Unmanned work system

Abstract

To provide an unmanned work system capable of performing predetermined work evenly in a work area with a simple configuration.SOLUTION: An unmanned work system to perform a predetermined work while moving a mobile body 1 around a work area A, gradually turning from the outside to the inside, acquires as a pixel count the height of a marker M in the image acquired by an imaging unit 2 in the forward and stop processing of the mobile body 1, compares the pixel count with a movement stop reference value which is a reference value for stopping the movement of the mobile body 1, moves the mobile body 1 forward if the pixel count<the movement stop reference value, stops the mobile body 1 if the pixel count≥the movement stop reference value, and further performs a stopping position correction process and a moving direction correction process every time the mobile body 1 turns.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to an unmanned operation system that autonomously moves a mobile object within a work area defined by a plurality of markers to perform a predetermined operation. [Background technology]

[0002] Conventionally, systems disclosed in Patent Documents 1 and 2 have been proposed as unmanned work systems in which a mobile object moves autonomously within a work area partitioned by a plurality of markers to perform a predetermined work.

[0003] The system disclosed in Patent Document 1 (hereinafter referred to as "Prior Art 1") comprises a number of markers placed at each corner of a work area and a lawnmower that moves within the work area. The markers are equipped with electromagnetic wave emitting means that emit long-wavelength electromagnetic waves, and the lawnmower receives the electromagnetic waves emitted from each electromagnetic wave emitting means with a rotating antenna and is equipped with a calculation means that determines the position of the lawnmower from the angle between the lawnmower's reference line and each marker. The system is characterized by its ability to recognize the traveling area within the work area with high accuracy, without being affected by weather or obstacles.

[0004] Furthermore, the system disclosed in Patent Document 2 (hereinafter referred to as "Prior Art 2") comprises an autonomous mobile robot equipped with a detection device that detects two signs placed ahead in the direction of travel, and a target path setting device that calculates the coordinates of the two signs with the current position as the origin based on the detection results of the detection device, and sets a target path for the autonomous mobile robot from these coordinates; this system is characterized in that by simply placing signs, the autonomous mobile robot will perform specified tasks while moving autonomously in a work area surrounded by these signs. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] JP 2014-149682 A [Patent Document 2] Patent Publication No. 2021-170287 Summary of the Invention [Problem to be solved by the invention]

[0006] However, the markers used in the system of Prior Art Example 1 are equipped with an electromagnetic wave emitting means, and the markers used in the system of Prior Art Example 2 are equipped with a distance measurement unit and a direction measurement unit. Since both systems use dedicated markers, the equipment costs are high and the markers are not easy to obtain, so that it is not possible to simply add more markers and re-zone the work area.

[0007] The present invention has been developed in consideration of the problems associated with the conventional examples, and aims to provide an unmanned work system that does not require dedicated markers, but can control the autonomous movement of a mobile body within a work area by using inexpensive, easily available markers, such as commercially available pylons (triangular cones), and can perform specified tasks evenly throughout the work area. [Means for solving the problem]

[0008] The gist of the present invention will now be described with reference to the accompanying drawings.

[0009] This unmanned work system autonomously moves a moving body 1 within a work area A partitioned by placing a plurality of markers M at each corner to perform a specified task, wherein the moving body 1 is equipped with an imaging unit 2 that images the area ahead, and a control unit that controls the movement of the moving body 1 based on images acquired by the imaging unit 2, and the control unit is configured to execute the processes a to e below, and the moving body 1 is configured to move in a circular motion from the outside to the inside of the work area A while repeatedly moving forward and turning. Note a marker recognition process for recognizing a marker M in the image acquired by the imaging unit 2 based on a trained model of object detection generated by machine learning; b. The height of the marker M recognized in the image is acquired in the number of pixels, and a value for stopping the moving object 1 in front of the marker M is obtained by multiplying the number of vertical pixels of the entire image acquired by the imaging unit 2 by a predetermined coefficient. character and a movement stop reference value obtained by comparing the number of pixels with the movement stop reference value obtained by the above-mentioned method. If the number of pixels is smaller than the movement stop reference value, the moving body 1 is moved forward, and if the number of pixels is equal to or greater than the movement stop reference value, the moving body 1 is stopped. c) When the moving body 1 stops, the moving body 1 is rotated in a predetermined direction to search for the marker M that is the next moving destination, and if it is found, The next destination A turning process for turning the moving body 1 in the direction of the marker M d. Each time the moving body 1 turns, the movement stop reference value in the forward / stop processing is subtracted, and the moving body 1 after the turn is moved to the destination. of Previous destination for marker M of Stop position correction process to stop the vehicle before the stop position for marker M e. Every time the moving body 1 turns, the moving direction of the moving body 1 is changed to a destination in the moving direction. of Movement direction correction process to correct the direction of movement to the inside by a specified amount relative to marker M

[0010] In the unmanned operation system according to claim 1, the movement stop reference value is calculated by the following formula (1), and the movement direction correction process calculates a correction angle by using the following formula (2): and correcting the moving direction of the marker M to the inside by the correction angle. The present invention relates to an unmanned operation system characterized by the above-mentioned. Note Equation (1) Movement stop reference value = marker height + (total image height - marker height) × (1-percentage of converted) Here, the turning ratio is (number of turns performed) / (target number of turns), and the target number of turns is the number of turns required for the moving body 1, set in the initialization process, to turn in the working area A. The marker height is the number of pixels in the image that the height of the marker M found as the next destination in the turning process occupies, and the total image height is the number of vertical pixels of the entire image in which the found marker M is captured. Equation (2) Correction angle = (maximum correction angle) × (Percentage of people converted) Here, the maximum correction angle is a preset value (angle), and the redirected ratio is the same as the formula (1).

[0011] The present invention relates to an unmanned operation system, characterized in that in the initial stage of the operation, the movement stop reference value is calculated by the following formula (1') instead of the formula (1). Note Equation (1') Movement stop reference value = marker height + (total image height - marker height) × ((1-converted rate)^2) Here, the converted ratio, the marker height, and the total image height are the same as in equation (1).

[0012] Also, claims 2, In the unmanned work system described in No. 3, the control unit estimates the length of the longest side of the work area A, and executes a target number of turns setting process to set the number of turns required for the rotating movement of the moving body 1 in the work area A as the target number of turns based on this length of the longest side, and when the number of turns of the moving body 1 reaches the target number of turns, the unmanned work system is configured to terminate the movement of the moving body 1.

[0013] Furthermore, in the unmanned operation system described in claim 4, the control unit executes a left turning movement process in which the moving body 1 turns left while performing the specified work, and a right turning movement process in which the moving body 1 turns right while performing the specified work, and when the number of turns of the moving body 1 in the first turning movement process reaches the target number of turns, the moving body 1 is moved to the vicinity of an arbitrary marker M, and from this movement position, the same turning movement process as the first time is executed in the opposite direction to the first time.

[0014] Furthermore, in the unmanned operation system described in any one of claims 1 to 3, the control unit is configured to execute a safety confirmation process to determine whether or not the moving body 1 can move forward after the marker recognition process, and this safety confirmation process is characterized in that at least one of the following is executed: checking the temperature of components of the control unit, checking whether or not the moving body 1 has come into contact with an obstacle, checking whether or not an image has been acquired by the imaging unit 2, and checking the safety around the moving body 1.

[0015] Furthermore, in the unmanned operation system described in claim 4, the control unit is configured to execute a safety confirmation process to determine whether or not the moving body 1 can move forward after the marker recognition process, and this safety confirmation process is characterized in that at least one of the following is executed: checking the temperature of the components of the control unit, checking whether or not the moving body 1 has come into contact with an obstacle, checking whether or not an image has been acquired by the imaging unit 2, and checking the safety of the surroundings of the moving body 1.

[0016] Furthermore, in the unmanned operation system described in claim 5, the control unit is configured to execute a safety confirmation process to determine whether or not the moving body 1 can move forward after the marker recognition process, and this safety confirmation process is characterized in that at least one of the following is executed: checking the temperature of the components of the control unit, checking whether or not the moving body 1 has come into contact with an obstacle, checking whether or not an image has been acquired by the imaging unit 2, and checking the safety of the surroundings of the moving body 1. Effect of the Invention

[0017] Since the present invention is configured as described above, it does not require dedicated markers, but instead uses inexpensive and easily available markers such as commercially available pylons (triangular cones), resulting in an unmanned work system that can control the autonomous movement of a mobile body within a work area and perform specified tasks evenly throughout the work area. [Brief description of the drawings]

[0018] [Figure 1]1 is an explanatory diagram showing an example of the operation of a moving body according to the present invention; [Diagram 2] FIG. 2 is a perspective view showing a moving body of the present embodiment. [Diagram 3] FIG. 2 is a bottom view showing the moving body of the present embodiment. [Figure 4] 4 is a flowchart illustrating an initialization process according to the embodiment. [Diagram 5] 1 is a flowchart showing a marker discovery process according to the present embodiment. [Figure 6] 11 is a flowchart of a main loop process according to the present embodiment. [Figure 7] 11 is a flowchart illustrating a safety confirmation process according to the present embodiment. [Figure 8] 4 is a flowchart showing a turning process according to the embodiment. [Figure 9] 4 is a flowchart illustrating a self-position estimation process according to the present embodiment. [Figure 10] 1A to 1C are explanatory diagrams showing an example of the movement of a moving object relative to a work area in a time series according to an embodiment of the present invention. [Figure 11] 1 is an explanatory diagram showing an example of a movement trajectory of a moving object relative to a work area in the present embodiment; DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0019] A preferred embodiment of the present invention will be briefly described below with reference to the drawings to illustrate the operation of the present invention.

[0020] In the present invention, when work is started, first, an imaging process is executed by the imaging unit 2 to capture an image of the area ahead of the moving body 1 (in the direction of movement of the moving body 1), and then a marker recognition process is executed by the control unit to recognize the marker M in the image acquired by the imaging process of the imaging unit 2.

[0021] Next, the size of the marker M recognized in the marker recognition process in the image is obtained as a numerical value, and this obtained numerical value is compared with a movement stop reference value which is a reference value for stopping the movement of the moving body 1. If the obtained numerical value is smaller than the movement stop reference value, the moving body 1 is moved forward, and if the obtained numerical value is equal to or greater than the movement stop reference value, a forward / stop process is executed which stops the moving body 1.

[0022] In the present invention, the imaging process, the marker recognition process, and the forward movement / stop process are repeatedly executed, and the moving body 1 moves forward toward the destination marker M until the acquired numerical value becomes equal to or greater than the movement stop reference value.

[0023] When the acquired numerical value becomes equal to or greater than the movement stop reference value and the forward movement of the moving body 1 stops, a turning process is executed in which the moving body 1 is rotated in a specified direction to search for the marker M that will be the next destination, and the moving body 1 is turned in the direction of the discovered marker M. In addition, in this turning process, a stop position correction process is executed in which the movement stop reference value in the forward / stop process is subtracted and the moving body 1 after turning is stopped before the stopping position for the previous destination marker M, and then the imaging process → the marker recognition process → the forward / stop process are repeatedly executed, and the moving body 1 moves forward toward the next destination marker M until the acquired numerical value becomes equal to or greater than the movement stop reference value.

[0024] In the present invention, the stop position correction process is executed every time the moving body 1 turns, so that the timing of the moving body 1 turning relative to the destination marker M becomes gradually faster each time it turns. As a result, the moving body 1 moves in a circular motion from the outside to the inside within the work area A while repeatedly moving forward and turning, and the specified work can be performed evenly throughout the work area A.

[0025] Specifically, for example, when working in a rectangular work area A as shown in Figure 1, markers M1, M2, M3, and M4 are placed at each corner of the work area A, and the work area A is divided by these markers M1, M2, M3, and M4.In the work area A divided by markers M1, M2, M3, and M4, the moving body 1 is set at the position of marker M1, which is the starting point, facing marker M2, which is the next destination, and processing begins.

[0026] After the process starts, the moving body 1 positioned at the marker M1 recognizes the destination marker M2, moves forward toward the marker M2 while a numerical value indicating the size of the recognized marker M2 in the image is smaller than a movement stop reference value, and stops moving when a numerical value indicating the size of the marker M2 in the image recognized in the marker recognition process becomes equal to or larger than the movement stop reference value.

[0027] When the moving body 1 stops, it executes a turning process at that position and moves forward toward the next destination marker M3 that has been recognized (discovered) in this turning process. At this time, since a stop position correction process is executed during the turning process, the stopping position for marker M3 is stopped before marker M2. When the moving forward toward marker M3 ends and the moving body 1 stops, the same process as when moving from marker M2 to marker M3 is executed, and thereafter, the control unit repeatedly executes each process described above, and as shown in FIG. 1, the moving body 1 alternately moves forward and turns to move from the outside to the inside of the working area A, in other words, from the outermost periphery to the center of the working area A, while autonomously moving and rotating to perform a predetermined task.

[0028] In this way, the present invention is an innovative unmanned work system that does not require dedicated markers, but instead uses inexpensive and easily available markers M, such as commercially available pylons (triangular cones), to control the autonomous movement of the moving body 1 within the work area A and perform specified tasks evenly throughout the work area A. EXAMPLES

[0029] A specific embodiment of the present invention will now be described with reference to the drawings.

[0030] This embodiment is an unmanned work system that autonomously moves a moving body 1 within a work area A that is partitioned by arranging a plurality of markers M at each corner to perform a predetermined task, and the moving body 1 is equipped with an imaging unit 2 that images the area ahead, and a control unit that controls the movement of the moving body 1 based on the image acquired by the imaging unit 2. The control unit performs a marker recognition process that recognizes the marker M in the image acquired by the imaging unit 2, acquires a numerical value representing the size of the marker M recognized in the image in question, compares the acquired numerical value with a movement stop reference value that is a reference value for stopping the movement of the moving body 1, moves the moving body 1 forward if the acquired numerical value is smaller than the movement stop reference value, and stops the moving body 1 if the acquired numerical value is equal to or greater than the movement stop reference value, and performs a forward / stop process that ... When the forward movement of the moving body 1 stops, a turning process is executed in which the moving body 1 is rotated in a predetermined direction to search for a marker M that will be the next destination, and the moving body 1 is turned in the direction of the found marker M. Each time the moving body 1 turns, the movement stop reference value in the forward movement / stop process is subtracted, and the moving body 1 after turning is stopped before the stopping position for the previous destination marker M with respect to the destination marker M. (Note that this also includes exceptional cases in which the moving body 1 does not stop before the stopping position for the previous destination marker M with respect to the destination marker M due to changes in conditions such as the shape of the working area A and the number of target turns.) The moving body 1 is configured to move in a circular motion from the outside to the inside of the working area A while repeatedly moving forward and turning.

[0031] In this embodiment, "turning" means that the moving body 1 faces the direction of the destination, and "rotation" means the action of the moving body 1 to "turn". In other words, the moving body 1 "turns" by one or more "rotations". Also, "turning" means the moving state of the moving body 1 with respect to the working area A.

[0032] In this embodiment, the moving body 1 is configured as a grass cutter and configured as a grass cutting / lawn mowing work system that is applied to grass cutting and lawn mowing work as the predetermined work. Note that this embodiment is not limited to the grass cutting / lawn mowing work system, and can be configured as various work systems such as snow removal work and cleaning work.

[0033] Each component of this embodiment will be described in detail below.

[0034] The markers M used in this embodiment can be appropriately adopted as long as they have an appropriate height and width and can be recognized (detected) in an image acquired by imaging by an imaging unit 2 provided on a moving body 1 described later, and for example, commercially available pylons (triangular cones) that are inexpensive and easily available are preferable. Note that the number of markers M can be appropriately changed depending on the shape of the work area A.

[0035] As shown in Figs. 2 and 3, the moving body 1 is configured as a tracked vehicle in which tracks 4 (also called crawlers or caterpillars) are provided on the left and right sides of the main body 3 as a traveling means.

[0036] In addition, the moving body 1 of this embodiment is provided with eccentric tires 5 on the outside of the front of each crawler belt 4, and is configured so that the eccentric tires 5 enable the moving body 1 to move over steps that the crawler belt 4 cannot overcome.

[0037] Specifically, the crawler belt 4 and the eccentric tire 5 are of an electric type driven by a motor, and the eccentric tire 5 is mounted on the drive shaft of the crawler belt 4, shares a common drive source with the crawler belt 4, and is configured to rotate eccentrically as the crawler belt 4 is driven.

[0038] In addition to the above-mentioned running means, the moving body 1 of this embodiment is provided with an auxiliary tire 6 on the front side of the main body 3, and is configured so that when the moving body 1 leans forward due to the downward step, this auxiliary tire 6 comes into contact with the ground, thereby preventing the moving body 1 from falling over.

[0039] The moving object 1 is also provided with an imaging unit 2 on top of the main body 3, which captures an image of the area ahead.

[0040] Specifically, the imaging unit 2 is provided at the center of the movable body 1 in the width direction via a support portion 2a provided on the movable body 1.

[0041] In detail, the imaging unit 2 is a Web camera (a wide-angle camera is used in this embodiment) that can transfer and process captured images in real time, and is mounted on the moving body 1 via a support part 2a that is configured to be freely bent into a L-shape, as shown in FIG. 2, and is configured so that the height position can be changed (adjusted) by changing the degree of bending of the support part 2a.

[0042] As described above, this embodiment is configured to capture images using one imaging unit 2 (wide-angle camera), but two cameras, a wide-angle camera and a telephoto camera, may be used as the imaging unit 2, and these may be used selectively to detect a more distant marker M. In this way, when the marker M is lost, the image captured by the wide-angle camera is made higher resolution to make it easier to find the marker M, and once the marker M is found, the camera is switched to the telephoto camera, and the image captured by the telephoto camera is made lower resolution for image recognition, making it possible to recognize the marker M at a long distance without reducing the processing speed.

[0043] Furthermore, the support part 2a is not limited to the configuration of this embodiment, but may be an extendable rod-shaped body, or may have several support parts 2a of different lengths and be replaceable with the support part 2a of the appropriate length.

[0044] Also, the imaging unit 2 may be provided via a rotation mechanism (for example, a servo motor) so that the orientation can be freely moved in the horizontal direction (left and right direction). This allows the offset angle to be larger than in a configuration in which the imaging unit 2 is fixed.

[0045] Further, the movable body 1 is provided with grass-cutting means 7 on the lower part (bottom part) of the main body 3, and is configured as a grass cutter that cuts weeds and grass within the work area A.

[0046] Specifically, the grass-cutting means 7 is an electric type driven by a motor, and is configured as a rotary blade type having multiple blade portions 7a for cutting weeds and grass, and entanglement prevention members 7b for preventing weeds from becoming entangled around the blade portions 7a, as shown in FIG. 3.

[0047] More specifically, the anti-entanglement members 7b are made of nylon cord and are provided between each blade portion 7a in an upward facing state (with their tips facing the bottom side of the main body portion 3) (the blade portions 7a and the anti-entanglement members 7b are arranged alternately).

[0048] The movable body 1 of this embodiment is equipped with two grass-cutting means 7 of the above-described configuration, which are arranged side by side in the width direction of the main body 3. When a foreign object such as a pebble hits the blade portion 7a, the foreign object is ejected toward the rear, thereby preventing damage to the blade portion 7a.

[0049] In addition, the moving body 1 is provided with a bumper portion 8 that protrudes forward from the main body portion 3, as shown in FIG.

[0050] Specifically, the bumper portion 8 is provided with a contact sensor and is configured to detect whether or not the bumper portion 8 has come into contact with an obstacle.

[0051] The moving body 1 also has a battery (not shown) that serves as a power source for each component, such as the running motor, the grass-cutting motor, and the imaging unit 2. Furthermore, a solar panel 9 is provided on the top surface of the main body 3, which is connected to the battery and charges the battery during operation.

[0052] The moving body 1 also includes a control unit (not shown), and the autonomous movement (turning movement) within the working area A is controlled by this control unit.

[0053] The control unit is a small computer (single board computer), and in this embodiment, the control unit is equipped with Raspberry Pi (registered trademark) from the Raspberry Pi Foundation (UK). Note that the computer serving as the control unit is not limited to the above computer.

[0054] The control unit of this embodiment is configured to execute an initialization process, a main loop process, a turning process, a safety confirmation process, and a marker discovery process.

[0055] The initialization process is a process executed immediately after the start of work, and a target number of turns setting process is executed to set the number of turns required for the moving body 1 to move evenly throughout the work area A as the target number of turns.

[0056] Specifically, in the initialization process, as shown in the flowchart of FIG. 4, first, a marker recognition process for recognizing the marker M in the image acquired by the imaging unit 2 is executed.

[0057] The marker recognition process recognizes (detects) the marker M in the image based on a trained model of object detection (marker detection) generated by machine learning (deep learning).

[0058] In this embodiment, the machine learning involves preparing approximately 1,000 to 3,000 images, labeling the markers M in the prepared images, and training the labeled images using a GPU (image processing device), but the training method is not limited to this.

[0059] In addition, in the marker recognition process, markers M with different shapes can be recognized as the same marker M. In this case, this can be achieved by using a method in which different markers M are trained as the same class during machine learning, or a method in which different markers M are trained separately by machine learning and these different markers M are recognized as the same marker M.

[0060] Furthermore, if an identifier (such as a simple number, letter, or figure that allows individual recognition) is attached to the marker M in the initialization process, a marker discovery process is further executed.

[0061] In this marker discovery process, the process shown in the flowchart of FIG. 5 is executed.

[0062] Once a marker M is recognized in the image, the length of the longest side of the working area A is estimated based on the size of the recognized marker M, and a target number of turns setting process is executed to set the number of turns required for the moving body 1 to turn in the working area A as the target number of turns based on the length of the longest side.

[0063] Specifically, in the target number of turns setting process, the height of the marker M first recognized after the process begins is calculated, specifically, the ratio (height occupancy rate) of the number of pixels in the height direction occupied by the marker M in the image to the number of pixels in the height direction of the entire image, and the reciprocal of this height occupancy rate of marker M multiplied by a constant is calculated as the target number of turns.

[0064] For example, if the number of pixels occupied by the height of marker M recognized in an image with a resolution of 320 x 240 is 24 (pixels), the number of pixels for the height of the entire image is 240 (pixels), so the height occupancy rate of marker M is 24 / 240 = 1 / 10, and its reciprocal is 10. This reciprocal 10 is multiplied by a constant (set to "10" in this embodiment), so the target number of turns becomes 10 x 10 = 100 (times).

[0065] It should be noted that the above calculation method assumes that the moving body 1 will be able to turn and move efficiently when the working area A is set to a square. However, even if the working area A is a rectangle, a triangle, a pentagon, or a hexagon, the calculation method includes an operating margin so that work can be carried out without problems, although the work efficiency will be somewhat reduced.

[0066] In the main loop process, processes related to forward movement, stopping, and turning of the moving body 1 and safety confirmation are executed, as shown in the flowchart of FIG.

[0067] Specifically, first, the marker M is recognized in the image captured by the imaging unit 2 through a marker recognition process, and then a safety confirmation process is executed to determine whether forward movement is permitted.

[0068] In the safety confirmation process, as shown in the flowchart of Figure 7, confirmation is performed of the surrounding safety, such as checking the temperature of system components such as the control unit, whether the moving object 1 is in contact with an obstacle, whether an image is being acquired by the imaging unit 2, and whether there are any people or other moving objects nearby.

[0069] If there is no problem in this safety confirmation process (if no malfunction is detected), the process proceeds to forward movement / stopping process, and forward movement process of the moving body 1 is executed.

[0070] In the forward / stop processing, the size of the marker M in the image recognized in the marker recognition processing, specifically, the number of pixels occupied by the height of marker M in the image, is obtained, and this obtained marker height (number of pixels) is compared with a movement stop reference value, which is a reference value for stopping the movement of the moving body 1.If the marker height (number of pixels) is smaller than the movement stop reference value, the moving body 1 is moved forward, and if the marker height (number of pixels) is equal to or greater than the movement stop reference value, the moving body 1 is stopped.

[0071] Specifically, the movement stop reference value is a value obtained by multiplying the number of vertical pixels of the entire image by a predetermined coefficient, and in this embodiment, the predetermined coefficient is set to 0.9 (90%).

[0072] Therefore, for example, when the image resolution is 320 x 240, the number of vertical pixels is 240 (pixels), so the movement stop reference value is 240 x 0.9 = 216 (pixels). In this case, in the main loop process, the moving body 1 is moved forward until the number of pixels occupied by the height of the marker M in the image reaches 216 (pixels), and the moving body 1 is stopped when the number of pixels occupied by the height of the marker M in the image reaches 216 (pixels).

[0073] In the above-described forward movement / stopping process, in this embodiment, the number of pixels occupied by the height of the marker M in the image is obtained, and this obtained marker height (number of pixels) is compared with a movement stop reference value that is a reference value for stopping the movement of the moving object 1, but for example, a horizontally long marker M may be used and the number of pixels of the width dimension of the marker M may be obtained. In this case, the movement stop reference value is a value obtained by multiplying the number of horizontal pixels of the entire image by a predetermined coefficient.

[0074] Also, in this main loop processing, when the moving body 1 moves forward toward the destination, it is determined whether or not there is a deviation in the moving direction relative to the destination, and if there is a deviation, a moving angle correction process is executed to correct the deviation.

[0075] The movement angle correction process is executed in two stages, one for when the deviation amount is large and one for when it is small, as shown in the flowchart of the main loop process in FIG.

[0076] Specifically, in this embodiment, first, it is determined whether the amount of deviation (the angle of deviation in the moving direction of the moving body 1 relative to the destination) is within 5 degrees, and if the amount of deviation is greater than 5 degrees, a process is executed to correct the deviation by rotating the left and right tracks 4 in opposite directions, and if the amount of deviation is within 5 degrees, a further process is executed to determine whether the amount of deviation is within 2 degrees.

[0077] If the deviation is within 2 degrees, forward movement processing is executed as is, and if the deviation is more than 2 degrees, processing is executed to correct the deviation by adjusting the drive output of the left and right crawlers 4.

[0078] In this forward movement / stop process, when the stop process is executed, the turning process is executed and the turning operation of the moving body 1 (operation to change the direction toward the next destination marker M) is started.

[0079] In this turning process, the process shown in the flowchart of FIG. 8 is executed.

[0080] Specifically, the turning process involves executing a process for rotating the moving body 1 by a predetermined angle for a fixed period of time, searching for the marker M that will be the next destination, and turning the moving body 1 in the direction of the marker M that is found.

[0081] In addition, in this turning process, a stop position correction process is executed for each process (each time this turning process is performed) in which a movement stop reference value in the forward / stop process is subtracted, and a movement direction correction process is executed in which the movement direction of the moving body 1 in this turning process is corrected inward by a predetermined amount relative to the destination marker M located in the movement direction.

[0082] Specifically, in the stop position correction process, the movement stop reference value is calculated using the following formula (1).

[0083] Movement stop threshold = marker height + (total image height - marker height) x (1 - converted rate) (1) Here, the turning ratio is (number of turns performed) / (target number of turns), and the target number of turns is the number of turns required for the moving body 1 to turn and move around the working area A, which is set in the target number of turns setting process in the initialization process. Also, the marker height is the number of pixels in the image that the height of the marker M discovered as the next destination in the turning process occupies, and the total image height is the number of vertical pixels of the entire image in which the discovered marker M is captured.

[0084] In the above, the turning ratio increases in value each time the moving body 1 turns, so in equation (1), (total image height - marker height) x (1 - turned ratio) decreases in value each time the moving body 1 turns, and as a result, the movement stop reference value is subtracted.

[0085] In addition, in the movement stop reference value, the amount of change in the marker height is large when the marker M is close to the moving body 1, and the amount of change is small when it is far away, so it is better to subtract a large amount in the early stages of the work to improve accuracy. To achieve this, it is preferable to make (1 - redirected ratio) in formula (1) nonlinear as ((1 - redirected ratio)^2).

[0086] By executing this stopping position correction process, the progress rate of the moving body 1 relative to the distance it can travel to its destination (destination) decreases, and the timing at which the moving body 1 stops at the destination marker M becomes earlier with each turn, allowing the moving body 1 to gradually move toward the center of the working area A.

[0087] In addition, the movement direction correction process is executed each time the moving body 1 turns, i.e., each time this turning process is executed, in which the moving direction of the moving body 1 is corrected (the destination is shifted) by a predetermined angle inward relative to the destination marker M located in the moving direction.

[0088] Specifically, in the movement direction correction process, a correction angle is calculated using the following formula (2), and a process is executed to correct the movement destination (target value) of the movement destination marker M inward by the amount of this correction angle.

[0089] Correction angle = (maximum correction angle) × (conversion rate) (2) Here, the maximum correction angle is a preset value (angle), and is set to "17" in this embodiment. The turning ratio is (number of turns performed) / (target number of turns), and the target number of turns is the number of turns required for the moving body 1 to turn in the working area A, which is set in the target number of turns setting process in the initialization process.

[0090] In the above, since the turning ratio increases in value every time the moving object 1 turns, the correction angle calculated by the formula (2) increases (is added) in value every time the moving object 1 turns.

[0091] By executing this movement direction correction process, the moving body 1 can move at approximately equal intervals along the outer circumferential trajectory, and the moving body 1 can be controlled to move evenly throughout the work area A so that no work is left unfinished (in this embodiment, so that no grass is left uncut).

[0092] In addition, in this embodiment, when the number of turns of the moving body 1 reaches the target number of turns in the turning process described above, the first turning movement process is terminated, and a process for performing a second turning movement process is executed.

[0093] Specifically, when the first turning movement process is completed (when the number of turns of the moving body 1 reaches the target number of turns), the number of turns in this turning process is reset and the turning direction is set to the opposite direction to the turning direction in the first turning movement process (for example, if the first turning movement process was set to counterclockwise, the turning direction will be leftward, and therefore the second turning movement process will be clockwise, so the turning direction is set to rightward), and then the process proceeds to a main loop process, in which the moving body 1 is moved to marker M, which is the starting point of the second turning movement process, and thereafter, each process is repeatedly executed in the same manner as the first turning movement process, and a turning movement process is performed in which the moving body 1 turns in the opposite direction to the first time, and if the number of turns in the turning process in this second turning movement process reaches the target number of turns, a process is executed to stop the system.

[0094] In this embodiment, for the sake of convenience in explaining each process, the process is explained separately as shown in the flowchart, but in reality, each process is executed within one main loop process.

[0095] In this embodiment, a self-position estimation process may be added to the control unit so that the position shift of the moving object 1 during work can be corrected with higher accuracy. This self-position estimation process is effective when an identifier is assigned to the marker M.

[0096] Specifically, the self-position estimation process is configured to execute the process shown in the flowchart of FIG. 9, estimate the position of the moving body 1, and correct the moving direction of the moving body 1 based on this estimated position.

[0097] Since this embodiment is configured as described above, there is no need for dedicated markers as in the conventional example. Instead, inexpensive and easily available markers M, such as commercially available pylons (triangular cones), are placed at each corner of the work area A, and the moving body 1 is simply set at marker M (marker M on one side of the longest side in the work area A) which serves as the starting point, facing the next destination marker M. With this simple preparation, the moving body 1 can autonomously move in the work area A to perform a specified task (grass cutting or lawn mowing in this embodiment).

[0098] In addition, in this embodiment, by executing the stopping position correction process and the moving direction correction process, the turning timing of the moving body 1 is advanced and the moving direction of the moving body 1 relative to the destination marker M is corrected each time the moving body 1 turns, so that work (grass cutting, lawn mowing) can be performed so as to fill the work area A neatly with a surface.

[0099] Moreover, in this embodiment, as shown in Figure 10, the moving body 1 performs two rotational movement processes, a counterclockwise rotational movement process (Figure 10(a) → Figure 10(b)) and a clockwise rotational movement process (Figure 10(c) → Figure 10(d)), relative to the work area A. Therefore, as shown in Figure 11, compared to a single rotational movement process, the moving body 1 moves more evenly within the work area A. This makes it possible to more neatly fill in the work area A with a surface and reduces uncut grass as much as possible, resulting in an innovative unmanned work system with excellent practicality.

[0100] It should be noted that the present invention is not limited to the present embodiment, and the specific configuration of each component can be designed as appropriate. [Explanation of symbols]

[0101] 1. Mobile 2. Imaging unit A Work Area M Marker

Claims

1. An unmanned work system in which a mobile body moves autonomously within a work area partitioned by arranging a plurality of markers at each corner to perform a predetermined task, the mobile body being equipped with an imaging unit that images an area ahead, and a control unit that controls the movement of the mobile body based on the image acquired by the imaging unit, the control unit performing a marker recognition process that recognizes a marker in the image acquired by the imaging unit, acquiring a numerical value representing the size of the marker recognized in the image, comparing the acquired numerical value with a movement stop reference value that is a reference value for stopping the movement of the mobile body, and if the acquired numerical value is smaller than the movement stop reference value, moving the mobile body forward, and controlling the movement of the mobile body based on the acquired numerical value. an advance / stop process for stopping the moving body when the detected value is equal to or greater than a movement stop reference value; a turning process for rotating the moving body in a predetermined direction when the forward movement of the moving body stops, searching for a marker that will be the next destination, and turning the moving body in the direction of the found marker; and a stop position correction process for subtracting the movement stop reference value in the advance / stop process each time the moving body turns, and stopping the moving body after turning shorter than the stopping position for the previous destination marker, wherein the moving body is configured to move in a circular motion from the outside to the inside of the work area while repeatedly moving forward and turning.

2. In the unmanned operation system described in claim 1, the control unit executes a movement direction correction process to correct the movement direction of the moving body a predetermined amount inward relative to a destination marker located in the movement direction each time the moving body turns, and further, in this movement direction correction process, the predetermined amount is configured to increase each time the moving body turns.

3. In the unmanned operation system described in claim 1, the acquired numerical value in the forward movement / stop processing is the number of pixels occupied by the height of the marker in the image acquired by the imaging unit, and the movement stop reference value is a value obtained by multiplying the number of vertical pixels of the entire image acquired by the imaging unit by a predetermined coefficient.

4. 3. The unmanned operation system according to claim 2, wherein the acquired numerical value in the forward movement / stop processing is the number of pixels occupied by the height of the marker in the image acquired by the imaging unit, and the movement stop reference value is a value obtained by multiplying the number of vertical pixels of the entire image acquired by the imaging unit by a predetermined coefficient.

5. In the unmanned work system described in any one of claims 1 to 4, the control unit is configured to estimate the length of the longest side of the work area, and based on this length of the longest side, execute a target number of turns setting process to set the number of turns required for the rotating movement of the moving body in the work area as a target number of turns, and when the number of turns of the moving body reaches the target number of turns, terminate the movement of the moving body.

6. In the unmanned operation system described in claim 5, the control unit executes a left rotation movement process in which the moving body performs the specified work while rotating in a left direction, and a right rotation movement process in which the moving body performs the specified work while rotating in a right direction, and when the number of turns of the moving body in a first rotation movement process reaches the target number of turns, the moving body is moved to the vicinity of an arbitrary marker, and from this movement position, a rotation movement process similar to the first time is executed in the opposite direction to the first time.

7. 5. The unmanned operating system according to claim 1, wherein the predetermined operation performed by the moving body is grass-cutting.

8. 6. An unmanned operating system according to claim 5, wherein the predetermined work carried out by the mobile body is grass-cutting work.

9. 7. An unmanned operating system according to claim 6, wherein the predetermined work carried out by the mobile body is grass-cutting work.

Images