 310a may be configured to include a map acquisition unit 311 and a command generation unit 312, which will be described later.

The map acquisition unit 311 acquires the three-dimensional map 420 shown in FIG. 4 as map information used by the control unit 310a to determine the position where the laser irradiation should be performed in the construction target area R1 of the pipe 2. Specifically, the map acquisition unit 311 acquires a three-dimensional map 420 in which the irradiation area 160 of the laser beam 16 is specified on the three-dimensional data 410 indicating the position coordinates of the inner surface of the pipe 2 in the construction target range R1. It is configured as follows. As will be described in detail with reference to FIGS. 9 to 13, the map acquisition unit 311 may acquire the three-dimensional map 420 as follows. That is, based on the detection result of the surface condition sensor 14 for the construction target range R1 in which the scanning of the surface condition sensor 14 is completed, the three-dimensional data 410 indicating the three-dimensional position coordinates of the inner surface of the pipe 2 in the construction target range R1 ( FIG. 4) is calculated by the procedure described later. Subsequently, the map acquisition unit 311 acquires the three-dimensional map 420 in which the irradiation area 160 of the laser beam 16 (160a and 160b in FIG. 4 and the like) is specified on the three-dimensional data 410 (FIG. 4).

In one exemplary embodiment, the map acquisition unit 311 creates a three-dimensional map 420 as follows based on the detection result of the surface condition sensor 14 for the construction target range R1 in which the scanning of the surface condition sensor 14 is completed. It may be configured to acquire. First, the map acquisition unit 311 calculates the three-dimensional data 410 representing the spatial shape of the inner surface of the pipe 2 in the construction target range R1. Subsequently, the map acquisition unit 311 sequentially superimposes a plurality of data representing the positions and ranges of the plurality of irradiation areas 160 (160a and 160b in FIG. 4 and the like) irradiated with the laser beam 16 on the three-dimensional data 410. Acquire a 3D map by going through. In other words, the map acquisition unit 311 performs a process of superimposing the three-dimensional data of the irradiation area 160 on the three-dimensional data 410 of the inner surface of the pipe 2 in the construction target range R1. As a result, the geometrical range of the irradiation area 160 is displayed on the three-dimensional map 420 by the three-dimensional data created by superimposing the three-dimensional data of the irradiation area 160 on the three-dimensional data 410 of the inner surface of the pipe 2. It is revealed.

In this embodiment, in the initial stage, three-dimensional data 410 representing the spatial shape of the inner surface of the pipe 2 in the construction target range R1 is calculated. Further, in this embodiment, in the second and subsequent stages following the initial stage, a plurality of data representing the positions and ranges of the plurality of irradiation areas 160 (160a and 160b in FIG. 4 and the like) to be irradiated with the laser beam 16 are three-dimensionally displayed. The three-dimensional map 420 is gradually generated so as to be sequentially superimposed on the data 410. As a result, according to this embodiment, the three-dimensional map 420 is efficiently synthesized by adding the geometric information about each of the plurality of irradiation areas 160 to the three-dimensional data 410 by an incremental method. can do.

Further, the command generation unit 312 is a laser irradiation unit so that the laser beam 16 is selectively irradiated to the irradiation area 160 in the construction target range R1 based on the three-dimensional map 420 received from the map acquisition unit 311. It is configured to send commands to 15 and the mobile robot 11 (as the second mobile robot 11b).

In some embodiments, three-dimensional data indicating the position coordinates of the inner surface of the pipe 2 when the inner surface of the pipe 2 is in the following state within the construction target range R1 of the pipe 2. The area 160 to be irradiated with the laser beam 16 is specified on the 410. For example, when a layer of oxide scale (rust on the inner surface, etc.) is formed on the inner surface of the pipe 2 within the construction target range R1 of the pipe 2, the adhesion area of the oxide scale and the thickness of the oxide scale are formed. It is highly necessary to perform construction by laser irradiation in places where the area is large. Therefore, such a location on the inner surface of the pipe 2 may be specified as the irradiation area 160. Further, when the construction system 1 in the pipe is used to remove the coating film formed on the inner surface of the pipe 2, the coating film is lifted or the thickness of the coating film is raised within the construction target range R1 of the pipe 2. In places where there is a large amount of space, it is highly necessary to perform construction by laser irradiation. Therefore, such a location on the inner surface of the pipe 2 may be specified as the irradiation area 160.

On the other hand, the three-dimensional CAD system 32 in the control system 3 stores the CAD data of the design shape of the pipe 2 obtained at the time of designing the shape of the pipe 2 in its own database 32a as the pipe shape design information. The three-dimensional CAD system 32 transmits the pipe shape design information to the self-position acquisition unit 310b in response to an access request from the self-position acquisition unit 310b in the control terminal 31. As will be described in detail below, the self-position acquisition unit 310b that has received the pipe shape information calculates the self-position of the mobile robot 11 that is moving in the pipe 2 as the second mobile robot 11b in the pipe 2. When performing processing, pipe shape design information is used as auxiliary information.

Further, in one exemplary embodiment, the process of calculating the three-dimensional data 410 and the three-dimensional map 420 acquired by the map acquisition unit 311 is executed by the arithmetic processing server 33 shown in FIG. 3A in the following procedure. May be done. First, the arithmetic processing server 33 receives measurement data from sensor devices including the surface state sensor 14 provided on the construction robot 10 and other sensor devices via the communication cable C2. Subsequently, by performing analysis processing and arithmetic processing on the measurement data, the control unit 310a generates control input information used for controlling each component device constituting the construction robot 10, and the control unit 310a generates control input information. Output. Further, by performing analysis processing and arithmetic processing on the measurement data from the sensor devices provided on the construction robot 10, the self-position acquisition unit 310b causes the self-position of the mobile robot 11 (as the second mobile robot 11b). The input information used for calculating the above is generated and output to the self-position acquisition unit 310b.

Next, a procedure for constructing the pipe 2 by the construction robot 10 and a control operation when the construction robot 10 is controlled by the control system 3 will be described according to an exemplary embodiment. First, as a preparation prior to explaining the construction operation of the construction robot 10 and the control operation by the control system 3, the detailed configuration of the construction robot 10 used in this embodiment will be described.

As shown in FIG. 2, the mobile robot 11 that moves in the pipe 2 has a tubular main body portion 17 that extends along the longitudinal direction (d1, d2) of the pipe 2, and follows different radial directions of the pipe 2. The arm portions 12 and 12'extend from the main body portion 17, respectively. Further, the crawler type traveling bodies 13 and 13'provided at the tips of the arms 12 and 12'are in contact with the inner surface of the pipe 2 so as to be able to travel on the inner surface of the pipe 2. Then, by driving the crawler type traveling bodies 13 and 13'with the drive motors 18 and 18', the mobile robot 11 can be moved along the longitudinal direction (d1, d2) of the pipe 2.

Further, a control signal for controlling the construction robot 10 is transmitted to and received from one end surface 17a of the main body 17 between the power cable C1 for supplying electric power to the construction robot 10 from an external power source and the control system 3. Communication cable C2 for this is connected. Further, on the other end surface 17b of the main body portion 17, among the laser irradiation portions 15, a nozzle portion 15c and a ring motor portion 15d, which will be described later, are arranged. Further, a surface state sensor 14 described later is provided on the upper wall surface constituting the side surface portion of the main body portion 17, and the surface state sensor 14 is located near the end surface 17b on the upper wall surface of the main body portion 17. positioned.

In one example, as shown in FIG. 5, three arm portions 12a, 12b, and 12c extend from the main body portion 17 along the circumferential direction of the pipe 2 at a distance of 120 ° along the radial direction of the pipe 2. The crawler-type traveling bodies 13a, 13b and 13c provided at the tips of the three arms 12a, 12b and 12c abut on the inner surface of the pipe 2 so as to be able to travel on the inner surface of the pipe 2. It may be configured as follows. The arms 12a, 12b and 12c may be elastically configured so that the distance between the main body 17 and the crawler-type traveling bodies 13a, 13b and 13c can be adjusted. As a result, the main body 17 and the crawler type traveling bodies 13a, 13b and 13c are brought into contact with the inner surface of the pipe 2 with an appropriate pressing force according to the inner diameter of the pipe 2. The distance between them is adjusted.

In the configuration of the mobile robot 11 shown in FIG. 5, a structure in which the crawler type traveling body 13 (13a to 13c) is supported by the arm portions 12 (12a to 12c) extending from the main body portion 17 is viewed from the front in the traveling direction of the mobile robot 11. FIG. 6 is shown in more detail as a front view. As described above, the mobile robot 11 has a main body portion 17 formed in an elongated shape extending along the longitudinal direction of the pipe 2, and in FIG. 6, the main body portion 17 has a circular cross section and the pipe 2 It is depicted as having a cylindrical field extending in the longitudinal direction of. Further, as shown in FIG. 6, a plurality of crawler type traveling bodies 13 (13a to 13c) arranged at different positions in the circumferential direction of the main body 17 and configured to travel on the inner surface of the pipe 2 A plurality of crawler type traveling bodies 13 (13a to 13c) are arranged at different positions in the circumferential direction of the main body portion 17, and are configured to support the crawler type traveling bodies 13 (13a to 13c) so as to be expandable and contractible along the radial direction of the pipe 2. It is configured to include a plurality of arm portions 12 (12a to 12c) each having a mechanism.

According to the above configuration shown in FIG. 6, a plurality of crawler-type traveling bodies 13 (13a to 13c) installed at a plurality of positions along the circumferential direction of the main body 17 of the mobile robot 11 are provided with a plurality of arm portions 12 ( The expansion and contraction mechanisms provided in each of 12a to 12c) can be expanded and contracted along the radial direction, so that the following technical advantages can be obtained.

For example, consider a case where the mobile robot 11 comes into contact with the inner surface of the pipe 2 or sits down at a bent portion of the pipe 2 and gets stuck in the pipe 2 so that the mobile robot 11 becomes inoperable. In this case, the plurality of arm portions 12 (12a to 12c) supporting the plurality of crawler type traveling bodies 13 (13a to 13c) can be contracted. Therefore, when the mobile robot 11 is collected from the pipe 2, the arm portions 12 (12a to 12c) protruding from the main body portion 17 and the crawler type traveling body 13 (13a to 13c) can be prevented from getting in the way. .. From the above, according to the above configuration, even if the mobile robot 11 becomes inoperable in the pipe 2 due to being stuck in the pipe 2, the mobile robot 11 can be smoothly collected.

Further, in the embodiment shown in FIG. 6, the expansion / contraction mechanism of each of the plurality of arm portions 12 (12a to 12c) is expanded / contracted with the spring member 121 (121a to 121c) and the cylinder device 122 (122a to 122c) described later. It may be composed of the quantity adjusting unit 123 (123a to 123c). Here, the spring member 121 (121a to 121c) is a member that expands and contracts so as to press the crawler type traveling body 13 (13a to 13c) against the inner surface of the pipe 2 by a spring urged along the radial direction of the pipe 2. Is. Further, the cylinder device 122 (122a to 122c) is a member configured to change the amount of expansion and contraction of the arm portion 12 (12a to 12c) with respect to the main body portion 17 by the expansion and contraction operation along the radial direction of the pipe 2. .. In one example, the cylinder device 122 (122a-122c) may be implemented using an air cylinder. Further, the expansion / contraction amount adjusting unit 123 (123a to 123c) expands and contracts the arm portion 12 (12a to 12c) by driving the cylinder device 122 (122a to 122c) in response to the expansion / contraction instruction signal received from the control unit 310a. It is configured to control the amount.

Therefore, according to this embodiment, when the user detects that the mobile robot 11 has been seated in the pipe 2 by a sensor or the like mounted on the mobile robot 11, the mobile robot 11 seated in the pipe 2 is piped as follows. The work of taking out from the inside can be facilitated. First, the user gives a command to the control unit 310a to contract the arm portions 12 (12a to 12c) inward in the radial direction. For example, the user may give such a command to the control unit 310a by operating the keyboard or the like of the control terminal 31. Then, the control unit 310a that has received the command transmits an expansion / contraction instruction signal to the expansion / contraction amount adjusting units 123 (123a to 123c) in response to the command. At that time, the expansion / contraction instruction signal transmitted from the control unit 310a instructs the expansion / contraction amount adjusting unit 123 (123a to 123c) of the expansion / contraction amount for maximally contracting the arm portion 12 (12a to 12c) inward in the radial direction. It may be a control signal to be used. In this way, the amount of expansion and contraction of the plurality of arm portions 12 (12a to 12c) is remotely controlled by remote control via the control unit 310a on the control terminal 31. As a result of such remote control, the plurality of arm portions 12 (12a to 12c) supporting the plurality of crawler type traveling bodies 13 (13a to 13c) can be contracted toward the main body portion 17. Therefore, when the mobile robot 11 is recovered from the pipe 2, the arm portions 12 (12a to 12c) protruding from the main body portion 17 and the crawler type traveling body 13 (13a to 13c) are performed by performing the remote control as described above. ) Can be out of the way.

Further, in the embodiment shown in FIG. 6, the control unit 310a is configured to adaptively select the expansion / contraction instruction signal given to the expansion / contraction amount adjusting units 123 (123a to 123c) according to the shape of the inner surface of the pipe 2. It may have been done. As a result, the control unit 310a adaptively selects the expansion / contraction instruction signal given to the expansion / contraction amount adjusting units 123 (123a to 123c), whereby the crawler type traveling body 13 (13a to 13a to) with respect to the inner surface of the pipe 2. The cylinder devices 122 (122a to 122c) can be controlled so as to secure the ground contact force required for traveling the 13c).

As described above, the expansion / contraction mechanism of each of the plurality of arm portions 12 (12a to 12c) has the structure as described above with reference to FIG. 6, and the control unit 310a is attached to the expansion / contraction amount adjusting unit 123 (123a to 123c). By adaptively selecting the expansion / contraction instruction signal to be given to the pipe 2 according to the shape of the inner surface of the pipe 2, the following technical advantages can be obtained. For example, when the shape of the pipe 2 is a T-shape having a bent portion bent at a right angle, the bent portion has a peculiar inner wall surface shape different from that of the straight pipe portion in the pipe 2. Therefore, in order for the mobile robot 11 to smoothly pass through the bent portion, it is necessary to appropriately expand and contract the arm portions 12 (12a to 12c) according to the shape of the inner wall surface. Therefore, in the above configuration, the spring members 121 (121a to 121c) and the cylinder devices 122 (122a to 122c), which respectively constitute the expansion / contraction mechanism of the plurality of arm portions 12 (12a to 12c), act on the inner surface of the pipe 2. On the other hand, the ground contact force required for traveling of the crawler type traveling bodies 13 (13a to 13c) is secured. Specifically, when the mobile robot 11 passes through the bent portion in the pipe, the spring member 121 (121a to 121c) and the cylinder device 122 (122a to 122c) work to generate the necessary ground contact force as follows. It will be secured. That is, by pressing the crawler type traveling body 13 (13a to 13c) against the inner surface of the pipe 2 according to the peculiar inner wall surface shape at the bent portion, the crawler type traveling body 13 (13a to 13a to 13a) is pressed against the inner surface of the pipe 2. It is possible to continue to maintain the ground contact force required for traveling in 13c).

When the construction work performed on the pipe 2 by the construction system 1 in the pipe is the work of removing the oxide scale formed on the inner surface of the pipe 2, the oxidation is performed by irradiating the laser beam from the laser irradiation unit 15. When the material scale layer is cauterized, dusty burning debris is produced. Therefore, in that case, a vacuum pump (not shown) connected to the nozzle portion 15c via a suction hose is further provided in the main body portion 17 of the mobile robot 11, and is dusty from the laser ejection port 15b at the tip of the nozzle 15c. It is possible to suck and collect the burning debris.

Subsequently, based on the above description regarding the configuration of the construction robot 10, the construction operation of the construction robot 10 and the control operation by the control system 3 will be described. The construction robot 10 inserted into the pipe 2 first executes the first work phase in order to construct the inside of the pipe 2. In the first work phase, the mobile robot 11 (as the first mobile robot 11a) moves on the movement path that spans the entire construction target range R1 in the pipe 2. As a result, the surface state sensor 14 mounted on the mobile robot 11 (as the first mobile robot 11a) scans the entire construction target range R1 in the pipe 2 while moving in the pipe 2.

In this way, when the surface condition sensor 14 scans the construction target range R1 in the pipe 2 and the scanning is completed over the entire construction target range R1, the surface condition sensor 14 covers the entire construction target range R1 for which the scanning is completed. Spatial measurement information representing the scanning result is output as sensor output data. In one exemplary embodiment, the surface condition sensor 14 that scans the entire construction target range R1 and outputs spatial measurement information representing the scanning result over the entire construction target range R1 is a space recognition sensor 14 as shown in FIG. It may be configured as. As shown in FIG. 7, the surface condition sensor 14 may be configured to include at least one 3D scanner 140. The three-dimensional scanner 140 includes an irradiation port 140a for irradiating the scanning signal wave in a form of scanning in all directions and a reflected wave detection sensor 140b for detecting the reflected wave of the irradiated scanning signal wave. In this embodiment, the three-dimensional scanner 140 operates as the first three-dimensional scanner 141 for measuring the surface state in the pipe 2 in the construction target range R1. In one example, the scanning signal wave emitted from the irradiation port 140a of the three-dimensional scanner 140 may be any signal wave including a range-finding laser, an electromagnetic wave, and an ultrasonic wave.

The first three-dimensional scanner 141 is configured to be movable in the pipe 2 in the construction target range R1 by the mobile robot 11 (as the first mobile robot 11a). Then, the first three-dimensional scanner 141 irradiates the scanning signal wave and scans the three-dimensional space in the pipe 2, so that the pipe is connected from the irradiation source of the scanning signal wave (for example, the irradiation port 140a shown in FIG. 7). The first point group data representing each distance and direction to a plurality of points constituting the space in 2 is acquired as space measurement information representing the space shape in the pipe 2. At this time, while moving along the axial direction of the pipe 2 in the construction target range R1 of the pipe 2, the first three-dimensional scanner 141 scans the scanning signal wave at each position along the axial direction of the pipe 2. Acquire point group data of the space in the vicinity of the construction robot 10. Then, by synthesizing the point cloud data acquired for the space near each position along the axial direction (longitudinal direction) of the pipe 2 along the axial direction (longitudinal direction) of the pipe 2, the entire construction target range R1 is obtained. The first point cloud data can be obtained.

In one exemplary embodiment, the first three-dimensional scanner 141 irradiates the distance-finding laser and scans the three-dimensional space in the pipe 2 to irradiate the distance-finding laser irradiation source (eg, FIG. 7). It may be realized as a SLAM device that acquires first point group data representing the respective distances and directions from the irradiation port 140a) shown in the above to a plurality of points forming the space in the pipe 2. In this embodiment, the first three-dimensional scanner 141 implemented as the SLAM device uses the point cloud data representing the entire internal space of the pipe 2 as the first point cloud data by the method described later with reference to FIG. It may be configured to be obtained as.

Referring to FIG. 8, the mobile robot 11 is in a state of moving in the traveling direction indicated by the arrow w1, and is located at the longitudinal position L1 in the pipe 2 at time T1 = t (FIG. 8A). At this time, the mobile robot 11 acquires the point cloud data related to the point cloud P (p1 to p8 shown in FIG. 8A) by performing spatial scanning with the scanning signal wave at the position L1 in the longitudinal direction. At the following time T2 = t + Δt, the mobile robot 11 moves to the longitudinal position L2 in the pipe 2 (FIG. 8B), and at the longitudinal position L2, spatial scanning by the scanning signal wave is performed to perform the point cloud Q ( The point cloud data related to q1 to q8 shown in FIG. 8B) is acquired. At the following time T3 = t + 2 × Δt, the mobile robot 11 moves to the longitudinal position L3 in the pipe 2 (FIG. 8C), and at the longitudinal position L3, a point cloud is performed by performing spatial scanning with a scanning signal wave. Point cloud data related to R (r1 to r6 shown in FIG. 8C, etc.) is acquired.

Then, by synthesizing the point cloud data relating to the point clouds P, Q and R along the arrow w1, it is possible to acquire the point cloud data representing the internal space of the pipe 2 over the range from the longitudinal positions L1 to L3. .. In this way, the mobile robot 11 acquires local point cloud data at each of the plurality of positions arranged along the longitudinal direction in the pipe 2, and the local point cloud data acquired at each of the plurality of positions is collected in the length of the pipe 2. By synthesizing along the direction, it is possible to acquire the point cloud data representing the entire internal space of the pipe 2 as the first point cloud data.

As will be described later, the spatial measurement information acquired as described above as the first point cloud data representing the internal space shape of the pipe 2 using the first three-dimensional scanner 141 is shown in FIG. It may be used to generate the dimension map 420. Further, the surface state sensor 14 may further include an image pickup device 142 that outputs an image of the inner surface of the pipe 2 to the control unit 31, and the image pickup device 142 includes a lens section 142a. In one embodiment, the image taken by the image pickup apparatus 142 detects the irradiation area 160 in the construction target range R1 by grasping the surface state of the inner surface of the pipe 2 in the construction target range R1 in detail by image recognition processing. May be used to Further, in another embodiment, the image taken by the image pickup apparatus 142 is a pipe in the construction target range R1 by grasping the surface state of the inner surface of the pipe 2 in the construction target range R1 in detail by image recognition processing. It may be used to control the irradiation condition when irradiating the inner surface of No. 2 with a laser beam.

In this embodiment, the spatial measurement information representing the scanning result over the entire construction target range R1 that has been scanned by the surface condition sensor 14 is the spatial shape in the pipe 2 by the first three-dimensional scanner 141 included in the surface condition sensor 14. It may be obtained as measurement data obtained by three-dimensional measurement. For example, this spatial measurement information is obtained by irradiating the scanning signal wave with the first three-dimensional scanner 141 and scanning the three-dimensional space in the pipe 2 to irradiate the irradiation source of the scanning signal wave (for example, the irradiation shown in FIG. 7). It may be obtained as first point group data representing the respective distances and directions from the port 140a) to the plurality of points forming the space in the pipe 2.

As described above, the spatial measurement information measured as the first point cloud data by the first three-dimensional scanner 141 in the surface state sensor 14 is transferred to the arithmetic processing server 33 via the communication cable C2 and the hub 35. Will be sent. The arithmetic processing server that received the first point cloud data performs geometric arithmetic processing on the first point cloud data to indicate the three-dimensional position coordinates of the inner surface of the pipe 2 in the construction target range R1. Dimensional data 410 may be calculated. Further, the arithmetic processing server 33 performs further spatial shape analysis processing on the first point cloud data to calculate a three-dimensional map 420 in which the irradiation area 160 of the laser beam 16 is specified on the three-dimensional data 410. You may. When the three-dimensional data 410 and the three-dimensional map 420 are calculated as described above, the arithmetic processing server 33 transmits the three-dimensional data 410 and the three-dimensional map 420 to the map acquisition unit 311 in the control unit 310a. As described above, the map acquisition unit 311 has the three-dimensional map 420 in which the irradiation area 160 of the laser beam 16 is specified on the three-dimensional data 410 indicating the three-dimensional position coordinates of the inner surface of the pipe 2 in the construction target range R1. Will be acquired.

In this embodiment, the process of calculating the three-dimensional data 410 and the three-dimensional map 420 as described above has been described as being executed by the arithmetic processing server 33. However, the process of calculating the three-dimensional data 410 and the three-dimensional map 420 as described above may be executed by the control unit 310a.

When the construction robot 10 completes the first work phase by moving in the pipe 2, and the map acquisition unit 311 acquires the three-dimensional data 410 and the three-dimensional map 420 for the construction target range R1 in the pipe 2, subsequently. , The second work phase described later is performed. In the second work phase, the construction robot 10 irradiates the inner surface of the pipe 2 in the construction target range R1 with the laser irradiation unit 15 while moving along the movement path over the entire construction target range R1 in the pipe 2. Do.

During the execution of the second work phase, the control unit 310a controls the laser irradiation unit 15 configured to be movable in the pipe 2 by the mobile robot 11 (as the second mobile robot 11b). At the same time, the control unit 310a controls the moving speed and the moving direction of the moving robot 11 (as the second moving robot 11b) while detecting the self-position of the moving robot 11 as the second moving robot 11b. As a result, the control unit 310a appropriately controls the irradiation position for irradiating the inner surface of the pipe 2 with the laser beam and the irradiation conditions for the laser light while the construction robot 10 is moving in the pipe 2 in the second work phase. To do. More specifically, the command generation unit 312 in the control unit 310a executes the following control operation.

First, the command generation unit 312 receives the three-dimensional data 410 and the three-dimensional map 420 for the construction target range R1 of the pipe 2 from the map acquisition unit 311 in the control unit 310a. Subsequently, the command generation unit 312 sets the laser irradiation unit 15 and (second movement) so that the laser light 16 is selectively irradiated to the irradiation area 160 within the construction target range R1 based on the three-dimensional map 420. A control command is sent to the mobile robot 11 (as the robot 11b). At that time, the command generation unit 312 in the control unit 310a may control the irradiation conditions of the laser beam based on the surface state in the pipe 2 detected by the surface state sensor 14. As a result, the type of construction work (removal of oxide scale, cleaning, removal of coating film, etc.) by irradiating the inner surface of the pipe 2 with laser, and the surface condition of the inner surface of the pipe (for example, the adhesion area of the oxide scale). The construction work by laser irradiation can be performed over the construction target range R1 while controlling the irradiation conditions of the laser beam 16 according to the thickness of the oxide scale, the floating of the coating film, the thickness of the coating film, etc.). As a result, it is possible to prevent the wall thickness of the pipe 2 from being scraped off more than necessary by the laser irradiation by performing the laser irradiation at the required output for the required time at each construction location in the pipe 2.

Next, a specific configuration of the laser irradiation unit 15 used in this embodiment will be described. The nozzle portion 15c constituting the laser irradiation portion 15 is configured to be rotatable along the circumferential direction in the pipe 2 so that the irradiation direction of the laser beam 16 rotates. As shown in FIG. 2, the nozzle portion 15c is configured to guide the laser light 16 from the laser light generator 15e to the laser emission port 15b by the reflection mirror 15a. The laser light generator 15e constituting the laser irradiation unit 15 uses the electric power supplied from the power cable C1 as energy for laser generation to generate the laser light 16 guided to the reflection mirror 15a of the nozzle unit 15c. It is a device and may be built in the main body 17. Here, by moving one end of the reflection mirror 15a in the direction y1 or the direction y2 inside the nozzle portion 15c, the direction in which the laser beam 16 is emitted from the laser ejection port 15b is the longitudinal direction (d1, d2) of the pipe 2. Can be fine-tuned along. Further, the ring motor portion 15d constituting the laser irradiation portion 15 connects the base portion of the nozzle portion 15c to the end portion 17b of the main body portion 17 so that the nozzle portion 15c can rotate along the circumferential direction in the pipe 2. It is a ring motor. That is, when the rotation speed of the ring motor portion 15d is controlled so that the irradiation direction of the laser beam 16 rotates at an angular velocity ω along the circumferential direction in the pipe 2, the nozzle portion 15c is also piped as the ring motor portion 15d rotates. It rotates along the circumferential direction in 2. By making the nozzle portion 15c rotatable along the circumferential direction by the ring motor portion 15d as described above, the entire laser light source including the laser light generator 15e does not rotate in the circumferential direction in the main body portion 17. Can be. By adopting a configuration in which the entire laser light source is fixed in the circumferential direction in the main body 17 without rotating in this way, it is possible to prevent entanglement of cables and the like of equipment.

Hereinafter, in the second work phase, the operation performed by the construction robot 10 in the pipe 2 and the control operation executed by the control unit 310a to control the operation of the construction robot 10 will be described in more detail. In the second work phase, the mobile robot 11 (as the second mobile robot 11b) moves on the movement path that spans the entire construction target range R1 in the pipe 2. At that time, the command generation unit 312 of the control unit 310a refers to the three-dimensional map 420 received from the map acquisition unit 311 of the control unit 310a, and the irradiation area 160 exists at any position in the construction target range R1 of the pipe 2. , To grasp the range of the irradiation area 160.

Subsequently, the command generation unit 312 sets the rotational angular velocity ω when rotating the irradiation direction of the laser beam 16 along the circumferential direction in the pipe 2 to the spread width (longitudinal direction) along the circumferential direction of the irradiation area 160. It is necessary to control according to the above. This is because the laser on the mobile robot 11 (second mobile robot 11b) is located at the axial position (position in the longitudinal direction of the pipe 2) in which the irradiation area 160 in which the laser irradiation should be performed in the circumferential direction of the pipe 2 occupies a wide angle range. This is because the irradiation unit 15 needs to focus on sweeping the laser irradiation along the circumferential direction in the pipe 2. On the contrary, in the axial position where the angle range occupied by the irradiation area 160 is narrow, the laser irradiation unit 15 on the mobile robot 11 (second mobile robot 11b) may complete the sweeping of the laser irradiation in a short time. Therefore, the command generation unit 312 rotates in the pipe 2 so that the irradiation direction of the laser beam 16 rotates at an appropriate rotation angular velocity ω according to the wide angle range occupied by the irradiation area 160 in the circumferential direction of the pipe 2. The rotation speed of the nozzle portion 15c along the circumferential direction of is controlled.

Further, the command generation unit 312 of the pipe 2 according to the angle range occupied by the irradiation area 160 in which the laser irradiation should be performed in the circumferential direction of the pipe 2 at each position (axial position) in the pipe 2 along the longitudinal direction. The traveling speed of the mobile robot 11 (second mobile robot 11b) along the longitudinal direction is controlled. This is because the mobile robot 11 (second mobile robot 11b) travels in a place where the irradiation area 160 occupies a wide angle range and it is necessary to focus on sweeping the laser irradiation along the circumferential direction in the pipe 2. This is because it is necessary to slow down.

In order to realize the above control function, the command generation unit 312 is a laser irradiation unit that is moved by the mobile robot 11 (second mobile robot 11b) on the three-dimensional map 420 that represents the spatial structure in the pipe 2. The processing operation of projecting the current self-position of 15 is performed. Then, the command generation unit 312 selectively emits the laser light 16 to the irradiation area 160 within the construction target range 10 specified by the three-dimensional map 420 according to the current self-position of the laser irradiation unit 15. A control command is sent to the laser irradiation unit 15 and the mobile robot 11 (second mobile robot 11b) so as to be irradiated. As a result, when the control unit 310a includes the self-position of the laser irradiation unit 15 moved by the mobile robot 11 (as the second mobile robot 11b) in the irradiation area 160 specified on the three-dimensional map 420. It is possible to control the laser irradiation unit 15 so as to irradiate the laser beam 16.

As described above, it is necessary to detect the self-position of the laser irradiation unit 15 moving by the mobile robot 11 (second mobile robot 11) as input information for the control operation performed by the command generation unit 312. Therefore, in the self-position acquisition unit 310b that operates in the control terminal 31, the laser irradiation unit 15 that is moved by the mobile robot 11 (second mobile robot 11b) according to the method described later with reference to FIGS. 9 and 10 is currently located. The self-position information indicating the self-position in the pipe 2 is calculated. Subsequently, whether the command generation unit 312 that has received the self-position information from the self-position acquisition unit 310b includes the self-position of the laser irradiation unit 15 that is moved by the mobile robot 11 (second mobile robot 11b) in the irradiation area 160. Judge whether or not. Then, the command generation unit 312 transmits a control command to the laser irradiation unit 15 so as to irradiate the laser beam 16 when it is determined that the self-position is included in the irradiation area 160.

FIG. 9 shows an example of calculating self-position information from data measured using only the second three-dimensional scanner 143 included in the construction robot 10. First, prior to explaining the specific calculation method of the self-position information shown in FIG. 9, the second three-dimensional scanner 143 used for detecting the self-position will be described in detail. The 3D scanner 140 included in the surface state sensor 14 described above not only operates as a first 3D scanner 141 for obtaining a 3D map 420, but also operates as a second 3D scanner 143 as described below. It may be configured to do so. That is, the second three-dimensional scanner 143 is realized by the three-dimensional scanner 140 configured to be movable in the pipe 2 in the construction target range R1 by the mobile robot 11 (as the second mobile robot 11b). Then, the second three-dimensional scanner 143 is used to calculate the self-position of the construction robot 10 while the self-position acquisition unit 310b is moving in the pipe 2. Further, the second three-dimensional scanner 143, like the first three-dimensional scanner 141, constitutes the space in the pipe 2 by irradiating the scanning signal wave and scanning the three-dimensional space in the pipe 2. It is configured to acquire the second point group data representing the respective distances to a plurality of points as spatial measurement information representing the spatial shape in the pipe 2.

As described above, in the construction robot 10, the first three-dimensional scanner 141 and the second three-dimensional scanner 143 are realized in the surface state sensor 14 by the same three-dimensional scanner 140. However, in another embodiment, the second 3D scanner 143 may be provided as an independent 3D scanner separate from the surface condition sensor 14 including the 3D scanner 140.

Based on the above description, a specific method for calculating self-position information according to the embodiment shown in FIG. 9 will be described. As shown in FIG. 9, the second three-dimensional scanner 143 is moving along the arrow w1 in the construction target range R1 of the pipe 2 by the mobile robot 11 (second mobile robot 11b). Then, the moving second three-dimensional scanner 143 irradiates a scanning signal wave and scans the three-dimensional space in the pipe 2, so that each distance to a plurality of points constituting the space in the pipe 2 is reached. The second point cloud data representing the above is sequentially measured. Every time the second point cloud data is measured at each position in the construction target range R1, the process of collating the second point cloud data with the three-dimensional map 420 in the pipe 2 received from the map acquisition unit 311 is performed. , Self-position information is calculated based on the result of this collation process.

Subsequently, when the mobile robot 11 (as the second mobile robot 11b) comes to the position L shown in FIG. 9, the mobile robot 11 slips in the pipe 2, and the position in the circumferential direction of the mobile robot 11 and / Or assume that the axial position is suddenly displaced. In that case, while the mobile robot 11 (as the second mobile robot 11b) continues to move from the position L, the self-position acquisition unit 310b uses the second point group data and the three-dimensional map 420 that are sequentially measured. The amount of positional deviation of the mobile robot 11 in the circumferential direction and / or the axial direction is calculated by the collation process with and. Then, the self-position acquisition unit 310b corrects the self-position information by using this position deviation amount. The control unit 310a, which has received the position deviation amount from the self-position acquisition unit 310b, issues a control command to the mobile robot 11 (the first) in order to correct the traveling trajectory of the mobile robot 11 so that the position deviation amount becomes zero. 2 It may be transmitted to the mobile robot 11b).

Subsequently, as yet another embodiment for calculating the self-position information, not only the data measured by the second three-dimensional scanner 143 included in the construction robot 10 but also the pipe shape design information is further utilized. An example of calculating the self-position information will be described. The pipe shape design information is three-dimensional CAD data representing the design shape of the pipe 2 generated at the time of designing the pipe 2, and is a self-position acquisition unit from the three-dimensional CAD system 32 in response to an access request from the self-position acquisition unit 310b. It is output to 310b. As shown in FIG. 9, while the mobile robot 11 (second mobile robot 11b) is moving along the arrow w1 in the construction target range R1 of the pipe 2, the second three-dimensional scanner 143 collects the second point cloud data. Each time it is obtained, the self-position acquisition unit 310b calculates the self-position information by collating the second point cloud data with the pipe shape design information.

From the above, according to this embodiment, by collating the second point cloud data with the pipe shape design information, not only the current axial position of the construction robot 10 but also the posture around the pipe axis of the construction robot 10 can be obtained. Can be identified. Therefore, even if the posture of the construction robot 10 changes when the construction robot 10 moves in the pipe 2, self-position information can be acquired in consideration of this, and the pipe shape design information is not used. By comparison, the irradiation area 160 of the laser beam 16 can be specified with high accuracy.

FIG. 10 shows an example of calculating self-position information using not only the data measured by the second three-dimensional scanner 143 included in the construction robot 10 but also the position of the marker formed on the pipe 2. ing. In the embodiment shown in FIG. 10, a welding line (mk11 to mk32 shown in FIG. 10) formed at the joint portion of the pipe 2 is used as a marker, and the welding lines mk11 to mk32 are in the axial direction of the pipe 2. Multiple pieces are arranged in regular positions along the line. Further, in this embodiment shown in FIG. 10, the control terminal 31 is connected to the pipe 2 while the mobile robot 11 (as the second mobile robot 11b) is moving in the construction target range R1 of the pipe 2 in the second work phase. A marker detection unit 310c for detecting the position of the formed marker is further provided. The marker detection unit 310c is connected to a marker detector 144 provided at the bottom of the mobile robot 11. In one embodiment, the marker detector 144 may be an image sensor 144 (FIG. 10) provided below the mobile robot 11 so as to be close to the inner surface of the pipe 2. Then, when the image sensor 144 passes in the vicinity of the marker as the mobile robot 11 moves, a signal indicating that the image sensor 144 has passed the marker is output from the image sensor 144 to the control terminal 31 at that time. Based on the timing of this signal output by the image sensor 144, the control terminal 31 can embed information indicating the marker position in the three-dimensional map 420 and the second point cloud data.

As shown in FIG. 10, after the start of the second work phase, the second three-dimensional scanner 143 moves in the construction target range R1 of the pipe 2 along the arrow w1 by the mobile robot 11 (second mobile robot 11b). To do. Then, the moving second three-dimensional scanner 143 irradiates a scanning signal wave and scans the three-dimensional space in the pipe 2, so that each distance to a plurality of points constituting the space in the pipe 2 is reached. The second point cloud data representing the above is sequentially measured. Further, each time a signal indicating that the marker has passed is obtained from the image sensor 144 connected to the marker detector 310c, the marker detection unit 310c displays the marker on the second point group data received from the self-position acquisition unit 310b. The detection positions of the above are overlapped and returned to the self-position acquisition unit 310b. Then, the marker detector 310c calculates the self-position information by collating the detection position of the marker superimposed on the second point cloud data with the marker position information included in the three-dimensional map 420.

Subsequently, when the mobile robot 11 (as the second mobile robot 11b) comes to the position L1 shown in FIG. 10, the mobile robot 11 slips in the pipe 2 and the circumferential position and / or axis of the mobile robot 11 Suppose that the directional position suddenly shifts. In that case, while the mobile robot 11 (as the second mobile robot 11b) continues to move from the position L1, the self-position acquisition unit 310b is superimposed on the sequentially measured second point group data. The amount of positional deviation of the mobile robot 11 in the axial direction is calculated by collation processing between the marker detection position and the marker position information included in the three-dimensional map 420. Then, the self-position acquisition unit 310b corrects the self-position information by using this position deviation amount. The control unit 310a, which has received the position deviation amount from the self-position acquisition unit 310b, issues a control command to the mobile robot 11 (the first) in order to correct the traveling trajectory of the mobile robot 11 so that the position deviation amount becomes zero. 2 It may be transmitted to the mobile robot 11b).

From the above, according to this embodiment, the detection position of the marker superimposed on the second point cloud data acquired by the second three-dimensional scanner 143, the position information of the marker included in the three-dimensional map 420, and the position information of the marker. By collating, the self-position can be accurately specified in relation to the three-dimensional map 420 with reference to the position of the marker.

Subsequently, as yet another embodiment for calculating the self-position information, not only the data measured by the second three-dimensional scanner 143 included in the construction robot 10 but also the gyro sensor 19a included in the construction robot 10 An example of calculating self-position information by further using the acceleration sensor 19b will be described. Inside the main body 17 of the construction robot 10, as a sensor used to detect the posture position of the construction robot 10 along the circumferential direction of the pipe 2 and the movement amount of the construction robot 10 along the axial direction of the pipe 2. , A gyro sensor and an acceleration sensor 19 are provided (FIG. 2). The gyro sensor 19a outputs the change in the posture of the construction robot 10 to the self-position acquisition unit 310b as a displacement vector represented by a three-dimensional polar coordinate system. Further, the acceleration sensor 19b outputs the position change of the construction robot 10 along the axial direction of the pipe 2 to the self-position acquisition unit 310b as a displacement vector represented by a three-dimensional Cartesian coordinate system. Therefore, by accumulating the displacement vectors sequentially output by the gyro sensor 19a and the acceleration sensor 19b from the start of movement of the construction robot 10, the posture of the construction robot 10 in the pipe 2 and the axial direction of the pipe 2 are obtained. The position along can be calculated. From the above, the self-position acquisition unit 310b can calculate the self-position information by collating the calculated posture and axial position of the construction robot 10 with the second point cloud data.

From the above, according to this embodiment, the gyro sensor 19a and the acceleration sensor 19b are mounted by performing simple arithmetic processing on a small amount of data obtained as a measured value from the gyro sensor 19a and the acceleration sensor 19b. The current self-position of the mobile robot 11 (second mobile robot 11b) can be specified. As a result, according to this embodiment, the self-position of the mobile robot 11 (second mobile robot 11b) can be accurately specified in a short time only by executing a simple arithmetic process. The applications of the gyro sensor 19a and the acceleration sensor 19b are not limited to the above applications, and the surface state sensor 14 scans the surface state in the pipe 2 while moving in the pipe 2 as the first mobile robot 11a. It may be used by the robot 11 as follows. That is, the mobile robot 11 traveling in the pipe 2 as the first mobile robot 11a may travel in the pipe 2 while recognizing its own posture by using the gyro sensor 19a and the acceleration sensor 19b. By doing so, the mobile robot 11 can travel while maintaining its posture in the pipe 2 in the correct direction. Similarly, the mobile robot 11 moving in the pipe 2 as the second mobile robot 11b while performing laser irradiation also has the gyro sensor 19a and the acceleration sensor 19b in order to travel in the pipe 2 while recognizing its own posture. Can be used.

Subsequently, as yet another embodiment for calculating the self-position information, not only the data measured by the second three-dimensional scanner 143 included in the construction robot 10 but also the drive for driving the construction robot 10. An example of calculating self-position information by further utilizing the output value of the encoder 20 attached to the motor will be described. The drive motors 18 and 18'for driving the crawler 13 in the construction robot 10 are equipped with encoders 20 and 20'that count the rotation amounts of the drive motors 18 and 18'and output measurement signals obtained. (Fig. 2). Since the encoder 20 outputs a measurement signal obtained by counting the rotation amounts of the drive motors 18 and 18'for running the mobile robot 11 of the construction robot 10 in the pipe 2, the count value output by the encoder 20 is output. By accumulating, the position of the construction robot 10 when viewed along the axial direction in the pipe 2 can be calculated. From the above, the self-position acquisition unit 310b can calculate the self-position information by collating the calculated axial position of the construction robot 10 with the second point cloud data.

Further, still another embodiment for calculating the self-position information is shown in FIG. In the embodiment shown in FIG. 11A, a plurality of sound sources ss (ss1 and ss2) that emit sound waves sw in response to the ringing command cr from the control unit 310a are provided in the pipe 2. In one example, a plurality of sound sources ss (ss1 and ss2) may be provided at the start end and the end of the construction target range R1 in the pipe 2. Further, in the embodiment shown in FIG. 11, the mobile robot 11 moving in the pipe 2 further includes a sound wave sensor array 25. Here, the sound wave sensor array 25 includes a plurality of sound wave observation means 25 (1) to 25 (M) for observing the sound wave sw, and is configured to output a sound wave signal from the observed sound wave sw. At that time, each of the plurality of sound wave observation means 25 (1) to 25 (M) may be a microphone, and the sound wave sensor array 25 may be a microphone array composed of the plurality of microphones.

On that basis, the self-position acquisition unit 310b that has received the timing T _cr of the ringing command cr from the control unit 310a may be configured to execute the following processing operation. First, to estimate the relative position of the ultrasonic sensor array 25 as viewed from the respective positions of the plurality of sound sources ss based on the sound source signal from the timing T _cr and sonic sensor array 25 of the sounding command cr (ss1 and ss2). Subsequently, the control unit 310a receives the pipe shape design information obtained at the time of shape design of the pipe 2 from the three-dimensional CAD system 32 in the control system 3. Subsequently, the control unit 310a collates the installation positions of the plurality of sound sources ss (ss1 and ss2) shown in the pipe shape design information in the pipe 2 with the relative positions of the sound wave sensor array 25, thereby causing self-position information. Is calculated.

According to the above configuration shown in FIG. 11 (A), even when the moving robot is contaminated by the detached material from the inside of the pipe and the visibility of the optical equipment and the photographing device of the moving robot is obstructed, the inside of the pipe is blocked. It is possible to detect the self-position of the mobile robot by relying on the sound waves from a plurality of sound sources provided in. Similarly, even if the reflected wave detection sensor that detects the reflected wave of the scanning signal wave on the mobile robot side becomes dirty and it becomes difficult to scan the inside of the pipe with the scanning signal wave in three dimensions, the sound wave from the sound source is relied on. It is possible to detect the self-position of the mobile robot. Further, in the above configuration, by appropriately controlling the timing and the type of sound emitted from a plurality of sound sources by a ringing command from the control unit, the self-position of the mobile robot based on the sound waves from the sound sources can be detected accurately. It can be carried out. Further, in this embodiment, not only the sound wave sw from the sound source ss but also the output value of the encoder 20 attached to the drive motor 18 for driving the construction robot 10 is further utilized to make the self-position information more accurate. It may be calculated.

Further, in one exemplary embodiment, in the above configuration shown in FIG. 11A, the self-position acquisition unit 310b includes an acoustic signal space processing unit 310b-1 and a sound source position estimation unit 310b-2, which will be described later. (See FIG. 11B). In the acoustic signal space processing unit 310b-1, the acoustic signal space processing unit receives a sound wave signal which is a multi-channel sound wave signal from the sound wave sensor array 25, and the timing T _cr of the ringing command cr, the multi-channel sound wave signal, and a plurality of sound waves. The sound wave separation process and the sound wave localization process are executed based on the positional relationship between the observation means 25 (1) to 25 (M). Subsequently, the sound source position estimation unit 310b-2 between the respective positions of the plurality of sound source ss (ss1 and ss2) and the position of the sound wave sensor array 25 based on the execution results of the sound source separation process and the sound source localization process described above. Estimate the relative positional relationship of.

In the above configuration shown in FIG. 11A, the multi-channel sound source signals obtained by the plurality of sound wave observation means 25 (1) to 25 (M) in the sound wave sensor array 25 are combined with the timing T _cr of the ringing command cr and a plurality of sound wave command cr. The sound source separation process and the sound source localization process are executed by using the sound wave observation means 25 (1) to 25 (M) in combination with the positional relationship. As a result, an acoustic space capable of expressing the arrival direction of sound waves and the distance to the sound source, as compared with the case where only the multi-channel sound source signals obtained by the plurality of sound wave observation means 25 (1) to 25 (M) are used alone. Information (spatial information of sound) can be calculated accurately. Further, not only multi-channel sound source signal only, by a combination of positional relationship between the timing T _cr, and a plurality of ultrasonic observation means 25 of the sounding command cr (1) ~25 (M) as auxiliary information, the sound source separation Processing and sound source localization processing can be executed easily and efficiently.

In one embodiment, in the above configuration shown in FIG. 11A, the microphone array 25a, which is the sound wave sensor array 25, may have the configuration shown in FIG. As shown in FIG. 12, the arrival direction of the sound wave is represented by an angle θ _sd , which is a relative angle offset when the direction in which the microphones 25 (1) to 25 (M) are arranged is viewed as a reference direction. To. In the embodiment shown in FIG. 12, the plurality of microphones 25 (1) to 25 (M) constituting the microphone array 25a observe sound waves arriving from the direction represented by the angle θ _sd . Subsequently, the microphone array 25a is connected to a plurality of delay filter circuits Dm (1 ≦ m ≦ M) with the sound wave signal obtained by observing the above-mentioned incoming sound wave as x _m (t) (1 ≦ m ≦ M). Output. Subsequently, the plurality of delay filter circuits Dm (1 ≦ m ≦ M) first generate sound wave signals x _m (t) (1 ≦ m ≦ M) from each of the plurality of microphones 25 (1) to 25 (M). A delay amount δ _m (1 ≦ m ≦ M) peculiar to each microphone is added to the microphone. At this time, the delay amount δ _m (1 ≦ m ≦ M) added to the sound wave signals x _m (t) (1 ≦ m ≦ M) is the delay amount for compensating for the phase difference of the incoming sound wave between the microphones. _Therefore , the phase difference of the incoming sound wave between the microphones is determined according to the angle θ _sd . Subsequently, the plurality of delay filter circuits Dm (1 ≦ m ≦ M) apply the sound wave signal x _m (t−δ _m ) (1 ≦ m ≦ M) to which the delay amount is added as described above for a short period at each time. A signal X _m (t−δ _m ) (1 ≦ m ≦ M) converted into a frequency domain representation by Fourier transform is generated. Subsequently, the signal X _m (t−δ _m ) (1 ≦ m ≦ M) is transmitted to the self-position acquisition unit 310b via the communication interface 28 (FIG. 11 (A)), and is transmitted in the self-position acquisition unit 310b. It is output to the acoustic signal space processing unit 310b-1 (FIG. 11B).

In the embodiment shown in FIG. 12, the acoustic signal space processing unit 310b-1 receiving the signal X _m (t−δ _m ) (1 ≦ m ≦ M) as the multiple channel sound source signal X _m (t). Sound source separation processing and sound source localization processing may be executed for −δ _m ) (1 ≦ m ≦ M). Subsequently, in this embodiment, the sound source position estimation unit 310b-2 sets the respective positions of the plurality of sound source ss (ss1 and ss2) and the sound wave sensor array 25 based on the execution results of the sound source separation process and the sound source localization process described above. The relative positional relationship with the position of may be estimated.

In the simplest embodiment of the sound source localization process described above with reference to FIG. 12, from the timing T _cr of the ringing command cr to the time when the sound signal x _m (t) (1 ≦ m ≦ M) is observed on the microphone array 25a. The time difference may be obtained, and the distance from this time difference to the sound source ss along the longitudinal direction of the pipe 2 may be estimated. Further, in the simplest embodiment of the sound source separation process described with reference to FIG. 12, the frequencies of the sound waves emitted by each of the plurality of sound sources ss (ss1 and ss2) are greatly different, and the bandpass filter in the microphone array 25a is used. May be used to discriminate mixed sounds from a plurality of sound sources ss (ss1 and ss2) for each frequency.

Further, in an alternative embodiment of the sound source localization process and the sound source separation process, the acoustic signal space processing unit 310b-1 multiple-channels the signal X _m (t−δ _m ) (1 ≦ m ≦ M) from the microphone array 25a. It may be implemented as a beamformer arithmetic circuit that receives as a sound source signal. In that case, the acoustic signal space processing unit 310b-1 may estimate the sound source direction from the acoustic steering vector obtained by the signal analysis processing for the signal X _m (t−δ _m ) (1 ≦ m ≦ M). .. Subsequently, the acoustic signal space processing unit 310b-1 that estimates the sound source direction, in addition to the sound source direction, has a relative positional relationship between the plurality of microphones 25 (1) to 25 (M) and the timing of the ringing command cr. The position of the sound source may be estimated by further considering T _cr .

Next, the flow of the processing operation executed by the construction system 1 in the pipe for the construction work in the pipe 2 will be described with reference to the flowchart of FIG. First, the processing of the flowchart of FIG. 13 is started from step S11, and the construction robot 10 thrown into the pipe 2 is started to move along the movement path over the entire construction target range R1. Then, while the surface condition sensor 14 is moved by the mobile robot 11 as the first mobile robot 11a, the surface condition sensor 14 scans the entire construction target range R1. As a result of this scanning performed over the entire construction target range R1, the sensing result regarding the surface state in the pipe 2 is output to the arithmetic processing server 33. For example, as a result of scanning the entire construction target range R1 by the first three-dimensional scanner 141, the first point cloud data showing the spatial shape in the pipe 2 is output as the sensing result.

Subsequently, the processing of the flowchart of FIG. 13 proceeds to step S12, and the arithmetic processing server 33 receiving the sensing result (first point cloud data) has a sensing result (first) corresponding to the detection result of the surface state inside the pipe 2. The three-dimensional data 410 is calculated from the point cloud data). The three-dimensional data 410 is three-dimensional data indicating the three-dimensional position coordinates of the inner surface of the pipe 2 in the construction target range R1. Subsequently, the process proceeds to step S13, and the arithmetic processing server 33 specifies the irradiation area 160 to be irradiated with the laser beam 16 from the sensing result (first point cloud data) according to the method described later. At this time, as information on the specified irradiation area 160, the position of the irradiation area 160 in the pipe 2, the range in which the irradiation area 160 extends, the irradiation condition of the laser light when irradiating the laser in the irradiation area 160, and the like are calculated. Will be done. Subsequently, the process proceeds to step S14, and the arithmetic processing server 33 superimposes one or more information on one or more specified irradiation areas 160 on the three-dimensional data 410 to generate the three-dimensional map 420. When the three-dimensional map 420 is transferred to the map acquisition unit 311 in the control unit 310a, the processing of the flowchart of FIG. 13 proceeds to step S15, and the mobile robot 11 as the second mobile robot 11b is the entire construction target range R1. Start moving along the movement path that straddles.

Subsequently, the process proceeds to step S16, and the self-position acquisition unit 310c calculates self-position information representing the self-position of the moving robot 11 moving in the construction target range R1 of the pipe 2 as the second mobile robot 11b. Subsequently, the process proceeds to step S17, and the command generation unit 312 that has received the self-position information from the self-position acquisition unit 310b is irradiated with the self-position of the laser irradiation unit 15 that is moved by the mobile robot 11 (second mobile robot 11b). It is determined whether or not it is included in the area 160. Then, the command generation unit 312 transmits a control command to the laser irradiation unit 15 so as to irradiate the laser beam 16 when it is determined that the self-position is included in the irradiation area 160.

As described above, in step S16 and step S17 of the flowchart of FIG. 13, the control unit 310a may be configured to execute the following processing in cooperation with the self-position acquisition unit 310b. That is, each time the mobile robot 11 is positioned at each of the plurality of positions lined up along the traveling direction of the mobile robot 11 over the construction target range R1, the control unit 310a is based on the detection result of the surface state sensor 14 and the pipe 2 The irradiation area 160 is specified from the inner surface of the robot, and the laser irradiation unit 15 is controlled so that the irradiation area 160 is selectively irradiated with the laser beam 16.

For example, in one embodiment, in step S16 and step S17, the control unit 310a executes the following processing in cooperation with the self-position acquisition unit 310b. That is, while the mobile robot 11 is moving in the pipe 2 in the traveling direction in the construction target range R1 (that is, moving along the longitudinal direction of the pipe 2 at the speed ν shown in FIG. 2), the surface state sensor 14 Based on the detection result, the irradiation area 160 is specified on the inner surface of the pipe 2, and the laser irradiation unit 15 is controlled so that the laser beam 16 is selectively irradiated to the irradiation area 160.

Further, in an alternative embodiment, the control unit 310a may be configured to perform the following operations by cooperating with the self-position acquisition unit 310b in steps S16 and S17. First, each time the mobile robot 11 arrives at each position of a plurality of irradiation areas 160 (160a, 160b, etc. shown in FIG. 4) arranged along the traveling direction of the mobile robot 11 over the construction target range R1, the mobile robot 11 pauses on the spot. The mobile robot 11 is controlled in this way. Subsequently, the laser irradiation unit 15 is controlled so as to irradiate the inner surface of the pipe 2 with the laser at the paused spot, and the mobile robot 11 is moved to the position of the next laser irradiation area 160 after the laser irradiation is completed. Control.

Subsequently, the process proceeds to step S18, and it is determined whether or not the mobile robot 11 (second mobile robot 11b) has completed traveling in the entire construction target range R1, and if it is determined that the traveling has been completed, FIG. End the flowchart. When it is determined that the mobile robot 11 (second mobile robot 11b) has not finished traveling in the entire construction target range R1, the processing of the flowchart of FIG. 13 returns to step S16, and steps S16 and S17 are executed again. Will be done.

In yet another embodiment, as shown in FIG. 14, the mobile robot 11 detects the state of the inner surface of the pipe 2 with respect to the construction target range R1 (first work phase) and the pipe 2 with respect to the construction target range R1. Laser irradiation to the inner surface (second working phase) may be executed by one reciprocating movement over the entire construction target range R1. Specifically, the mobile robot 11 formed by the first mobile robot 11a and the second mobile robot 11b has an outward path w1 that passes the construction target range R1 in the first direction d1 along the longitudinal direction of the pipe 2 and construction. It may be configured to move on a movement path including a return path w2 passing through the target range R1 in the direction d2 opposite to the first direction d1.

In that case, even if the map acquisition unit 311 is configured to acquire the three-dimensional map 420 in which the irradiation area 160 is specified based on the detection result of the surface state sensor 14 while the mobile robot 11 is moving on the outward path w1. Good. Further, after the command generation unit 312 acquires the three-dimensional map 420, the mobile robot 11 starts the movement of the return path w2, and the mobile robot 11 moves the entire construction target range R1 on the return path w2 in three dimensions. Based on the map 420, a command is sent to the laser irradiation unit 15 and the mobile robot 11 as the second mobile robot 11b so that the laser beam 16 is selectively irradiated to the irradiation area 160 in the construction target range R1. It may be configured in.

In yet another embodiment, as shown in FIG. 15, the first mobile robot 11a and the second mobile robot 11b may be formed as separate mobile robots that move the entire construction target range R1. That is, the first mobile robot 11a that moves the surface state sensor 14 in order to complete the scanning of the surface state sensor 14 over the construction target range R1 and the laser irradiation unit 15 that performs laser irradiation in the pipe 2 over the construction target range R1 are moved. The second mobile robot 11b to be made may be formed as a separate mobile robot.

The map acquisition unit 311 is configured to acquire a three-dimensional map 420 in which the irradiation area 160 is specified based on the detection result of the surface state sensor 14 while the first mobile robot 11a moves over the entire construction target range R1. May be done. Further, after the command generation unit 312 acquires the three-dimensional map 420, the second mobile robot 11b starts moving within the construction target range R1, while the second mobile robot 11b moves the entire construction target range R1. To send a command to the laser irradiation unit 15 and the second mobile robot 11b so that the laser beam 16 is selectively irradiated to the irradiation area 160 in the construction target range R1 based on the three-dimensional map 420. It may be configured.

In yet another embodiment, as shown in FIG. 16, the surface state sensor 14 for detecting the state of the inner surface of the pipe 2 moves forward in the traveling direction of the mobile robot 11 that moves along the longitudinal direction of the pipe 2. It is installed. Further, in the embodiment shown in FIG. 16, the laser irradiation unit 15 for irradiating the laser beam 16 toward the inner surface of the pipe 2 moves rearward in the traveling direction of the mobile robot 11 that moves along the longitudinal direction of the pipe 2. It is installed. Therefore, in the configuration shown in FIG. 14, the work stage and the laser irradiation unit 15 for detecting the surface state in the pipe 2 by the surface state sensor 14 while the moving robot 11 in the construction target range R1 is moving in the traveling direction. Even if the work step of irradiating the inner surface of the pipe 2 with the laser is executed at the same time, the surface condition sensor 14 passes through the irradiation area 160, which is the target of the laser irradiation, before the temperature rises due to the laser irradiation. Become. As a result, in the configuration shown in FIG. 14, the work stage of detecting the surface state in the pipe 2 and the work stage of irradiating the inner surface of the pipe 2 with laser are singlely moved without exposing the surface state sensor 14 to high heat. By simultaneously executing the work using the robot 11, the construction work in the pipe 2 can be completed in a short time.

Further, in the embodiment shown in FIG. 16, the control unit 310a may be configured to perform the following operations. That is, each time the mobile robot 11 is positioned at each of the plurality of positions lined up along the traveling direction of the mobile robot 11 over the construction target range R1, the control unit 310a is based on the detection result of the surface state sensor 14 and the pipe 2 The irradiation area 160 is specified from the inner surface of the robot, and the laser irradiation unit 15 is controlled so that the irradiation area 160 is selectively irradiated with the laser beam 16.

For example, in one embodiment, the control unit 310a is moving in the pipe 2 in the traveling direction of the moving robot 11 in the construction target range R1 (that is, along the longitudinal direction of the pipe 2 at the speed ν shown in FIG. 2). (While moving), the irradiation area 160 is specified on the inner surface of the pipe 2 based on the detection result of the surface condition sensor 14, and the laser irradiation unit is such that the laser beam 16 is selectively irradiated to the irradiation area 160. 15 is controlled.

Further, in an alternative embodiment, the control unit 310a may be configured to perform the following operations. First, each time the mobile robot 11 arrives at each position of a plurality of irradiation areas 160 (160a, 160b, etc. shown in FIG. 4) arranged along the traveling direction of the mobile robot 11 over the construction target range R1, the mobile robot 11 pauses on the spot. The mobile robot 11 is controlled in this way. Subsequently, the laser irradiation unit 15 is controlled so as to irradiate the inner surface of the pipe 2 with the laser at the paused spot, and the mobile robot 11 is moved to the position of the next laser irradiation area 160 after the laser irradiation is completed. Control.

As a result, according to the embodiment shown in FIG. 16, by controlling the laser irradiation unit 15 so that the laser light 16 is selectively irradiated to the irradiation area 160, a place in the pipe 2 where laser irradiation is required. It is possible to perform laser irradiation only at. As a result, the construction work in the pipe 2 can be completed in a short time by using the single mobile robot 11, and the wall thickness of the pipe 2 is laser-irradiated in a place where laser irradiation is not required in the pipe 2. This makes it possible to prevent wasteful scraping.

Further, in one exemplary embodiment, the mobile robot 11 may be configured to include a plurality of imaging units 22 shown in FIG. 17 (A). Specifically, a plurality of imaging units 22 (22a to 22e) arranged at different positions along the circumferential direction of the main body 17 of the mobile robot 11 are provided, and each imaging unit 22 (22a to 22e) is provided. , Each has a wide-angle lens. Further, each of the wide-angle lenses possessed by the plurality of imaging units 22 (22a to 22e) has a field of view vw (vw (a) to 22e) of each of the plurality of angular ranges partially overlapping along the circumferential direction in the pipe 2. It may be included in vw (e)). As a result, the plurality of imaging units 22 (22a to 22e) are configured to be able to capture an image over the inner surface of the pipe 2 in the entire circumferential direction.

Further, in this embodiment, the control unit 310a is configured so that a plurality of imaging units 22 (22a to 22e) generate inner surface state information indicating the inner surface state of the pipe 2 from an image taken inside the pipe 2. You may. Then, the control unit 310a is configured to generate a control command to the laser irradiation unit while identifying the position and range of the area in the pipe 2 to be laser-irradiated based on the inner surface state information. May be done.

Therefore, in this embodiment, as shown in FIG. 17B, a plurality of wide-angle lenses arranged in front of the moving robot 11 in the traveling direction while the mobile robot 11 is moving in the pipe 2 are shown in FIG. 17B. Images covering the fields of view vw1 and vw2 can be taken. As shown in FIG. 17B, the fields of view vw1 and vw2 can cover the field of view from the immediate front to the immediate rear of the mobile robot 11 over the entire circumferential direction of the inner surface of the pipe 2.

Further, in still another embodiment, the plurality of imaging units 22 (22a to 22e) may be provided on the mobile robot 11 as functional expansion parts of the surface state sensor 14 shown in FIG. Further, in this embodiment, the plurality of imaging units 22 (22a to 22e) may be provided in front of the mobile robot 11 in the traveling direction together with other components of the surface state sensor 14 shown in FIG. Further, in this embodiment, laser irradiation is performed from the laser irradiation unit 15 arranged behind the traveling direction of the mobile robot 11 as shown in FIG. 16 in parallel with the photographing in the pipe 2 by these wide-angle lenses. Can be done. Therefore, according to the above configuration, the position and range of the area in the pipe 2 to be irradiated with the laser can be determined without having to generate a three-dimensional map in the pipe 2 by scanning the inside of the pipe 2 with the surface state sensor 14. It becomes possible to identify. Further, according to the above configuration, it is possible to identify the position and range of the area in the pipe 2 to be irradiated with the laser without having to detect the self-position of the mobile robot 11.

Next, a method of accurately detecting the surface state in the pipe 2 to be detected by the surface state sensor 14 will be described with reference to FIG. Such detection of the surface state is necessary not only when specifying which location on the three-dimensional data 410 as the irradiation area 160, but also the laser beam 16 is detected by the control unit 310a according to the surface state in the pipe 2. It is also required when controlling the irradiation conditions. That is, depending on the surface condition of the inner surface of the pipe 2, whether or not the construction by laser irradiation is necessary and how much laser output and how long the laser irradiation should be performed will change. .. Therefore, in order to specify a place where it is determined that construction by laser irradiation is necessary on the three-dimensional data 410 measured by the surface state sensor 14 (first three-dimensional scanner 141) as an irradiation area, the surface state is high. Accurate detection is required. Further, in order to determine the specified irradiation area 160, how much laser output and how long the laser irradiation should be performed, highly accurate detection of the surface state is required.

Further, in this embodiment, the surface state in the pipe 2 to be detected by the surface state sensor 14 is, for example, the adhesion area of the oxide scale, the degree of erosion generated on the inner surface of the pipe 2 by the oxide scale, and oxidation. The shape state of the object scale layer (whether the surface shape of the scale layer is rolled into flakes, etc.), the floating of the coating film, the thickness of the coating film, etc. It is difficult to accurately detect the surface state as described above only by analyzing the first point cloud data measured by the three-dimensional scanner 140 as the spatial measurement information representing the spatial shape inside the pipe 2. Therefore, in this embodiment, a mechanism is disclosed in which the image pickup device 142 included in the surface state sensor 14 detects the surface state in the pipe 2 based on an image obtained by photographing the inner surface of the pipe 2.

Specifically, in this embodiment, the control unit 310a generates inner surface state information indicating the inner surface state in the irradiation area 160 indicated by the three-dimensional map 420 by image recognition from the image Im obtained by photographing the inner surface of the pipe 2. An image processing unit 313 is further provided, and the command generation unit 312 that has received the inner surface state information from the image processing unit 313 generates a control command to be transmitted to the laser irradiation unit 15 based on the inner surface state information. To do. In one exemplary embodiment, the image processing unit 313 may generate inner surface state information by image recognition from an image Im obtained by photographing the inner surface of the pipe 2 as follows. That is, the image processing unit 313 may include the neural network 500 shown in FIG. Then, the image processing unit 313 inputs the image Im into the neural network 500 shown in FIG. 18 and processes it to execute the image recognition process, and generates the inner surface state information from the output of the neural network 500. You may.

As described above, for the purpose of controlling the irradiation condition of the laser beam 16 by the control unit 310a according to the surface condition inside the pipe 2, the pipe 2 is a mechanism for accurately detecting the surface condition of the inner surface of the pipe 2. The reason why the image recognition process for the image obtained by photographing the inner surface of the light is preferable is mainly for the following reasons. Conventionally, when the construction work on the base material surface is manually performed by irradiating the laser beam, the laser beam irradiation conditions for each construction site are determined based on the surface condition of the base metal surface visually visually recognized by the operator. By making adjustments, appropriate construction was performed according to the surface condition of the base metal. For example, the operator adjusts the intensity of the laser beam output to each construction site or adjusts the sweep speed of the laser irradiation based on the surface condition of the surface of the base material visually recognized by the naked eye. Therefore, in this embodiment, an image of the inner surface of the pipe 2 taken by the image pickup apparatus 142 is acquired as information that can be mechanically recognized and processed corresponding to the surface state of the surface of the base material visually visually recognized, and the image is taken from the image. By image recognition, the inner surface state information indicating the inner surface state in the irradiation area 160 in the pipe 2 is generated. As a result, according to this embodiment, by generating a command to the laser irradiation unit 15 based on the inner surface state information, the laser irradiation condition according to the surface state of the base material, which has been manually performed conventionally. Adjustment operations can be automated for construction work in the piping.

Further, the image processing unit 313 included in the control unit 310a is input from the outside when generating the inner surface state information about the irradiation area 160 shown by the three-dimensional map 420 from the image Im obtained by photographing the inner surface of the pipe 2. A machine learning process that improves the accuracy of image recognition according to the teacher signal Tc may be executed, and the learning result obtained by executing the machine learning process may be used to perform image recognition. In one embodiment shown in FIG. 18, the image processing unit 313 includes a neural network 500 and a feedback signal generation unit 600 used for image recognition of an image Im obtained by photographing the inner surface of the pipe 2, and is an image processing unit. The 313 is connected to the display device Dp. Then, the image processing unit 313 may improve the accuracy of image recognition by causing the neural network 500 to execute the machine learning process described later with reference to FIG.

In this configuration, the image recognition for generating the inner surface state information indicating the inner surface state in the irradiation area 160 in the pipe 2 from the image Im obtained by capturing the irradiation area 160 is the learning result obtained by executing the machine learning process. Is executed using. At that time, this machine learning process is executed so as to improve the accuracy of the image recognition according to the teacher signal Tc input from the outside. Therefore, according to this configuration, the image processing unit 313 as the image recognition execution subject is trained by the machine learning process executed according to the teacher signal Tc so that the above-mentioned inner surface state can be recognized with the required recognition accuracy. The accuracy of image recognition can be improved.

Hereinafter, a method of executing a machine learning process on the neural network 500 used for image recognition of the image Im obtained by photographing the inner surface of the pipe 2 and improving the image recognition accuracy by the neural network 500 in the image processing unit 313. Will be described. In the embodiment shown in FIG. 18, an image Im obtained by photographing the inner surface of the pipe 2 is acquired as a sample image for learning, and the data (FIG. 15) of the sample image Im for learning is transferred to the neural network 500 shown in FIG. It is the initial input of. Then, in the embodiment shown in FIGS. 15 and 16, a machine learning process for improving the accuracy of image recognition is executed by executing deep learning by the neural network 500. After the machine learning process is completed, the image processing unit 313 is configured to perform image recognition using the learning result obtained by executing this deep learning.

This deep learning is executed as follows. Is there an error in the neuron output output by the output layer of the neural network 500 after the data of the image Im obtained by photographing the inner surface of the pipe 2 as a sample image for learning is input as the initial input to the neural network 500? Judge whether or not. Then, when an error is included, a feedback signal for correcting the error is fed back to the neural network 500 according to a teacher signal Tc input from the outside by a skilled worker of the construction work in the pipe 2. Subsequently, the image Im data is input again to the neural network 500 after the error is corrected by the feedback signal, and the same processing as described above is repeated.

The process of executing this deep learning will be described in more detail as follows. First, as a sample image for learning, the image Im captured by the image pickup device 142 on the inner surface of the pipe 2 is read from the image pickup device 142 into the image processing unit 313. Then, the plurality of image blocks included in the image Im are encoded to generate the input data iS1 to iS8.

Subsequently, in order to cause the multi-layered neural network 500 shown in FIG. 18 to execute deep learning, input data iS1 to iS8 are input to each neuron nn constituting the input layer 510 of the neural network 500. Subsequently, the output signals from each neuron nn constituting the input layer 510 are processed by the inner layers 520, 530 and 540, and are output from the output layer as neuron outputs OS1 to OS4. The neuron outputs OS1 to OS4 are signals representing irradiation conditions of the laser beam 16 to be instructed to the laser irradiation unit 15 when the surface state of the inner surface of the pipe 2 is the surface state represented by the image Im. At that time, the irradiation condition defines whether or not the laser irradiation should be performed on the portion represented by the image Im of the inner surface of the pipe 2, and how much laser output and how long the laser irradiation should be performed. It is a condition to do.

Subsequently, a worker skilled in the work of irradiating the surface of the base material with a laser beam to construct the surface of the base material visually visually recognizes the image Im displayed on the display screen Dp connected to the image processing unit 313. The irradiation conditions of the laser beam 16 output as the neuron output OS1 to OS4 are read. Subsequently, the above-mentioned skilled worker determines whether or not the irradiation conditions of the laser light 16 output as the neuron output OS1 to OS4 are correct based on the result of visually recognizing the image Im displayed on the display screen. For example, a skilled worker may determine whether or not the irradiation conditions (laser irradiation output and laser irradiation time) of the laser beam 16 represented by the neuron outputs OS1 to OS4 are appropriate from the visual results of the image Im displayed on the display screen Dp. To judge.

When the irradiation conditions of the laser beams 16 output as the neuron outputs OS1 to OS4 are incorrect, the teacher signal Tc for correcting the error in the irradiation conditions is input to the feedback signal generator 600 provided in the image processing unit 313. To do. For example, a skilled worker may find that the irradiation conditions (laser irradiation output and laser irradiation time) of the laser beam 16 represented by the neuron outputs OS1 to OS4 are insufficient from the visual results of the image Im displayed on the display screen Dp. Suppose you decide. In that case, the skilled worker inputs the teacher signal Tc corresponding to the irradiation conditions modified so that the laser irradiation output and the laser irradiation time are sufficient to the feedback signal generator 600.

Subsequently, the feedback signal generation unit 600 generates feedback signals Fd1, Fd2, and Fd3 for correcting the error of the neuron outputs OS1 to OS4 according to the teacher signal Tc, and generates feedback signals Fd1, Fd2, and Fd3 for the inner layers 520, 530, and 540 of the neural network 500. And input the feedback signal. When the feedback signals Fd1, Fd2 and Fd3 are input to the inner layers 520, 530 and 540, the binding strength coefficient stored in each of the plurality of neurons constituting the inner layers 520, 530 and 540 is the feedback signal Fd1. , Fd2 and Fd3.

Subsequently, the input data iS1 to iS8 are input to each neuron nn constituting the input layer 510. Subsequently, the output signals from each neuron nn constituting the input layer 510 are processed by the inner layers 520, 530 and 540, and are output from the output layer as neuron outputs OS1 to OS4. Then, whether or not the irradiation conditions of the laser beams 16 output as the neuron outputs OS1 to OS4 are correct is determined again by a skilled worker who is viewing the image Im. Then, when it is determined that the irradiation conditions of the laser beams 16 output as the neuron outputs OS1 to OS4 are incorrect, the same processing procedure as described above is repeated, whereby the inner layers 520, 530 and 540 are separated. The value of the connection strength coefficient of each constituent neuron gradually converges to the correct value. On the contrary, when the skilled worker determines that the irradiation conditions of the laser light 16 output as the neuron output OS1 to OS4 are correct, the irradiation conditions of the laser light 16 output as the neuron output OS1 to OS4 are supported. Surface state information is generated and output to the command generation unit 312.

In this way, by applying the feedback signals Fd1, Fd2, and Fd3 to each of the plurality of neurons constituting the inner layers 520, 530, and 540, respectively, the binding strength coefficient in these neurons approaches the optimum value. Is corrected to. Then, each time the feedback signals Fd1, Fd2 and Fd3 are applied to the inner layers 520, 530 and 540 based on the neuron outputs OS1 to OS4 obtained by inputting the input data iS1 to iS8, the above-mentioned correction of the coupling strength coefficient is performed. By repeating this, the machine learning process progresses. At that time, the machine learning processes that proceed in the inner layers 520, 530, and 540, respectively, may correspond to learning processes that learn different learning objects.

For example, in one exemplary embodiment, the machine learning processes that proceed in the inner layers 520, 530, and 540, respectively, may correspond to any one or more of the following learning processes g1 to g3.
(G1) A learning process for correctly recognizing the boundary between a rusted region (a region in which an oxide scale layer is formed on the inner surface) and a normal region without rust on the inner surface of the pipe 2.
(G2) A learning process for correctly recognizing the difference in surface color between a rusted area and a normal area.
(G3) A learning process for correctly recognizing the difference in surface shape (whether the surface is flaky or rough due to minute irregularities on the surface, etc.) between the rusted region and the normal region.

As described above, in the embodiment shown in FIG. 18, in order to improve the accuracy of the image recognition described above, the neural network in which the error of the neuron output is corrected by the teacher signal Tc as a machine learning process executed according to the teacher signal Tc. Perform deep learning using. Therefore, according to this embodiment, in the machine learning process for improving the accuracy of the image recognition described above, the feature amount is obtained from the image obtained by photographing the inner surface of the pipe according to an engineering method such as spectroscopic spectrum analysis or color information analysis. No need to extract. Therefore, according to this embodiment, when the inner surface state information is generated from the image Im obtained by capturing the irradiation area 160, it is necessary to perform image recognition in which it is difficult to extract the feature amount from the image based on the engineering method. Even in this case, it is possible to easily realize image recognition in which the inner surface state can be recognized with the required recognition accuracy.

In one exemplary embodiment, the construction system 1 in the pipe is applied to remove rust generated on the inner surface of the TCA pipe that circulates cooling air for cooling the rotor of a large gas turbine. It is possible. Here, the TCA pipe is a pipe for cooling air for cooling the air discharged from the rotor of the large gas turbine and circulating it in the vehicle interior of the large gas turbine for cooling the rotor.

1 Construction system in the pipe 2 Pipe 3 Control system 10 Construction robot 11 Mobile robot 11a First mobile robot 11b Second mobile robot 12 (12a, 12b, 12c, 12') Arm 13 (13a, 13b, 13c, 13') ) Crawler 14 Surface condition sensor 15 Laser irradiation unit 15a Reflection mirror 15c Nozzle unit 15d Ring motor unit 15b Laser emission port 15e Laser light generator 16 Laser light 17 Main body 17a, 17b End face 18 Drive motor 19 Gyro sensor and Acceleration sensor
20 Encoder 22 (22a to 22e) Imaging unit 25 Sound wave sensor array 25a Microphone array 25 (1) to 25 (M) Sound wave observation means (microphone)
28 Communication interface 31 Control terminal 32 3D CAD system 33 Arithmetic processing server 34 Communication network 35 Hub 140 (141,143) 3D scanner 140a Irradiation port 140b Reflected wave detection sensor 142 Imaging device 142a Lens unit 144 Image sensor 160 Irradiation area 310a Control unit 310b Self-position acquisition unit 310c Marker detection unit 311 Map acquisition unit 312 Command generation unit 313 Image processing unit 500 Neural network 510 Input layer 520 Internal layer 600 Feedback signal generation unit vw (vw (a) to vw (e)) Wide angle Lens field C1 Power cable C2 Communication cable

Claims (24)

A surface condition sensor that detects the state of the inner surface by scanning the inner surface of the pipe over a preset construction target range in the longitudinal direction of the pipe.
A first mobile robot configured to move the surface condition sensor in the longitudinal direction inside the pipe in a construction target range in the longitudinal direction of the pipe.
A laser irradiation unit that irradiates the inner surface of the pipe with a laser beam,
A second mobile robot configured to move the laser irradiation unit in the longitudinal direction inside the pipe in the construction target range.
A control unit configured to control the laser irradiation unit and the second mobile robot is provided.
The control unit
Based on the detection result of the surface condition sensor for the construction target range in which the scanning of the surface condition sensor is completed, three-dimensional data indicating the position coordinates of the inner surface of the pipe in the construction target range is acquired. , A map acquisition unit that acquires a three-dimensional map in which the irradiation area of the laser beam is specified on the three-dimensional data.
Based on the three-dimensional map, a command generation unit that sends a command to the laser irradiation unit and the second mobile robot so that the laser light is selectively irradiated to the irradiation area within the construction target range. ,
In-piping construction system by mobile robot including. The map acquisition unit is based on the detection result of the surface condition sensor for the construction target range in which the scanning of the surface condition sensor is completed.
Three-dimensional data representing the spatial shape of the inner surface of the pipe in the construction target range is calculated.
It is characterized in that one or more data representing the positions and ranges of one or more irradiation areas to be irradiated with the laser beam are superimposed on the three-dimensional data to acquire a three-dimensional map. The in-pipe construction system using the mobile robot according to claim 1. The first mobile robot and the second mobile robot form a single mobile robot, and the mobile robot has an outward path extending the construction target range in the first direction along the longitudinal direction of the pipe, and the construction target. The range is configured to move with the return path extending in the direction opposite to the first direction.
The map acquisition unit is configured to acquire the three-dimensional map in which the irradiation area is specified based on the detection result of the surface state sensor while the mobile robot moves on the outbound route.
After acquiring the three-dimensional map, the command generation unit displays the three-dimensional map while the mobile robot starts moving on the return route and the mobile robot moves the construction target range on the return route. Based on this, it is configured to send a command to the laser irradiation unit and the mobile robot as the second mobile robot so that the laser beam is selectively irradiated to the irradiation area within the construction target range. The in-pipe construction system using a mobile robot according to claim 1 or 2, wherein the robot is characterized in that. The first mobile robot and the second mobile robot are formed as separate mobile robots.
The map acquisition unit is configured so that the first mobile robot acquires the three-dimensional map by moving the construction target range.
After acquiring the three-dimensional map, the command generation unit selectively selects the laser with respect to the irradiation area based on the three-dimensional map while the second mobile robot moves in the construction target range. The in-pipe construction system by a mobile robot according to claim 1 or 2, wherein a command is sent to the laser irradiation unit and the second mobile robot so that light is irradiated. The laser irradiation unit includes a nozzle unit that is rotatably configured along the circumferential direction in the pipe so that the irradiation direction of the laser light rotates.
The control unit is the second mobile robot along the longitudinal direction of the pipe at each position in the pipe along the longitudinal direction according to the angle range occupied by the irradiation area in the circumferential direction of the pipe. The in-pipe construction system by a mobile robot according to any one of claims 1 to 4, characterized in that it is configured to control the traveling speed of the above. Any of claims 1 to 5, wherein the control unit is configured to control the irradiation conditions of the laser beam based on the surface condition in the pipe detected by the surface condition sensor. The in-pipe construction system using the mobile robot described in item 1. The first mobile robot is configured to be movable in the pipe within the construction target range, and by irradiating the scanning signal wave and scanning the three-dimensional space in the pipe, from the irradiation source of the scanning signal wave. Further provided with a first three-dimensional scanner that acquires first point group data representing the distances and directions to a plurality of points constituting the space in the pipe as spatial measurement information representing the space shape in the pipe.
The control unit according to any one of claims 1 to 6, wherein the control unit is configured to acquire the three-dimensional map generated based on the first point cloud data. In-pipe construction system using a mobile robot. Further, a self-position acquisition unit for acquiring self-position information representing the self-position of the second mobile robot in the pipe is provided.
The control unit determines whether or not the self-position based on the self-position information acquired by the self-position acquisition unit is included in the irradiation area on the three-dimensional map, and the self-position is within the irradiation area. The piping by a mobile robot according to any one of claims 1 to 7, wherein the laser irradiation unit is configured to control the laser irradiation unit so as to irradiate the laser beam when included in the above. Internal construction system. The second mobile robot is configured to be movable in the pipe within the construction target range, and by irradiating a scanning signal wave and scanning the three-dimensional space in the pipe, the space in the pipe is configured. Further equipped with a second three-dimensional scanner that acquires second point group data representing the respective distances to a plurality of points as spatial measurement information representing the spatial shape in the pipe.
The self-position information is calculated by collating the second point cloud data acquired by the second three-dimensional scanner with the pipe shape design information obtained at the time of designing the shape of the pipe. , The in-pipe construction system using the mobile robot according to claim 8. The three-dimensional map includes position information of a marker formed on the pipe in the construction target range.
A marker detector that detects the marker and
The second mobile robot is configured to be movable in the pipe within the construction target range, and the space in the pipe is configured by irradiating a scanning signal wave and scanning the three-dimensional space in the pipe. Further provided with a second three-dimensional scanner that acquires second point cloud data representing the respective distances to a plurality of points as spatial measurement information representing the spatial shape in the pipe.
The self-position information is calculated by collating the detection position of the marker superimposed on the second point cloud data with the position information of the marker included in the three-dimensional map. The in-pipe construction system using the mobile robot according to claim 8, which is characterized. Further provided with a gyro sensor and an acceleration sensor configured to be movable in the pipe within the construction target range by the second mobile robot.
The in-pipe construction system by a mobile robot according to claim 8, wherein the self-position information is calculated using the measurement results of at least one of the gyro sensor and the acceleration sensor. A plurality of sound sources that emit sound waves in response to a ringing command from the control unit are provided.
The mobile robot, which is at least one of the first mobile robot and the second mobile robot, further includes a sound wave sensor array.
The sound wave sensor array includes two or more sound wave observation means for observing the sound wave, and is configured to output a sound wave signal from the observed sound wave.
The self-position acquisition unit that has received the timing of the ringing command from the control unit
Based on the timing of the ringing command and the sound source signal from the sound wave sensor array, the relative position of the sound wave sensor array as seen from each position of the plurality of sound sources is estimated.
The self-position information is calculated by collating the installation positions of the plurality of sound sources with the relative positions of the sound wave sensor array.
The in-pipe construction system using a mobile robot according to claim 8, further comprising the above method. The self-position acquisition unit
The sound source signal, which is a multi-channel sound source signal, is received from the sound wave sensor array, and the sound source separation process and the sound source are performed based on the timing of the ringing command, the positional relationship between the multi-channel sound source signal and the two or more sound wave observation means. An acoustic signal space processing unit that executes localization processing,
A sound source position estimation unit that estimates the relative positional relationship between each position of the sound source and the position of the sound wave sensor array based on the execution results of the sound source separation process and the sound source localization process.
12. The in-pipe construction system using a mobile robot according to claim 12. The surface state sensor includes an imaging device that photographs the inner surface of the pipe and outputs the captured image to the control unit.
The control unit
From the image, image recognition is used to generate inner surface state information indicating the inner surface state in the irradiation area indicated by the three-dimensional map.
The piping by a mobile robot according to any one of claims 1 to 13, characterized in that it is configured to generate a command to the laser irradiation unit based on the inner surface state information. Internal construction system. When the control unit generates the inner surface state information about the irradiation area indicated by the three-dimensional map from the image, the control unit
A machine learning process that improves the accuracy of the image recognition is executed according to the teacher signal input from the outside.
The in-pipe construction system by a mobile robot according to claim 14, wherein the image recognition is executed by using the learning result obtained by executing the machine learning process. When the control unit generates the inner surface state information about the irradiation area indicated by the three-dimensional map from the image, the control unit
The machine learning process is executed by performing deep learning using a neural network in which the image is used as the initial input and the error of the neuron output is corrected by the teacher signal.
The in-pipe construction system by a mobile robot according to claim 15, wherein the image recognition is executed by using the learning result obtained by executing the deep learning. A mobile robot configured to move in the pipe along the longitudinal direction of the pipe, and
A surface condition sensor mounted on the mobile robot in front of the mobile robot in the traveling direction and detecting the state of the inner surface of the pipe,
A laser irradiation unit mounted on the mobile robot behind the surface state sensor and irradiating the inner surface of the pipe with a laser beam.
The laser irradiation unit and the control unit configured to control the mobile robot are provided.
The laser irradiation unit
A laser light generator built in the main body of the mobile robot,
A laser irradiation port located behind the main body in the traveling direction and for emitting laser light from the laser light generator.
Including
The control unit
Each time the mobile robot is positioned at each of one or more positions arranged along the traveling direction of the mobile robot over the construction target range, the laser irradiation unit is controlled based on the detection result of the surface state sensor. An in-pipe construction system using a mobile robot configured in. A mobile robot configured to move in the pipe along the longitudinal direction of the pipe, and
A surface condition sensor mounted on the mobile robot in front of the mobile robot in the traveling direction and detecting the state of the inner surface of the pipe,
A laser irradiation unit mounted on the mobile robot behind the surface state sensor and irradiating the inner surface of the pipe with a laser beam.
The laser irradiation unit and the control unit configured to control the mobile robot are provided.
The control unit
Each time the mobile robot is positioned at each of one or more positions arranged along the traveling direction of the mobile robot over the construction target range, the laser irradiation unit is controlled based on the detection result of the surface state sensor.
Is configured as
The mobile robot includes a main body portion formed in an elongated shape extending along the longitudinal direction of the pipe.
A plurality of the surface state sensors are arranged at different positions in the circumferential direction of the main body portion, and include a plurality of imaging units each having a wide-angle lens.
Each of the wide-angle lenses possessed by the plurality of imaging units includes each of a plurality of angular ranges partially overlapping along the circumferential direction in the pipe in the field of view, thereby covering the entire inner surface of the pipe. It is configured to be able to take images over the direction,
The control unit
The plurality of imaging units generate inner surface state information indicating the inner surface state of the pipe from images taken inside the pipe.
Based on the inner surface condition information, transfer you, characterized in that it is configured to generate a control command for the position and extent of the area of the inner pipe to be subjected to laser irradiation to the laser irradiation unit with identification In-pipe construction system using a moving robot. The control unit is moving in the pipe in the traveling direction of the mobile robot in the construction target range.
Based on the detection result of the surface condition sensor, the irradiation area on the inner surface of the pipe is specified.
The in-pipe construction system by a mobile robot according to claim 18, wherein the laser irradiation unit is configured to be configured to selectively irradiate the irradiation area with the laser beam. The mobile robot
A main body formed in a long shape extending along the longitudinal direction of the pipe, and
A plurality of crawler-type traveling bodies arranged at different positions in the circumferential direction of the main body portion and configured to be able to travel on the inner surface of the pipe.
A plurality of telescopic mechanisms are arranged at different positions in the circumferential direction of the main body portion, and each has a telescopic mechanism configured to support the crawler type traveling body with respect to the main body portion so as to be expandable and contractible along the radial direction of the pipe. Arms and
The in-pipe construction system by a mobile robot according to any one of claims 1 to 19, wherein the system is configured to include. The expansion / contraction mechanism of each of the plurality of arms is
A spring member that expands and contracts so as to press the crawler type traveling body against the inner surface of the pipe by a spring urged along the radial direction of the pipe.
A cylinder device configured to change the amount of expansion and contraction of the arm portion with respect to the main body portion by the expansion and contraction operation along the radial direction of the pipe.
An expansion / contraction amount adjusting unit configured to control the expansion / contraction amount of the arm portion by driving the cylinder device in response to an expansion / contraction instruction signal received from the control unit.
The in-pipe construction system by a mobile robot according to claim 20, wherein the mobile robot is provided. The control unit adaptively selects the expansion / contraction instruction signal given to the expansion / contraction amount adjusting unit according to the shape of the inner surface of the pipe, so that the crawler type traveling body has the crawler type traveling body relative to the inner surface of the pipe. The in-pipe construction system using a mobile robot according to claim 21, further configured to control the cylinder device so as to secure a ground contact force required for traveling. A step of detecting the state of the inner surface of the pipe while moving the surface condition sensor in the longitudinal direction inside the pipe in the construction target range in the longitudinal direction of the pipe.
On the scanning of the surface condition sensor is based on the detection result of the surface condition sensor for the working target range was complete, acquired three-dimensional data indicating the position coordinates of the inner surface of the pipe before SL Construction Scope Then, the step of acquiring the three-dimensional map in which the irradiation area of the laser beam is specified on the three-dimensional data, and
By controlling the laser irradiation unit at each of one or more positions in the pipe along the longitudinal direction over the construction target range, the irradiation of the laser light toward the inner surface of the pipe is controlled. Steps to do and
With
In the step of controlling the irradiation of the laser beam, the laser irradiation unit is controlled so that the laser beam is selectively irradiated to the irradiation area within the construction target range based on the three-dimensional map. An in-pipe construction method characterized by this. A mobile robot, a surface state sensor mounted on the mobile robot in front of the mobile robot in the traveling direction, and a laser irradiation unit mounted on the mobile robot behind the surface state sensor in the traveling direction. Steps to put the construction robot including, into the pipe,
A step of moving the construction robot in the traveling direction in the piping,
A step of detecting the state of the inner surface of the pipe by the surface state sensor while the construction robot is moving in the traveling direction in the construction target range in the longitudinal direction of the pipe.
By controlling the laser irradiation unit based on the detection result of the surface state sensor each time the construction robot is positioned at each of one or more positions arranged along the traveling direction over the construction target range. A step of controlling the irradiation of laser light from the laser irradiation unit toward the inner surface of the pipe, and
Equipped with a,
The laser irradiation unit
A laser light generator built in the main body of the mobile robot,
A laser irradiation port located behind the main body in the traveling direction and for emitting laser light from the laser light generator.
In- pipe construction method characterized by including <br />.

Images