Control method for autonomous mobile robots and autonomous mobile robots

Abstract

To improve the accuracy of the stopping position when an autonomous mobile robot stops at a target location. [Solution] The autonomous mobile robot autonomously moves along a first target path to a first stop position just before the target stop position, while measuring and comparing the surrounding conditions using laser scanning (S1, S2). While stopped at the first stop position, it recognizes the position of a marker placed near the target stop position (S3), generates a second target path approaching the marker (S4), performs odometry movement to the second stop position following the generated second target path (S5, S6), recognizes the marker while stopped at the second stop position (S7), estimates the error with the target trajectory to the target stop position, performs odometry movement to correct the error (S8, S9), and stops at the target stop position (S10).

Description

Technical Field

[0001] The present invention relates to the technology of autonomous mobile robots.

Background Art

[0002] In recent years, as a successor to the AGV (Automatic Guided Vehicle) that has been active as an automated guided vehicle, an autonomous mobile robot called AMR (Autonomous Mobile Robot) has attracted attention. This AMR is a robot that makes its own judgments and automatically avoids people and obstacles to move. It has advantages such as reducing the movement distance of on-site workers and enabling introduction in a shorter period compared to AGVs, making it easier to introduce robots.

[0003] On the other hand, in the fields of AGVs and AMRs, a robot that autonomously moves along a preset path may have a target stop position set in order to stop at a predetermined position such as for power supply. For example, in Patent Document 1, it is described that by photographing markers at multiple points until reaching the target position and measuring the positions of the photographed markers, the error at the target position can be reduced.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] However, in the conventional technology, for example, when performing autonomous driving by laser scan matching, it was difficult to improve the stop position accuracy beyond a level of about ±25 mm due to laser measurement errors and scan match accuracy. And there was a problem that with this level of accuracy, it was impossible to meet the accuracy required, for example, at the time of stopping for power supply.

[0006] The object of the present invention is to provide a method for stopping an autonomous mobile robot and an autonomous mobile robot that can further improve the accuracy of stopping at a target stopping position. [Means for solving the problem]

[0007] A brief overview of some of the representative inventions disclosed in this application is as follows:

[0008] A typical embodiment of the present invention provides a control method for an autonomous mobile robot, While measuring the surrounding conditions using laser scanning and comparing the measurement results with the stored surrounding conditions, the system autonomously moves along the first target path, moving to a first stop position before the target stop position. In the stopped state at the first stopping position, the marker is imaged, and the coordinate position and direction of travel of the autonomous mobile robot relative to the marker are measured. A second target path is generated from the first stop position to approach the marker, and the vehicle moves along the second target path from the first stop position to the second stop position, recording the history of rudder angle and speed, based solely on odometry information. In the stopped state at the second stopping position, the marker is imaged again, and the coordinate position and direction of travel of the autonomous mobile robot relative to the marker are remeasured. Based on the coordinate position and direction of travel measured at the second stopping position, and the history of rudder angle and speed recorded during movement from the first stopping position to the second stopping position, the estimated coordinate position and direction of travel at the first stopping position are calculated by reverse calculation. This is a control method for autonomous mobile robots.

[0009] Furthermore, an autonomous mobile robot according to a typical embodiment of the present invention is An autonomous movement control unit that autonomously moves along a first target path to a first stop position while measuring the surrounding conditions using laser scanning and comparing the measurement results with the stored surrounding conditions, An imaging unit that images a marker at the first stop position, A self-position estimation unit estimates position coordinates and orientation based on a marker image acquired by the imaging unit, A target path generation unit generates a second target path that approaches the marker from the first stop position, When moving along the second target path to the second stopping position, an odometry calculation unit performs odometry calculations based on the steering angle and speed history acquired by the steering angle sensor and wheel rotation sensor, A rudder angle offset estimation unit calculates the position coordinates and orientation based on a marker image captured again at the second stop position, and the estimated value of the first stop position based on the history. It is an autonomous mobile robot equipped with [specific features / equipment]. [Effects of the Invention]

[0010] The effects obtained by some of the representative inventions disclosed in this application can be briefly explained as follows.

[0011] In other words, according to a typical embodiment of the present invention, the accuracy of the stopping position when an autonomous mobile robot stops near a marker can be improved. [Brief explanation of the drawing]

[0012] [Figure 1] This is a block diagram showing the main components of the AMR used in the AMR system in this embodiment. [Figure 2] This is a perspective view illustrating the overall overview of the AMR system in this embodiment. [Figure 3] This is a plan view illustrating the case where the AMR's driving path deviates. [Figure 4] This is a plan view illustrating the relationship between misrecognition of marker angle and misrecognition of self-position. [Figure 5] This is a plan view illustrating the impact of misidentification of one's own position on the travel path. [Figure 6] This is a plan view illustrating a route setting method that suppresses route deviation. [Figure 7] It is a plan view for explaining the position for switching the traveling mode of the AMR, etc. [Figure 8] It is a diagram for explaining a method of estimating the rudder angle offset value. [Figure 9] It is a diagram for explaining a method of determining the rudder angle offset amount. [Figure 10] It is a flowchart for explaining the flow of processing performed by the AMR.

Mode for Carrying Out the Invention

[0013] (Embodiment 1) Hereinafter, embodiments of the present invention will be described while referring to the drawings. Each embodiment described below is an example for realizing the present invention and does not limit the technical scope of the present invention. In the examples, members having the same function are denoted by the same reference numerals, and repeated descriptions thereof are omitted unless particularly necessary.

[0014] <Summary> FIG. 1 is a block diagram showing the main configuration of an AMR (Autonomous Mobile Robot: autonomous traveling and transporting robot) used in the AMR system in the present embodiment. As shown in FIG. 1, the AMR 1 includes a drive device 10, a path following unit 20, a marker photographing camera 30, a self-position estimation unit 31, a target path generation unit 32, and a rudder angle offset estimation unit 33.

[0015] Further, the AMR 1 includes an odometry calculation unit 34, a wheel rotation sensor 40, a rudder angle sensor 50, a self-position estimation unit 35, and a Lidar 60.

[0016] Of the above, the self-position estimation unit 31, the target path generation unit 32, the rudder angle offset estimation unit 33, the odometry calculation unit 34, and the self-position estimation unit 35 may each use different hardware resources (processors such as microcontrollers, CPUs, and GPUs, and associated storage media such as RAM), or they may be implemented as program modules using the same hardware resources.

[0017] The AMR1's drive unit 10 includes a motor as a drive source, a drive circuit that drives the motor, wheels that are rotated by the motor's driving force (see Figure 7 as appropriate), a transmission mechanism such as gears and shafts that transmit the motor's driving force to the wheels, and a steering mechanism for changing or adjusting the direction of the AMR1 by changing the direction of the wheels.

[0018] Based on the calculation results of the target path generation unit 32, odometry calculation unit 34, and self-position estimation unit 31, which will be described later, the path following unit 20 outputs steering angle command values ​​and speed command values ​​to the drive unit 10 so that the AMR1 travels along the target path TW on the road surface (see Figure 3, etc., as appropriate).

[0019] The marker imaging camera 30 is a camera that captures and acquires an image of the marker 100M (see Figure 2, etc., as appropriate), and various imaging devices such as a CCD camera or a C-MOS camera can be used. The captured image of the marker 100M is supplied from the marker imaging camera 30 to the self-position estimation unit 31.

[0020] The self-position estimation unit 31 determines, for example, the distance to marker 100M and the orientation of AMR1 relative to marker 100M (in other words, the angle it makes with respect to the target path TW on the road surface) from the image of marker 100M supplied from the marker-capturing camera 30, and estimates the current position of AMR1 on the road surface (hereinafter also referred to as self-position). Thus, the information (numerical value) of AMR1's self-position estimated based on the image of marker 100M is supplied from the self-position estimation unit 31 to the target path generation unit 32 and the steering angle offset estimation unit 33, respectively.

[0021] The target path generation unit 32 calculates the deviation (length and direction of the offset) from the target path TW (see Figure 3, etc., as appropriate) based on the self-position information (numerical value) of the AMR1 supplied by the self-position estimation unit 31, and supplies the calculated value to the path tracking unit 20. In this embodiment, the target path generation unit 32 also plays a role in setting the second target path STW, which will be described later in Figure 6.

[0022] The rudder angle offset estimation unit 33 calculates the directional deviation (offset angle) relative to the target path TW (see Figure 3, etc., as appropriate) from the self-position information (numerical value) of the AMR1 supplied by the self-position estimation unit 31, and supplies the calculated value to the odometry calculation unit 34.

[0023] The wheel rotation sensor 40 is a sensor that detects the rotational speed of the wheels (front wheels 2) of the AMR1, which will be described later in Figure 7. For example, it includes a gear-shaped rotor provided on the wheel's rotation axis, and a sensor (magnetic encoder) consisting of a coil and magnetic poles on the outer circumference of the rotor. The wheel rotation sensor 40 supplies the detected rotational speed to the odometry calculation unit 34.

[0024] The steering angle sensor 50 is a sensor that detects the steering angle of the front wheel 2 of the AMR1, which will be described later in Figure 7. For example, it includes a torque sensor provided on the gear of the steering shaft of the front wheel 2. The steering angle sensor 50 supplies the detected steering angle signal (hereinafter also simply referred to as steering angle) to the odometry calculation unit 34.

[0025] The odometry calculation unit 34 uses the calculated value (offset angle) supplied from the steering angle offset estimation unit 33, the rotational speed supplied from the wheel rotation sensor 40, and the steering angle supplied from the steering angle sensor 50 to determine the amount of movement of the AMR1, i.e., the direction and length of movement, and estimates the position of the AMR1 from the cumulative calculation of this amount of movement. The odometry calculation unit 34 supplies the estimated position of the AMR1 to the path following unit 20.

[0026] Based on the calculated deviation from the target path TW supplied by the target path generation unit 32 and the estimated position of the AMR1 supplied by the odometry calculation unit 34, the path following unit 20 outputs speed command values ​​and rudder angle command values ​​to the drive unit 10 so that the AMR1 approaches the target path TW and the position of the AMR1 lies on the target path TW.

[0027] On the other hand, Lidar60 is a device that uses electromagnetic waves with shorter wavelengths than radio waves (radar) to measure the distance to an object, its position, and its shape in three dimensions. Although not shown in the diagram, it includes an electromagnetic wave irradiation unit (such as a laser beam), a receiving unit, and a calculation unit. Lidar60 can capture the environment around AMR1 (walls, equipment, etc.) by irradiating electromagnetic waves from the irradiation unit in a pulsed manner, for example, and measuring the time difference between when the waves are reflected by the object and when they are received by the receiving unit using the calculation unit.

[0028] The self-position estimation unit 35 pre-stores map data that includes 3D position information of the road surface and surrounding environment (walls, equipment, etc.) along the path that the AMR1 will travel (drive) along. By performing laser scan matching using this map data and the calculation results of the Lidar 60, the self-position estimation unit 35 estimates the self-position of the AMR1 (current coordinate position on the road surface and direction of travel) by comparing the 3D information of the surrounding environment captured by the Lidar 60 with the 3D position information of the map data. The self-position estimation unit 35 outputs the estimated self-position of the AMR1 to the path-following unit 20. Here, the path-following unit 20 uses the self-position of the AMR1 (estimated value) output from the self-position estimation unit 35 to output speed command values ​​and steering angle command values ​​to the drive unit 10 so that the AMR1 approaches the target path TW and the position of the AMR1 is on the target path TW.

[0029] As described above, the AMR1 used in the AMR system of this embodiment includes a self-position estimation unit 35 that estimates the AMR1's own position using a Lidar 60, and a self-position estimation unit 31 that estimates the AMR1's own position by recognizing images of markers captured by a marker camera 30. The path-following unit 20 of the AMR1 in this embodiment uses, or rather, differentiates between, the estimation results obtained by laser scan matching from the self-position estimation unit 35 and the estimation results obtained by image recognition from the self-position estimation unit 31, and outputs speed command values ​​and rudder angle command values ​​to the drive unit 10 so that the AMR1's position is on the target path TW.

[0030] Figure 2 is a perspective view illustrating the overall overview of the AMR system in this embodiment. Note that the AMR1 shown on the left and right sides of Figure 2 is the same vehicle, and here we assume that it is traveling from left to right. In this example, equipment 100 is located on the left side of the road surface in the direction of travel, and a marker 100M is installed on the road-facing wall of equipment 100. Marker 100M has an arbitrary pattern that is the target of image recognition, and is installed so that the patterned side is at an angle to the wall of equipment 100 so that it can be easily photographed by the marker-capturing camera 30 (see also Figure 7 as appropriate). AMR1 stops along the side wall of equipment 100, performs predetermined work while stopped, and then moves again to the right side in Figure 2 to proceed to the next work area.

[0031] In one specific example of this system, as a preliminary step, the AMR1 is manually controlled to navigate the course where autonomous navigation is desired (i.e., a teaching run is performed), and the surrounding environment of the robot is measured and recorded using a laser rangefinder (LIDAR). Then, during autonomous navigation, scan matching is performed using Lidar60, and the surrounding environment measured by Lidar60 is compared with the data measured and recorded during the teaching run to correct the position of the AMR1 (see the driving state on the left in Figure 2).

[0032] On the other hand, due to the laser's ranging error and scan matching accuracy, the stopping position accuracy was such that it was about ±25 mm. When an error of this level occurs, there is a problem in that the required accuracy cannot be met, for example, when stopping to supply power from the power supply device of equipment 100 to the AMR1.

[0033] Therefore, in order to further improve the stopping position accuracy of the AMR1, when approaching a stopping position that requires high-precision positioning, it is conceivable that the AMR1 would then photograph a marker (marker 100M) placed near the stopping position with a marker-capturing camera 30, switch to a positioning mode based on the AMR1's own position obtained through image recognition, and then travel to the stopping position.

[0034] On the other hand, the technology for recognizing the AMR1's own position using such image loss recognition had the following challenges.

[0035] First, the method of photographing the 100M marker and measuring the self-position from the image has a drawback: the accuracy of the measurement decreases as AMR1 moves away from the 100M marker, and AMR1 is more likely to deviate from its path. This will be explained later with reference to Figure 2 and subsequent figures.

[0036] Furthermore, since marker 100M is positioned to the side, avoiding the AMR1's path, if the marker-capturing camera 30 is a monocular camera, marker 100M will be outside the field of view at the target stopping position. This presents a challenge, as it necessitates navigation using only odometry in areas outside the field of view.

[0037] Furthermore, if position measurement is performed while the AMR1 is in motion, both the marker camera 30 and the data processing processor require high speed, leading to increased costs. Considering these costs, if a relatively low-cost camera such as a monocular camera is used for marker photography, it is thought that stopping the AMR1 to photograph the marker 100M and measure its own position would improve measurement accuracy. However, considering productivity and other factors, it is thought that the number of times the AMR1 is stopped should be kept to a minimum.

[0038] In relation to the above issues, the factors that cause AMR1 to deviate from its path while photographing marker 100M will be explained with reference to Figures 3 to 6.

[0039] In the examples shown in Figures 3 to 6, the target path TW that the AMR1 should follow is indicated on the road surface by a dashed line. This target path TW is set as a roughly straight path slightly to the left of the center of the road surface. In the example shown in Figure 3, the AMR1 is traveling to the right of the target path TW. Subsequently, the path-following unit 20 of the AMR1 controls the drive unit 10 to move the vehicle body to the left and travel along the target path TW, as shown by the dashed arrow in Figure 3, based on images of the marker 100M and the equipment 100 on which the marker 100M is installed, taken by the marker-capturing camera 30.

[0040] However, while the driving trajectory shown by the dashed arrow in Figure 2 can be achieved relatively easily when a human manually remotely controls the AMR1, it was found that errors are likely to occur when autonomous driving is performed using a method that estimates the self-position based on images taken of marker 100. The reason for this will be explained below with reference to Figure 4 and subsequent figures.

[0041] In the example shown in Figure 4, AMR1 is traveling along the target path TW, and, as in the example shown in Figure 3, is photographing marker 100M with a marker-capturing camera 30 (not shown in the figure). At this time, AMR1 measures the relative positional relationship between AMR1 and marker 100M, and the relative attitude (angle) of the marker-capturing camera 30 and, consequently, the AMR1's body relative to marker 100M, from the image of marker 100M. Of these, the latter, that is, the measurement of the relative attitude (angle) of the marker-capturing camera 30 and, consequently, the AMR1's body relative to marker 100M, is prone to errors if the distance from marker 100M is large.

[0042] For example, in the case shown in Figure 4, the line connecting the marker-capturing camera 30 mounted on the AMR1 and the marker 100M positioned on the left edge of the road surface in the direction of travel is not parallel to the target path TW shown by the dashed line in Figure 4, but has an angle with respect to the target path TW. This angle is equivalent to (but actually in the opposite direction to) the angle shown by the auxiliary line for convenience shown by the dotted line in Figure 4. Due to this angle, the AMR1 sometimes misinterpreted the vehicle's attitude relative to the marker 100M. And when the AMR1's vehicle attitude relative to the marker 100M is misinterpreted, the position of the AMR1's vehicle is misinterpreted as a result. In the example shown in Figure 4, the AMR1 mistakenly perceives that the vehicle's attitude is (more) to the left than it actually is, and that the vehicle is traveling to the right of the target path TW in the direction of travel (see the AMR shown in the dashed box 1E in Figures 4 and 5).

[0043] Therefore, AMR1 is controlled to travel along the path indicated by symbol ER in Figure 5, in other words, along the calculated corrected path ER based on the misrecognized attitude and position. As a result, AMR1 travels along the path indicated by symbol AR in Figure 5, which is a path with approximately the same shape as the calculated corrected path ER. In this example, as shown by the dashed frame 1A in Figure 5, there were problems such as the body of AMR1 protruding to the left of the road surface.

[0044] The inventors investigated these problems from various perspectives and obtained the following findings.

[0045] When position measurement is performed using images captured by a camera, the accuracy of position measurement decreases as the camera's position, more specifically the position of the AMR1 equipped with the marker-capturing camera 30, moves further away from the marker (100M).

[0046] One possible method to eliminate the errors caused by the angles mentioned above is to place marker 100M on the target path. However, from a practical standpoint and for collision prevention, it is desirable to place marker 100M to the side, avoiding the path. Therefore, if the marker-capturing camera 30 mounted on the AMR1 is a monocular camera, and set to the target stopping position as shown in Figure 3, the marker-capturing camera 30 will be in front of marker 100M, thus excluding marker 100M from the field of view. Consequently, in such cases, self-position estimation by image recognition cannot be performed near the target stopping position, and navigation will be required using odometry alone. Furthermore, if position measurement is performed while the AMR1 is in motion, both the marker-capturing camera 30 and the hardware (processor, etc.) that processes data for position measurement will require high speed, resulting in high costs. On the other hand, if position measurement is performed using images of marker 100M with the AMR1 stopped, the accuracy of self-position estimation improves, but from a productivity standpoint, it is desirable to minimize the number of times the AMR1 is stopped.

[0047] Based on the above findings, in this embodiment, the AMR1 is stopped at two locations: a first stopping position which is a certain distance from marker 100M, and a second stopping position which is the closest point to marker 100M. Marker 100M is then imaged using the marker imaging camera 30, and the position is measured from the captured image.

[0048] Furthermore, in this embodiment, as shown in Figure 6, when moving from the first stopping position to the second stopping position, the AMR1 sets a straight line as the second target path STW, which connects the current position recognized by the machine (the left-right center position of the frame indicated by reference numeral 1E) and the foot of the perpendicular line drawn from marker 100M to the target trajectory. The AMR1 then travels in accordance with the set second target path STW.

[0049] Next, a specific example of the setting values ​​for the first and second stopping positions will be explained with reference to Figure 7. The first stopping position, indicated by the symbol FSP in Figure 7, is set to a distance D that is as close as possible to the marker 100M and within a range that allows for correction of the deviation from the target path (for example, about ±5cm) detected by image-based position measurement up to the target stopping position TSP.

[0050] In the example shown in Figure 3 above, the target stopping position TSP was set relative to, for example, the area slightly behind the center where the AMR1's power connector is located. However, in the example shown in Figure 7, the target stopping position TSP is set relative to the front of the AMR1 vehicle. There are no particular restrictions on which position the target stopping position TSP is set relative to; any position that allows control to bring the AMR1 to a stationary position is acceptable.

[0051] Furthermore, the reason for setting the first stopping position FSP as close as possible to the target stopping position TSP and marker 100M is, as mentioned above, that the accuracy of position measurement based on images taken of marker 100 decreases as the distance from marker 100M increases, and conversely, the accuracy of position measurement increases as the distance from marker 100M increases.

[0052] After conducting various experiments, the inventors found that it is advantageous to set the distance D to marker 100M or target stopping position TSP to approximately 2m. In this example, the first stopping position is set to a point approximately 2m away from marker 100M, and the AMR1 makes its first stop at this first stopping position.

[0053] On the other hand, the second stopping position is defined as a point on the aforementioned path that is close to marker 100M, within the range where marker 100M remains within the field of view of the marker-capturing camera 30 (see symbol SSP in Figure 7). In other words, the second stopping position SSP is a point near the limit where marker 100M remains within the field of view of the marker-capturing camera 30 as AMR1 travels along the second target path STW described in Figure 6. AMR1 makes its second stop at this second stopping position SSP.

[0054] Furthermore, AMR1 uses only odometry information to travel from the first stopping position to the second stopping position SSP, and from the second stopping position SSP to the final target stopping position TSP (see Figure 7). When AMR1 stops at the second stopping position SSP, it estimates the rudder angle offset based on the difference (error) between the odometry information from the first stopping position to the second stopping position SSP and the position information from the image. Then, during travel from the second stopping position SSP to the target stopping position TSP, AMR1 uses the estimated rudder angle offset to follow the target trajectory (target path TW). For tracking the target trajectory, known algorithms such as the purepursuit algorithm can be used. For position measurement of marker 100M, known algorithms can be used.

[0055] Next, the control of travel from the second stopping position SSP to the target stopping position TSP will be explained with reference to Figures 8 and 9. Figure 8 is a diagram illustrating the control content for travel from the second stopping position to the target stopping position, using a front-wheel drive, front-wheel steering ARM as an example.

[0056] The AMR1 shown in Figure 8 is an example of a vehicle equipped with three wheels: a front wheel 2, a left rear wheel 3, and a right rear wheel 4, and a drive system 10 (see Figure 1 as appropriate) that drives and steers the front wheel 2. In Figure 8, δ represents the steering angle of the front wheel 2, μ represents the speed of the front wheel 2, and lw represents the wheelbase, i.e., the distance between the axis of the front wheel 2 and the axes of the left and right rear wheels 3 and 4.

[0057] In this case, the journey from the second stopping position to the target stopping position is performed using odometry alone because marker 100M is outside the field of view of the marker-capturing camera 30. Therefore, if the values ​​of the steering angle δ and the moving speed μ cannot be obtained accurately, the trajectory will deviate significantly from the expected value. In addition, the steering angle δ is susceptible to the effects of wheel mounting errors and other factors.

[0058] Therefore, in this embodiment, after estimating the self-position at the second stopping position SSP, the rudder angle δ is estimated so as to minimize the difference between the amount of movement based on the self-position estimation result from images taken of marker 100M at the first stopping position and the second stopping position SSP, and the amount of movement obtained from odometry during travel between the first and second stopping positions. Then, by using the estimated value of rudder angle δ in the odometry calculation from the second stopping position to the target stopping position, the deviation from the assumed trajectory can be reduced.

[0059] The rudder angle can be calculated specifically as follows.

[0060] As shown in Figure 8, the coordinate system fixed to the road surface is defined as a two-dimensional coordinate system of 0-xy, and the velocity and turning angular velocity of the left and right center positions (xc,yc) of the rear wheels 3 and 4 in the 0-xy coordinate system are expressed by the following equations 1, 2, and 3.

[0061]

number

[0062]

number

[0063]

number

[0064] Figure 9 illustrates the method for determining the rudder angle offset. As shown in Figure 9, the measurement value at the first stop position of AMR1 is assumed to be (xc1, yc1) on the XY coordinate axis and to have an angle of ψ1 with respect to the target trajectory. Furthermore, the measurement value at the second stop position SSP of AMR1 is assumed to be (xc2, yc2) on the XY coordinate axis and to have an angle of ψ2 with respect to the target trajectory.

[0065] In this case, the rudder angle offset amount can be determined by performing calculations using the following equations 4 to 8.

[0066] First, using the history of μ and δ from the second positioning data, the first stopping position and direction are calculated in reverse using the following equations 4, 5, and 6.

[0067]

number

[0068]

number

[0069]

number

[0070] Here, we assume that the actual rudder angle is the measured value of the rudder angle plus an offset error, and we use equation 7 to add the offset amount δoff to the observed value δm.

[0071]

number

[0072] Then, the rudder angle offset amount is determined so as to minimize the value of J in Equation 8 below, which represents the quadratic error between the position estimated by the reciprocal and the value actually obtained in the first positioning measurement.

[0073]

number

[0074] Thus, the AMR1 of this embodiment has a configuration that switches the method of following the target trajectory and path depending on the distance from the target stopping position TSP and marker 100M, and a configuration that estimates the steering angle from the position measurement at multiple points and the history of wheel rotation speed and steering angle between points.

[0075] In other words, in this embodiment, marker 100M is photographed while the AMR1 is stopped to estimate its relative position to marker 100M and, consequently, its own position, and this process is performed only the minimum number of times necessary. Furthermore, in this embodiment, the steering angle error is estimated from the position measurements at multiple points and the history of wheel rotation speed and steering angle between points, and a correction is made according to the estimation result.

[0076] With an AMR system configured in this way, the accuracy of estimating the vehicle's own position (i.e., the two-dimensional coordinate position and orientation of the vehicle) based on image recognition is improved, and consequently, the accuracy of the stopping position and orientation when AMR1 stops near marker 100M is improved.

[0077] Figure 10 is a flowchart showing the procedures for self-localization and switching of driving modes in the AMR1.

[0078] In step S1, the AMR1 activates the Lidar 60 and self-position estimation unit 35 described above in Figure 1, and begins autonomous driving using self-position estimation by laser rangefinder (LiDAR).

[0079] During autonomous driving in step S1, the AMR1 uses Lidar 60 to measure the three-dimensional positions of objects around it using a laser rangefinder (LiDAR) and records them in a storage medium of its choice. The self-position estimation unit 35 then compares the surrounding objects measured by Lidar 60 during autonomous driving with the records from the teaching drive (for example, those pre-stored in the memory of the self-position estimation unit 35) to estimate the AMR1's own position on the path. The path-following unit 20 then uses the AMR1's own position estimated by the self-position estimation unit 35 to output speed command values ​​and steering angle command values ​​to the drive unit 10 so that the AMR1 approaches the target path TW and its position is on the target path TW. By performing this path-driving mode operation, the AMR1 performs autonomous driving while correcting its position on the path.

[0080] In the subsequent step S2, the AMR1 controls the motor of the drive unit 10 via the path-following unit 20 so that it stops at a predetermined distance D (see Figure 7 as appropriate) upstream of the target stop position TSP, in this example, 2 m before the target stop position TSP.

[0081] In the next step S3 while stopped, the AMR1 captures a photograph of marker 100M with the marker-capturing camera 30, and the self-position estimation unit 31 analyzes the captured image (pattern) of marker 100M. The self-position estimation unit 31 measures the relative position to marker 100M from the captured image (pattern) of marker 100M and estimates its own position, that is, its two-dimensional coordinate position on the path (road surface).

[0082] In step S4, which remains stopped, the AMR1 uses the target path generation unit 32 described above to generate a linear target path STW, which is a target trajectory up to the vicinity of marker 100M, that is, a path leading to the waypoint where the target path TW and the extension line passing through a part of marker 100M intersect, as explained in Figure 6, and outputs the generated target path STW to the path following unit 20.

[0083] In step S5, the path-following unit 20 of the AMR1 outputs speed command values ​​and rudder angle command values ​​to the drive unit 10 so that the AMR1 follows the target trajectory (target path STW), starts odometry driving, and enters stop positioning mode.

[0084] During this odometry driving, the odometry calculation unit 34, regardless of the output of the steering angle offset estimation unit 33, uses the rotational speed supplied from the wheel rotation sensor 40 and the steering angle supplied from the steering angle sensor 50 to determine the amount of movement of the AMR1, i.e., the direction and length of movement, and estimates the position of the AMR1 from the cumulative calculation of this amount of movement. Then, regardless of the output of the target path generation unit 32, the path following unit 20 uses the position of the AMR1 (estimated value) supplied from the odometry calculation unit 34 to output speed command values ​​and steering angle command values ​​to the drive unit 10 so that the AMR1 drives following the target trajectory (target path STW).

[0085] In step S6, the path-following unit 20 outputs speed command values ​​and rudder angle command values ​​to the drive unit 10 so that the AMR1 stops at the position just before the marker 100M goes out of the frame of the captured image, or in other words, so that the AMR1 stops just before the marker 100M goes out of the field of view of the marker-capturing camera 30. Through this control and operation, the AMR1 stops at the second stopping position SSP, which is before the waypoint shown in Figure 7.

[0086] Then, in this second stopped state, AMR1 performs the following steps S7 and S8. In step S7 while stopped, AMR1 photographs marker 100M with the marker-capturing camera 30, analyzes the image (pattern) of the captured marker 100M to measure its relative position to marker 100M, and estimates its own position. This image measurement process is the same as in step S3 described above.

[0087] Furthermore, in step S8 while stopped, the odometry calculation unit 34 estimates the steering angle error and tire circumference error from the relative position with respect to the marker 100M measured in step S4 and the steering and rotational speed history of the front wheels 2 recorded during the odometry run in step S5. Based on these estimated values, it calculates a correction value for the steering angle of the front wheels 2 and outputs it to the path following unit 20. In addition, at this time, the target path generation unit 32 may generate a corrected path that matches the direction and position of the target path TW, based on the relative position with respect to the marker 100M measured in step S4 and the self-position of the AMR1 supplied by the self-position estimation unit 31, and output it to the path following unit 20.

[0088] In the following step S9, the path-following unit 20 of the AMR1 performs odometry driving by outputting speed command values ​​and steering angle command values ​​to the drive unit 10 so that the AMR1 moves along the target path TW to the target stop position TSP. In step S10, the path-following unit 20 controls the AMR1 to stop at the target stop position TSP.

[0089] By performing this type of control, the stopping position at the target stopping position TSP, and in this example, the relative position and orientation of the AMR1 with respect to the equipment 100, can be achieved with high precision.

[0090] Furthermore, in step S10, which occurs when the system is stopped, a predetermined task is performed. Here, the predetermined task is not particularly limited and includes various tasks that can be performed by this system, such as charging the AMR1 using a charging device provided on equipment 100, and loading and unloading cargo from the AMR1 onto equipment 100.

[0091] After step S10, AMR1 may, as appropriate, return to autonomous driving using LiDAR according to a pre-set route and program (see Figure 7), in which case the process returns to step S1. Alternatively, after step S10, AMR1 may terminate the series of processes.

[0092] Thus, in this embodiment, during the first journey to the first stop position FSP, where the distance from the target stop position TSP and the marker 100M located in its vicinity is still far, the AMR1 autonomously moves along the first target path TW while measuring and comparing the surrounding conditions using a laser scan (Lider 60). Then, while stopped at the first stop position FSP, the AMR1 images the marker 100M, recognizes (estimates) the position of the marker 100M and, consequently, its own relative position from the captured image, generates a second target path STW that approaches the marker 100M, and follows this second target path STW to the second stop position SSP using odometry. Then, while stopped at the second stopping position SSP, AMR1 again images marker 100M, recognizes the position of marker 100M and its relative position from the captured image, estimates the error (rudder angle offset) with the target path TW to the target stopping position TSP, performs odometry movement to correct the error, and stops at the target stopping position TSP.

[0093] This configuration allows for cost reduction of the cameras and hardware installed in the AMR1, while stabilizing the travel trajectory and minimizing the number of stops.

[0094] It should be noted that the present invention is not limited to the embodiments described above, and various modifications are included. For example, the embodiments described above are described in detail to make the present invention easier to understand, and are not necessarily limited to those having all the configurations described. Furthermore, it is possible to replace a part of the configuration of one embodiment with the configuration of another embodiment, and it is also possible to add the configuration of another embodiment to the configuration of one embodiment.

[0095] Furthermore, it is possible to add, delete, or replace some of the configurations in each embodiment with other configurations. In addition, some or all of the above configurations, functions, processing units, processing means, etc., may be implemented in hardware, for example, by designing them as integrated circuits. Alternatively, the above configurations, functions, etc., may be implemented in software by having a processor interpret and execute programs that realize each function. Information such as programs, tables, and files that realize each function can be stored in memory, a recording device such as a hard disk or SSD (Solid State Drive), or a recording medium such as an IC card, SD card, or DVD.

[0096] Furthermore, the AMR system according to the above-described embodiment improves the accuracy of the stopping position and orientation of the AMR1 when it stops, thereby improving work efficiency during the AMR1 stopping operation and reducing unnecessary processes or costs, thus reducing overall energy consumption. Therefore, it is possible to contribute to the United Nations-led Sustainable Development Goals (SDGs), particularly Goal 7 (Ensure access to affordable, reliable, sustainable and modern energy) and Goal 13 (Take urgent action to combat climate change and its impacts). [Explanation of Symbols]

[0097] 1...AMR, 10...Drive unit, 20...Path tracking unit, 30...Marker camera, 31...Self-position estimation unit, 32...Target path generation unit, 33...Rudder angle offset estimation unit, 34...Odometry calculation unit, 35...Self-position estimation unit, 40...Wheel rotation sensor, 50...Rudder angle sensor, 60...Lider, 100...Equipment, 100M...Marker, TSP...Target stopping position, TW...Target path (first target path), FSP...First stopping position, SSP...Second stopping position, STW...Second target path.

Claims

1. While measuring the surrounding conditions using laser scanning and comparing the measurement results with the stored surrounding conditions, the system autonomously moves along the first target path, moving to a first stop position before the target stop position. In the stopped state at the first stopping position, the marker is imaged, and the coordinate position and direction of travel of the autonomous mobile robot relative to the marker are measured. A second target path is generated from the first stop position to approach the marker, and the vehicle moves along the second target path from the first stop position to the second stop position, recording the history of rudder angle and speed, based solely on odometry information. In the stopped state at the second stopping position, the marker is imaged again, and the coordinate position and direction of travel of the autonomous mobile robot relative to the marker are remeasured. Based on the coordinate position and direction of travel measured at the second stopping position, and the history of rudder angle and speed recorded during movement from the first stopping position to the second stopping position, the estimated coordinate position and direction of travel at the first stopping position are calculated by reverse calculation. A method for controlling an autonomous mobile robot.

2. Assuming that the difference between the estimated value and the coordinate position and direction of travel measured at the first stopping position is due to the steering angle offset, the steering angle offset amount that minimizes this difference is calculated. Based on the calculated rudder angle offset, the odometry movement is corrected, and the vehicle stops at the target stopping position. A method for controlling an autonomous mobile robot according to claim 1.

3. An autonomous movement control unit that autonomously moves along a first target path to a first stop position while measuring the surrounding conditions using laser scanning and comparing the measurement results with the stored surrounding conditions, An imaging unit that images a marker at the first stop position, A self-position estimation unit estimates position coordinates and orientation based on a marker image acquired by the imaging unit, A target path generation unit generates a second target path that approaches the marker from the first stop position, When moving along the second target path to the second stopping position, an odometry calculation unit performs odometry calculations based on the steering angle and speed history acquired by the steering angle sensor and wheel rotation sensor, A rudder angle offset estimation unit calculates the position coordinates and orientation based on the marker image captured again at the second stop position, and the estimated value of the first stop position based on the history. Equipped with, Autonomous mobile robot.

4. A path-following unit corrects odometry movement based on the calculated rudder angle offset and stops at the target stopping position. Equipped with, The autonomous mobile robot according to claim 3.

Images