JP2017130006A - Autonomous mobile body control device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an autonomous mobile body control device capable of autonomously moving an autonomous mobile body with high accuracy in a movement area.SOLUTION: A CPU 1 of the invention calculates a first virtual value and calculates plural second virtual values by a stochastic method. The CPU 1 determines a main specific second virtual value from each second virtual value and replaces the main specific second virtual value with the first virtual value and sets the first virtual value to a determined value when a second dispersion state of each second estimation y coordinate does not exceed a second reference state, and a third dispersion state of each second estimation posture θ does not exceed a third reference state. On the other hand, the CPU 1 determines a sub specific second virtual value from each second virtual value and replaces the sub specific second virtual value with the first virtual value and sets the first virtual value to a determined value when a first dispersion state exceeds a first reference state, a second dispersion state exceeds a second reference state, or a third dispersion state exceeds a third reference state.SELECTED DRAWING: Figure 3

Description

  The present invention relates to an autonomous mobile control device.

  Conventionally, many techniques have been proposed in which a specific moving path is formed by laying a magnetic tape, a rail, or the like in a moving area, and a moving body is automatically moved along the moving path. On the other hand, in recent years, research on an autonomous mobile body that autonomously moves within a movement area without forming a specific movement route in the movement area has also been conducted. Such an autonomous mobile body is disclosed in Patent Document 1, for example.

  The autonomous mobile body disclosed in Patent Document 1 includes a drive device, a storage device, a self-movement amount detection device, and an autonomous mobile body control device. The drive device is composed of a plurality of wheels, motors, and the like. The self-movement amount detection device detects the movement amount of the autonomous mobile body in the movement area. The storage device stores map information of the moving area. The environmental information detection device detects environmental information around itself in the moving area. The autonomous mobile control device controls the drive device, the storage device, the self-movement amount detection device, and the environment information detection device.

  Specifically, the autonomous mobile body control device operates the environmental information detection device to detect environmental information, and generates geometric shape data based on the geometric features around the autonomous mobile body from the environmental information. Then, the geometric shape data and the map information read from the storage device are collated, and the coordinates and orientation of the location that most overlaps the geometric shape data in the map information are obtained by a probabilistic method. The autonomous mobile body control device recognizes the coordinates and posture obtained as described above as the current coordinates and posture of the autonomous mobile body in the moving area, and controls the driving device and the self-movement amount detection device to move autonomously. Move your body within the moving area. By repeating these processes, the autonomous mobile body control device autonomously moves the autonomous mobile body within the moving area while correcting the current coordinates and posture of the autonomous mobile body within the moving area.

  By the way, the moving area is not limited to an environment having only a large number of geometric features, and an environment having a large number of geometric features and an environment having few geometric features can be mixed. In an environment where such geometric features are scarce, it is difficult for the autonomous mobile control device to obtain the current coordinates and posture of the autonomous mobile by a probabilistic method. This is because a plurality of coordinates and postures are calculated by a probabilistic method in an environment where geometric features are scarce.

  Therefore, in the above autonomous mobile body control device, in an environment where there are many geometric features, the autonomous mobile body is autonomously moved by obtaining the coordinates and orientation of the autonomous mobile body in the moving area by a probabilistic method. On the other hand, in an environment where geometric features are scarce, the autonomous mobile body is moved in the direction of travel so as not to touch obstacles in the moving area detected from the environmental information without calculating the coordinates and posture of the autonomous mobile body using a probabilistic method. To move autonomously.

  As described above, the autonomous mobile body control device switches the control of the drive device, the storage device, the self-movement amount detection device, and the environment information detection device according to the environment in the movement area, thereby reaching the destination in the movement area. An autonomous mobile body can be moved autonomously.

JP 2014-67223 A

  However, in the conventional autonomous mobile control device, if the time or distance for which the autonomous mobile body is moved so as not to touch an obstacle becomes longer, the coordinates and posture of the autonomous mobile body in the movement area and the movement area The deviation from the true coordinates and posture of the autonomous mobile body becomes large. As a result, in this autonomous mobile body control device, there is a possibility that the autonomous mobile body cannot be autonomously moved with high accuracy within the movement area.

  The present invention has been made in view of the above-described conventional situation, and it is an issue to be solved to provide an autonomous mobile control apparatus capable of autonomously moving an autonomous mobile in a moving area with high accuracy. Yes.

The autonomous mobile body control device of the present invention is used for an autonomous mobile body that autonomously moves within a preset movement area,
The autonomous mobile body includes a driving device, a storage device that stores map information of the moving area, a self-moving amount detecting device that detects a moving amount of the self within the moving area, and a surrounding area of the self within the moving area. An environmental information detecting device for detecting environmental information of
An autonomous mobile body control device that controls the drive device, the storage device, the self-movement amount detection device, and the environment information detection device while recognizing the x-coordinate, y-coordinate, and posture of the autonomous mobile body in the movement area Because
A first calculation unit that calculates a first virtual value composed of a first estimated x-coordinate, a first estimated y-coordinate, and a first estimated attitude of the autonomous moving body based on a movement amount of the autonomous moving body;
A second computing unit that computes a plurality of second virtual values composed of a second estimated x-coordinate, a second estimated y-coordinate, and a second estimated attitude of the autonomous mobile body based on the environmental information by a probabilistic method;
Whether the first variance state of each second estimated x coordinate exceeds a first reference state, whether the second variance state of each second estimated y coordinate exceeds a second reference state, A determination unit that determines whether or not the third distributed state of the second estimated posture exceeds the third reference state;
Each of the first distributed state does not exceed the first reference state, the second distributed state does not exceed the second reference state, and the third distributed state does not exceed the third reference state. A main determination unit for determining a main specific second virtual value from the second virtual value;
A first output mode unit that executes a first output mode in which the main specific second virtual value is replaced with the first virtual value to be a definite value, and is output to the driving device;
When the first distributed state exceeds the first reference state, or when the second distributed state exceeds the second reference state, or the third distributed state exceeds the third reference state A sub-determination unit that determines a sub-specific second virtual value from each of the second virtual values;
And a second output mode unit that executes a second output mode in which the sub-specific second virtual value is replaced with the first virtual value to obtain a definite value and outputs the second output mode to the driving device.

  In the autonomous mobile body control device of the present invention, the first calculation unit calculates the first virtual value of the autonomous mobile body based on the movement amount detected by the self-movement amount detection device. Further, based on the environment information detected by the environment information detection device, the second calculation unit calculates a plurality of second virtual values of the autonomous mobile body by a probabilistic method. The second virtual value includes a second estimated x coordinate, a second estimated y coordinate, and a second estimated posture of the autonomous mobile body. Thus, in each 2nd virtual value calculated by the probabilistic method, each 2nd estimated x coordinate, each 2nd estimated y coordinate, and each 2nd estimated attitude | position are disperse | distributed in a fixed range, respectively. . This distributed state changes depending on the environment information detected by the environment information detecting device, that is, the geometric feature around the autonomous mobile body in the moving area.

  Here, in this autonomous mobile control device, the first distributed state of each second estimated x coordinate in each second virtual value does not exceed the first reference state, and the second distributed state of each second estimated y coordinate is the first. When the second reference state is not exceeded and the third distributed state of each second estimated posture does not exceed the third reference state, the main determination unit determines a main specific second virtual value from each second virtual value. The first output mode unit executes the first output mode. In the first output mode, the main specific second virtual value is replaced with the first virtual value to obtain a fixed value that is the x-coordinate, y-coordinate, and posture of the autonomous mobile body in the moving area, and this fixed value is driven. Output to the device. Thus, the autonomous mobile body control device autonomously moves the autonomous mobile body within the movement area.

  On the other hand, when the first dispersion state exceeds the first reference state, or when the second dispersion state exceeds the second reference state, or when the third dispersion state exceeds the third reference state, This means that it is difficult to uniquely obtain the x-coordinate, y-coordinate and attitude of the autonomous mobile body in the moving area by the probabilistic method. Therefore, in such a case, in the autonomous mobile control device, the sub-determining unit determines the sub-specific second virtual value from each second virtual value. Then, the second output mode unit executes the second output mode. In the second output mode, the sub-specific second virtual value is replaced with the first virtual value as a final value, and this final value is output to the driving device. Thus, the autonomous mobile body control device autonomously moves the autonomous mobile body within the movement area.

  Thus, in this autonomous mobile control apparatus, when there are many geometric features, not only the main specific second virtual value is determined from each second virtual value obtained by the probabilistic method, but also the geometric Even if the scientific characteristics are poor, the sub-specific second virtual value is determined from each second virtual value. Thereby, in this autonomous mobile body control device, while calculating the first virtual value based on the movement amount of the autonomous mobile body, the first virtual value is replaced with the main specific second virtual value or the sub-specific second virtual value. It can be a definite value. For this reason, in this autonomous mobile body control apparatus, it becomes possible to recognize the x coordinate, y coordinate, and attitude | position of an autonomous mobile body in a movement area with high precision.

  Therefore, according to the autonomous mobile body control device of the present invention, it is possible to autonomously move the autonomous mobile body with high accuracy within the movement area.

  The autonomous mobile control device of the present invention determines a first virtual value based on a fixed value, a third calculator that calculates a displacement amount between the fixed value and the first virtual value, and if the displacement does not exceed a threshold value. A third output mode that executes the third output mode as a value and outputs the third output mode, and a correction mode that executes the first output mode or the second output mode if the displacement exceeds a threshold value. Is preferred.

  When the displacement amount of the first virtual value does not exceed the threshold value with respect to the fixed value obtained in the first output mode or the second output mode, that is, the x coordinate of the autonomous mobile body in the moving area based on the fixed value, When the accuracy of the first estimated x coordinate, the first estimated y coordinate, and the first estimated posture is high with respect to the y coordinate and the posture, the first virtual value is further set as the main specific second virtual value or the sub specific second virtual. There is little need to replace the value with the final value. Therefore, when the displacement amount of the first virtual value with respect to the confirmed value does not exceed the threshold value, the autonomous mobile body control device reduces the processing load by using the first virtual value as it is as the confirmed value. Can be moved autonomously.

  The main determining unit preferably determines the main specific second virtual value by weighting each second virtual value and a matching rate between each second virtual value and the first virtual value. Similarly, it is preferable that the sub-determining unit determines the sub-specific second virtual value by weighting each second virtual value and the matching rate between each second virtual value and the first virtual value. In these cases, the optimum value among the second virtual values can be the main specific second virtual value and the sub specific second virtual value. For this reason, it becomes possible to obtain the determined value with higher accuracy.

  The probabilistic method is preferably a particle filter based on the SLAM algorithm. In this case, it is possible to calculate a plurality of second virtual values with high accuracy relatively easily.

  The autonomous mobile body is preferably an industrial vehicle that travels autonomously. In moving areas such as factories and warehouses where industrial vehicles run autonomously, the layout often changes, so the geometrical characteristics of the environment around the industrial vehicles are often poor. It may be difficult to uniquely determine the coordinates, y-coordinate, and orientation. For this reason, if the autonomous mobile body control apparatus of this invention is used for an industrial vehicle, it will become possible to make an industrial vehicle autonomously drive with high precision within a movement area, and it will become possible to improve work efficiency.

  According to the autonomous mobile body control device of the present invention, it is possible to autonomously move the autonomous mobile body with high accuracy within the movement area.

FIG. 1 is a schematic diagram illustrating an industrial vehicle in which the autonomous mobile control device of the embodiment is employed. FIG. 2 is a schematic diagram showing a state in which an industrial vehicle autonomously travels within a moving area. FIG. 3 is a control flow when the autonomous mobile control device of the embodiment causes the industrial vehicle to autonomously travel within the movement area. FIG. 4 is a schematic diagram showing the state of dispersion of specific particles at point A in FIG. FIG. 5 is a list showing the second estimated x-coordinate, the second estimated y-coordinate, and the second estimated posture of each specific particle based on the dispersion state of the specific particle in FIG. FIG. 6 is a graph showing the distribution of each second estimated x-coordinate based on the dispersion state of the specific particles in FIG. FIG. 7 is a graph showing the distribution of each second estimated y coordinate based on the dispersion state of the specific particles in FIG. FIG. 8 is a schematic diagram showing the state of dispersion of specific particles at point B in FIG. FIG. 9 is a list showing the second estimated x-coordinate, the second estimated y-coordinate, and the second estimated posture of each specific particle based on the dispersion state of the specific particle in FIG. FIG. 10 is a graph showing the distribution of each second estimated x coordinate based on the dispersion state of the specific particles in FIG. FIG. 11 is a graph showing the distribution of each second estimated y coordinate based on the dispersion state of the specific particles in FIG.

  DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiments embodying the present invention will be described below with reference to the drawings.

  As shown in FIG. 1, the CPU 1 of the embodiment is mounted on an automatic transport vehicle 3. The CPU 1 is an example of the autonomous mobile control device of the present invention. The automated guided vehicle 3 is an industrial vehicle that can autonomously travel in the warehouse 5 shown in FIG. 2 without driving by an operator, and is an example of the autonomous mobile body of the present invention. The warehouse 5 is an example of a moving area according to the present invention.

  In the warehouse 5, first to fourth chambers 51 to 54 having different sizes are partitioned by a plurality of wall surfaces 50, and a corridor 55 connected to the first to fourth chambers 51 to 54 is partitioned. In the first to fourth chambers 51 to 54, a work desk 57 and the like are arranged in addition to various luggage shelves 56.

  The corridor 55 includes first to fourth corridors 55a to 55d. The first corridor 55a is connected to the first chamber 51 and extends in the north-south direction. The second hallway 55b is connected to the third chamber 53, and is connected to the first hallway 55a, the third hallway 55c, and the fourth hallway 55d, and extends in the east-west direction. The third hallway 55c is connected to the fourth chamber 54 and to the second hallway 55b, and extends in the north-south direction. The fourth hallway 55d is connected to the second chamber 52 and to the second hallway 55b, and extends in the north-south direction.

  As shown in FIG. 1, the automatic guided vehicle 3 includes a plurality of drive wheels 7, a ROM 9, a RAM 11, an encoder 13, an external sensor 15, and an information input unit 17 in addition to the CPU 1 described above. Although not shown, the automated guided vehicle 3 includes a known steering device and an electric motor.

  Each drive wheel 7 constitutes a drive device 19 together with a steering device, an electric motor, and the like, so that the automatic guided vehicle 3 can run in the warehouse 5.

  The ROM 9 stores a control program for autonomously driving the automatic guided vehicle 3. Specifically, the control program is a SLAM (Simultaneous Localization and Mapping) algorithm, and executes a particle filter as a probabilistic method.

  The RAM 11 is an example of a storage device in the present invention, and stores map information of the warehouse 5. This map information is coordinate information in the warehouse 5 when the east-west direction of the warehouse 5 shown in FIG. 2 is the x axis and the north-south direction of the warehouse 5 is the y axis. Further, the RAM 11 shown in FIG. 1 stores first to third reference states and threshold values described later. Further, the RAM 11 can temporarily store the first and second virtual values, the final value, and the like in addition to the route information input to the information input unit 17 described later.

  The encoder 13 is an example of a self-movement amount detection device according to the present invention, and can detect the movement amount of the automatic guided vehicle 3 in the warehouse 5 by measuring the rotation speed of each drive wheel 7 and the steering angle of the steering device. It has become. The encoder 13 may be configured to detect the amount of movement of the automatic guided vehicle 3 in the warehouse 5 by measuring the number of rotations of the electric motor. Further, instead of the encoder 13, a GPS device or the like may be adopted as a self-movement amount detection device.

  The external sensor 15 is an example of an environmental information detection device according to the present invention, and can detect environmental information around the automated guided vehicle 3 in the warehouse 5. Specifically, the external sensor 15 is attached to the lower part of the automatic transport vehicle 3 and irradiates a laser in a certain range in the traveling direction of the automatic transport vehicle 3 as shown by hatching in FIG. Then, the geometric features around the automatic transport vehicle 3 based on the wall surface 50 and the luggage rack 56 are detected as environmental information by the laser reflected on the wall surface 50 and the luggage rack 56 and returned to the external sensor 15. At the same time, the external sensor 15 measures the distance from the wall surface 50 or the like to the automatic transport vehicle 3.

  The information input unit 17 includes an operation display, a keyboard, and the like, and an operator can input route information such as the current location and the destination of the automatic transport vehicle 3.

  The CPU 1 controls the driving device 19, the ROM 9, the RAM 11, the encoder 13, the external sensor 15, and the information input unit 17 to make the automatic conveyance vehicle 3 autonomously travel. Specifically, the CPU 1 includes a first calculation unit, a second calculation unit, a third calculation unit, a determination unit, a main determination unit, a sub determination unit, a first output mode unit, a second output mode unit, Each calculation process is performed as a three-output mode section and a correction mode section. Thereby, CPU1 makes the automatic conveyance vehicle 3 autonomously run in the warehouse 5, performing the 1st output mode or the 2nd output mode or the 3rd output mode mentioned later.

  Next, the automatic transport vehicle 3 in the warehouse 5 is taken as an example in the case where the automatic transport vehicle 3 transports the load (not shown) in the load shelf 56 in the first chamber 51 to the load shelf 56 in the fourth chamber 54. Each arithmetic processing of CPU1 at the time of making it run autonomously is explained concretely.

  As shown in FIG. 3, when autonomously driving the automatic guided vehicle 3, first, the CPU 1 automatically performs automatic processing based on the map information of the warehouse 5 stored in the RAM 11 and the route information input from the information input unit 17. The coordinates on the map information of the current position of the transport vehicle 3 and the fourth chamber 54 as the destination are calculated. And CPU1 starts the driving | running | working of the automatic conveyance vehicle 3 toward the 4th chamber 54 by controlling the drive device 19 (step S101).

  As the automatic conveyance vehicle 3 travels, the encoder 13 detects the amount of movement of the automatic conveyance vehicle 3 by measuring the rotation speed of each drive wheel 7 and the steering angle of the steering device. And based on this moving amount, CPU1 calculates the 1st virtual value which consists of the 1st estimated x coordinate of the automatic conveyance vehicle 3, 1st estimated y coordinate, and 1st estimated attitude | positioning (step S102). This posture θ is an inclination in the y-axis direction with respect to the x-axis direction in the traveling direction of the automatic conveyance vehicle 3. Note that the first virtual value calculated by the CPU 1 is stored in the RAM 11.

  Further, when the automatic transport vehicle 3 travels, the external sensor 15 detects a geometric feature around the automatic transport vehicle 3 (step S103). Then, the CPU 1 calculates a plurality of second virtual values composed of the second estimated x-coordinate, the second estimated y-coordinate, and the second estimated posture θ of the automatic transport vehicle 3 by the particle filter based on the geometric feature. Specifically, the CPU 1 disperses a large number of particles that are virtual bodies of the automatic transport vehicle 3 around the automatic transport vehicle 3. From each particle, a plurality of specific particles that match the geometric feature detected by the external sensor 15 with a certain likelihood are extracted. From the x-coordinate, y-coordinate, and orientation θ of each specific particle thus extracted, the above-described second virtual values, that is, each second estimated x-coordinate, each second estimated y-coordinate, and each second estimated orientation θ are calculated. (Step S104). Each second virtual value is also stored in the RAM 11.

  Next, the CPU 1 determines whether or not the first output mode or the second output mode has been executed based on the information stored in the RAM 11 (step S105). If the first output mode or the second output mode has been executed (step S105: YES), the CPU 1 is based on the determined value obtained in the first output mode or the second output mode that has already been performed. Then, the displacement amount of the first virtual value is calculated (step S106). Specifically, the displacement amount is calculated by calculating the first estimated x-coordinate, the first estimated y-coordinate, the first estimated attitude θ, and the x-coordinate of the current automated guided vehicle 3 in the warehouse 5 based on the determined value, y This is done by comparing the coordinates and orientation θ. Details of the first output mode, the second output mode, and the final value will be described later.

  With respect to the displacement amount of the first virtual value calculated in this way, the CPU 1 determines whether or not the threshold value preset and stored in the RAM 11 is exceeded (step S107). Here, if the displacement amount does not exceed the threshold value (step S107: NO), the first estimated x coordinate, the first coordinate with respect to the current x coordinate, y coordinate, and attitude θ of the automated guided vehicle 3 in the warehouse 5 are set. This means that the accuracy of the estimated y coordinate and the first estimated posture θ is high. For this reason, the CPU 1 executes the third output mode (step S108). In the third output mode, the CPU 1 uses the first estimated x coordinate, the first estimated y coordinate, and the first estimated attitude θ of the first virtual value as the current x coordinate, y coordinate, and attitude θ of the warehouse 5 automatic transport vehicle 3. Let it be a certain fixed value. Thus, the PCU 1 recognizes the current x-coordinate, y-coordinate and attitude θ of the automated guided vehicle 3 in the warehouse 5. Then, the CPU 1 stores the determined value in the RAM 11 and outputs it to the driving device 19 to cause the automatic transport vehicle 3 to autonomously travel. In the third output mode, the previous determined value obtained in the first output mode or the second output mode is not erased from the RAM 11.

  On the other hand, when the displacement amount exceeds the threshold value (step S107: YES), the first estimated x-coordinate and the first estimated value are obtained with respect to the determined value obtained in the first output mode or the second output mode already performed. This means that the difference between the y coordinate and the first estimated orientation θ is large, and the accuracy of the first virtual value is low. If the first virtual value is a fixed value up to such a case, the x-coordinate, y-coordinate and attitude θ of the automatic transport vehicle 3 in the warehouse 5 recognized by the CPU 1 and the current state of the automatic transport vehicle 3 in the warehouse 5 The error between the x coordinate, the y coordinate, and the orientation θ increases. Therefore, when the displacement amount exceeds the threshold value, the CPU 1 does not execute the third output mode, but executes a process after step S109, which will be described later, as the correction mode unit, and sets the first virtual value to the second virtual value. An arithmetic process is performed to correct the first virtual value by obtaining a fixed value by replacing it with a value. In addition, even when the first output mode or the second output mode has never been executed (i.e., immediately after the automatic guided vehicle 3 is autonomously driven) (step S105: NO), the CPU 1 performs the above step S106. The processes after step S109 are executed without performing the processes of .about.S108.

  By the way, since each specific particle extracted in the above step S104 is dispersed within a certain range, each second estimated value calculated by the particle filter has each second estimated x coordinate and each second estimated y coordinate. And the second estimated postures θ are dispersed within a certain range. This distributed state changes depending on the geometric feature around the automatic guided vehicle 3 detected by the external sensor 15.

  Specifically, in the first to fourth chambers 51 to 54 shown in FIG. 2, a work desk 57 and the like are arranged in addition to many luggage shelves 56 in the room. There are many features. Thus, in an environment where there are many geometric features around the automatic transport vehicle 3, the accuracy of each second virtual value calculated by the particle filter is high. Therefore, for example, if the particle filter is executed at the point A of the first chamber 51 where the x coordinate, the y coordinate, and the orientation θ on the map information are “xA: yA: θA”, as shown in FIG. The specific particles P <b> 1 to P <b> 10 are dispersed in a range close to the point A. That is, in an environment where there are many geometric features, such as point A in the first chamber 51, the first distributed state, which is the distributed state of each second estimated x coordinate, and the distributed state of each second estimated y coordinate. The second dispersion state and the third dispersion state, which is the dispersion state of each second estimated posture θ, become smaller.

  On the other hand, in the environment where only the wall surface 50 exists around the automatic transport vehicle 3 as in the corridor 55 shown in FIG. 2, the geometric features detected by the external sensor 15 are poor. Thus, in an environment where the geometric features around the automatic transport vehicle 3 are scarce, the accuracy of each second virtual value calculated by the particle filter is low. Therefore, for example, if the particle filter is executed at the point B of the second corridor 55b where the x coordinate, the y coordinate, and the orientation θ on the map information are “xB: yB: θB”, as shown in FIG. The specific particles P <b> 1 to P <b> 10 are dispersed in a range far from the point B. In other words, in an environment where geometric features are scarce, the first distributed state of each second estimated x coordinate, the second distributed state of each second estimated y coordinate, and each 2 The third distributed state of the estimated posture θ increases. In particular, in an environment where only the wall surface 50 exists continuously in the east-west direction of the warehouse 5 as in the second corridor 55b, the first distribution state of the second estimated x-coordinate which is the traveling direction of the automatic transport vehicle 3 becomes large. . In addition, the dispersion state of each specific particle P1-P10 shown in FIG.4 and FIG.8 is an example, Even if it exists in the 1st-4th chambers 51-54, the geometric feature detected by the external sensor 15 is scarce. There may be cases. Similarly, even in the hallway 55, the external sensor 15 may detect a large number of geometric features.

  Therefore, as illustrated in FIG. 3, the CPU 1 determines whether or not the first distribution state of each second estimated x coordinate exceeds the first reference state, and the second distribution state of each second estimated y coordinate is the second reference state. It is determined whether or not the state has been exceeded and whether or not the third distributed state of each second estimated posture θ has exceeded the third reference state (step S109).

  In determining whether the first dispersion state exceeds the first reference state, in this embodiment, first, the CPU 1 exists at an absolute value of 2 or more away from the average value of each second estimated x coordinate. The ratio of the second estimated x coordinate is obtained. And about the ratio of the 2nd estimated x coordinate calculated | required in this way, CPU1 makes the 1st reference state made into "the ratio of the 2nd estimated x coordinate which exists 2 or more away in absolute value below 50%". Judge whether it meets. This first reference state is preset and stored in the RAM 11. Here, if the ratio of the second estimated x-coordinates that are separated by 2 or more in absolute value is less than 50%, the CPU 1 determines that the first distributed state does not exceed the first reference state. Note that the absolute values and the numerical values in the first reference state are examples, and other numerical values can be used.

  Similarly, even when determining whether or not the second dispersion state exceeds the first reference state, the CPU 1 determines that the second estimated y existing at an absolute value of 2 or more from the average value of each second estimated y coordinate. Find the percentage of coordinates. Whether or not the ratio of the second estimated y-coordinate obtained in this way satisfies the second reference state that “the ratio of the second estimated y-coordinate existing at an absolute value of 2 or more is 30% or less” Judge whether or not. This second reference state is also set in advance and stored in the RAM 11. Here, if the ratio of the second estimated y-coordinates that are separated by 2 or more in absolute value is less than 30%, the CPU 1 determines that the second distributed state does not exceed the second reference state. Note that the absolute values and the numerical values in the second reference state are examples, and other numerical values can be used.

  Further, even when determining whether or not the third dispersion state exceeds the third reference state, the CPU 1 keeps the second estimated posture θ existing at an absolute value of 3 or more away from the average value of each second estimated posture θ. Find the percentage of Whether or not the ratio of the second estimated attitude θ obtained in this way satisfies the third reference state in which “the ratio of the second estimated attitude θ that is separated by 3 or more in absolute value is 10% or less” Judge whether or not. This third reference state is also set in advance and stored in the RAM 11. Here, if the ratio of the second estimated posture θ that is separated by 3 or more in absolute value is less than 10%, the CPU 1 determines that the third distributed state does not exceed the third reference state. Note that the absolute values and the numerical values in the third reference state are examples, and other numerical values can be used.

  Based on these, it is determined whether or not the first to third dispersion states exceed the first to third reference states. Specifically, at the point A, as shown in FIGS. 5 and 6, the average value of the second estimated x-coordinates when the value of the x-coordinate (xA) at the point A is zero is “−0.2. " Since the number of second estimated x-coordinates that are separated from the average value by 2 or more in absolute value is one, the ratio is 10%. For this reason, the CPU 1 determines that at the point A, the first distributed state does not exceed the first reference state. In this determination, the fractional part of the average value is rounded up.

  Further, as shown in FIGS. 5 and 7, the average value of the second estimated y coordinates when the value of the y coordinate (yA) of the point A is zero is “zero”. Then, since the number of second estimated y-coordinates that are separated from the average value by two or more in absolute value is two, the ratio is 20%. For this reason, the CPU 1 determines that the second dispersion state does not exceed the second reference state at the point A.

  Further, as shown in FIG. 5, the average value of the second estimated postures θ when the value of the posture θ (θA) at point A is zero is “0.05”. Then, there is no second estimated posture θ that is separated from this average value by 3 or more in absolute value. For this reason, at the point A, the CPU 1 determines that the third distributed state does not exceed the third reference state.

  Thus, at the point A, the first distribution state of each second estimated x coordinate in each second virtual value does not exceed the first reference state, and the second distribution state of each second estimated y coordinate is The second reference state is not exceeded, and the third distributed state of each second estimated posture θ does not exceed the third reference state (step S109: NO).

  In this case, the CPU 1 determines the main specific second virtual value from each second virtual value calculated by the particle filter, that is, from each second estimated x coordinate, each second estimated y coordinate, and each second estimated posture θ. (Step S110). The main specific second virtual value is determined by the CPU 1 weighing each second virtual value and the matching rate between each second virtual value and the first virtual value. The main specific second virtual value is stored in the RAM 11.

  Then, the CPU 1 executes the first output mode (step S111). In the first output mode, the CPU 1 sets a fixed value by replacing the main specific second virtual value with the first virtual value. That is, the CPU 1 determines that the second estimated x coordinate, the second estimated y coordinate, and the second estimated attitude θ in the main specific second virtual value are the current x coordinate, y coordinate, and attitude θ of the automatic transport vehicle 3 in the warehouse 5. Recognize that there is. Then, the CPU 1 stores the determined value in the RAM 11 and outputs it to the driving device 19 to cause the automatic transport vehicle 3 to autonomously travel. Here, when the previous confirmed value is already stored in the RAM 11, the newly obtained confirmed value is overwritten. The CPU 1 also stores in the RAM 11 that the first output mode has been executed.

  On the other hand, at the point B, as shown in FIGS. 9 and 10, the average of the second estimated x-coordinates when the value of the x-coordinate (xB) at the point B is zero is “zero”. The number of second estimated x-coordinates that are separated from the average value by 2 or more in absolute value is seven, and the ratio is 70%. For this reason, CPU1 judges that the 1st distribution state has exceeded the 1st standard state in B point.

  As shown in FIGS. 9 and 11, the average of the second estimated y coordinates when the value of the y coordinate (yB) at the point B is zero is “zero”. Then, since the number of second estimated y-coordinates that are separated from the average value by two or more in absolute value is two, the ratio is 20%. For this reason, the CPU 1 determines that the second dispersion state does not exceed the second reference state at the point B.

  Furthermore, as shown in FIG. 9, the average of the second estimated postures θ when the value of the posture θ (θB) at point B is zero is “0.05”. Then, there is no second estimated posture θ that is separated from this average value by 3 or more in absolute value. For this reason, at the point B, the CPU 1 determines that the third dispersion state does not exceed the third reference state.

  In this way, at the point B, in each second virtual value, the second distribution state of each second estimated y coordinate does not exceed the second reference state, and the third distribution state of each second estimated posture θ. However, the first distributed state of each second estimated x-coordinate exceeds the first reference state (step S109: YES).

  In this case, the CPU 1 determines a sub-specific second virtual value from each second estimated y coordinate and each second estimated posture θ among the second virtual values calculated by the particle filter (step S112). That is, unlike the case where the main specific second virtual value is determined, each second estimated x-coordinate in which the first distributed state exceeds the first reference state is excluded in determining the sub specific second virtual value. The Further, even when determining the sub-specific second virtual value, the CPU 1 performs weighting of each second virtual value and the matching rate between each second virtual value and the first virtual value. This sub-specific second virtual value is stored in the RAM 11.

  Then, the CPU 1 executes the second output mode (step S113). In the second output mode, the CPU 1 replaces the sub-specific second virtual value with the first virtual value, thereby setting a fixed value that is the current x-coordinate, y-coordinate, and attitude θ of the automated guided vehicle 3 in the warehouse 5. . Here, the second estimated x-coordinates are excluded in determining the sub-specific second virtual value. Therefore, the CPU 1 recognizes that the second estimated y-coordinate and the second estimated attitude θ in the sub-specific second virtual value are the current y-coordinate and attitude θ of the automated guided vehicle 3 in the warehouse 5, and the first virtual The first estimated x coordinate in the value is recognized as the current x coordinate of the automated guided vehicle 3 in the warehouse 5. Then, the CPU 1 stores the determined value in the RAM 11 and outputs it to the driving device 19 to cause the automatic transport vehicle 3 to autonomously travel. Also in this case, if the previous confirmed value is already stored in the RAM 11, the newly obtained confirmed value is overwritten. Further, the CPU 1 also stores in the RAM 11 that the second output mode has been executed.

  As described above, at point B, only the first dispersion state exceeds the first reference state, but at other points with poor geometric features, for example, only the second dispersion state is the first. In some cases, the result may exceed two reference states. In such a case, in determining the sub-specific second virtual value, each second estimated y-coordinate is excluded, and the sub-specific second virtual value is determined from each second estimated x-coordinate and each second estimated orientation θ. The Rukoto. Similarly, when the third dispersion state exceeds the third reference state, each second estimated posture θ is excluded when determining the sub-specific second virtual value.

  Thus, if the automatic guided vehicle 3 has not reached the fourth chamber 54 as the destination (step S114: NO), the CPU 1 repeats the above steps S102 to S113 to thereby change the first to third output modes. Do one. Thereby, as shown by the arrow in FIG. 2, the CPU 1 reaches the second hallway 55b from the first chamber 51 through the first hallway 55a, and moves from the second hallway 55b to the third hallway 55c. It is possible to autonomously travel to the fourth chamber 54 through a route passing through to the fourth chamber 54. And if the automatic conveyance vehicle 3 arrives at the 4th chamber 54 (step S114: YES), CPU1 will complete | finish control of the automatic conveyance vehicle 3. FIG.

  As described above, in the CPU 1, when there are many geometric features around the automatic transport vehicle 3 like the point A of the first chamber 51, the CPU 1 calculates the main value from each second virtual value obtained by the particle filter. In addition to determining the specific second virtual value, even if the geometric features around the automatic transport vehicle 3 are poor, such as point B in the second corridor 55b, the secondary specific value is determined from each second virtual value. A second virtual value is determined. Thereby, in this CPU1, while calculating the 1st virtual value based on the movement amount of the automatic conveyance vehicle 3, this 1st virtual value is replaced with the main specific 2nd virtual value or the subspecific 2nd virtual value, can do.

  In particular, in the automated guided vehicle 3 in which the CPU 1 is used, in the warehouse 5 or the like that automatically runs, there are many cases where the geometric features of the surrounding environment are poor as in the second corridor 55b. Even in such a case, the CPU 1 can replace the first virtual value with the sub-specific second virtual value to obtain the determined value. Here, by replacing the first virtual value with the sub-specific second virtual value and determining the final value, the CPU 1 automatically transfers the second estimated y coordinate and the second estimated attitude θ in the sub-specific second virtual value in the warehouse 5. Although recognized as the current y coordinate and attitude θ of the vehicle 3, the current x coordinate of the automated guided vehicle 3 in the warehouse 5 is recognized as the first estimated x coordinate in the first virtual value. In this way, when the first virtual value is replaced with the sub-specific second virtual value and determined, the first virtual value that has not been corrected, in other words, the displacement amount has a threshold value. The exceeded first virtual value is included. However, even in this case, the true x-coordinate, y-coordinate, and orientation θ of the automatic guided vehicle 3 in the warehouse 5 are compared with the case where only the first virtual value whose displacement exceeds the threshold value is used as the fixed value. And the error can be reduced. As a result, in this CPU 1, not only when there are many geometric features around the automatic transport vehicle 3 detected by the external sensor 15, but also when the geometric features around the automatic transport vehicle 3 are poor. However, the x-coordinate, y-coordinate and attitude θ of the automated guided vehicle 3 in the warehouse 5 can be recognized with high accuracy. For this reason, the CPU 1 can improve the working efficiency of the automatic guided vehicle 3.

  Therefore, according to the CPU 1 of the present invention, the automatic guided vehicle 3 can be autonomously driven with high accuracy in the warehouse 5.

  In particular, if the CPU 1 has already executed the first output mode or the second output mode and the RAM 11 stores the determined value obtained at that time, the displacement of the first virtual value based on the determined value. Calculate the quantity. If the displacement amount does not exceed the threshold value, the third output mode in which the first virtual value is the final value is executed. As a result, when the accuracy of the first virtual value is high and it is not necessary to replace the first virtual value with the main specific second virtual value or the sub-specific second virtual value to make the final value, the processing load is reduced. It is possible to make the automatic guided vehicle 3 autonomously travel while reducing the above.

  Further, the CPU 1 calculates each second virtual value by a particle filter based on the SLAM algorithm. Therefore, the CPU 1 can calculate a plurality of second virtual values with high accuracy relatively easily.

  Further, when determining the main specific second virtual value and the sub-specific second virtual value, the CPU 1 matches each second virtual value calculated by the particle filter and the matching rate between each second virtual value and the first virtual value. And are weighted. For this reason, among the second virtual values, the optimum value can be the main specific second virtual value or the sub specific second virtual value, and the CPU 1 can obtain the determined value with higher accuracy. Yes.

  While the present invention has been described with reference to the embodiments, it is needless to say that the present invention is not limited to the above-described embodiments and can be appropriately modified and applied without departing from the spirit thereof.

  For example, the CPU 1 may be provided outside the automatic conveyance vehicle 3, and the CPU 1 may be configured to remotely control the drive device 19, the ROM 9, the RAM 11, the encoder 13, the external sensor 15, and the information input unit 17 by wireless communication.

  In the embodiment, one CPU 1 includes a first calculation unit, a second calculation unit, a third calculation unit, a determination unit, a main determination unit, a sub determination unit, a first output mode unit, a second output mode unit, a third Each calculation process is performed as an output mode section and a correction mode section. However, the present invention is not limited to this, and a plurality of CPUs 1 may be provided, and each CPU 1 may share the above arithmetic processing.

  Further, the CPU 1 may calculate a plurality of second virtual values by a probabilistic method other than the particle filter, or may calculate each second virtual value by combining a plurality of probabilistic methods.

  Moreover, the automatic conveyance vehicle 3 may be comprised so that switching between the case where it carries out said autonomous driving | running | working and the case where it drive | works by a driving | operation of an operator is possible.

  The present invention can be used for robots, automobiles, industrial vehicles, and the like capable of autonomous movement.

1 ... CPU (autonomous mobile body control device, first calculation unit, second calculation unit, third calculation unit, determination unit, main determination unit, sub determination unit, first output mode unit, second output mode unit, third Output mode section, correction mode section)
3 ... Automatic transportation vehicles (autonomous vehicles, industrial vehicles)
5 ... Warehouse (moving area)
11 ... RAM (storage device)
13 ... Encoder (Self-movement amount detection device)
15 ... External sensor (environmental information detection device)
19 ... Drive device

Claims (6)

Used for autonomous moving bodies that move autonomously within a preset moving area,
The autonomous mobile body includes a driving device, a storage device that stores map information of the moving area, a self-moving amount detecting device that detects a moving amount of the self within the moving area, and a surrounding area of the self within the moving area. An environmental information detecting device for detecting environmental information of
An autonomous mobile body control device that controls the drive device, the storage device, the self-movement amount detection device, and the environment information detection device while recognizing the x-coordinate, y-coordinate, and posture of the autonomous mobile body in the movement area Because
A first calculation unit that calculates a first virtual value composed of a first estimated x-coordinate, a first estimated y-coordinate, and a first estimated attitude of the autonomous moving body based on a movement amount of the autonomous moving body;
A second computing unit that computes a plurality of second virtual values composed of a second estimated x-coordinate, a second estimated y-coordinate, and a second estimated attitude of the autonomous mobile body based on the environmental information by a probabilistic method;
Whether the first variance state of each second estimated x coordinate exceeds a first reference state, whether the second variance state of each second estimated y coordinate exceeds a second reference state, A determination unit that determines whether or not the third distributed state of the second estimated posture exceeds the third reference state;
Each of the first distributed state does not exceed the first reference state, the second distributed state does not exceed the second reference state, and the third distributed state does not exceed the third reference state. A main determination unit for determining a main specific second virtual value from the second virtual value;
A first output mode unit that executes a first output mode in which the main specific second virtual value is replaced with the first virtual value to be a definite value, and is output to the driving device;
When the first distributed state exceeds the first reference state, or when the second distributed state exceeds the second reference state, or the third distributed state exceeds the third reference state A sub-determination unit that determines a sub-specific second virtual value from each of the second virtual values;
An autonomous mobile body comprising: a second output mode unit that executes a second output mode in which the sub-specific second virtual value is replaced with the first virtual value to obtain a definite value and outputs the second output mode to the driving device. Control device. The autonomous mobile body control device, based on the determined value, a third calculation unit that calculates a displacement amount of the determined value and the first virtual value;
If the displacement amount does not exceed a threshold value, a third output mode unit that executes a third output mode in which the first virtual value is a definite value and outputs the third output mode to the drive device;
The autonomous mobile control apparatus according to claim 1, further comprising: a correction mode unit that executes the first output mode or the second output mode if the displacement amount exceeds the threshold value.   The main determination unit determines the main specific second virtual value by weighting each second virtual value and a matching rate between each second virtual value and the first virtual value. Or the autonomous mobile body control apparatus of 2 description.   The said sub determination part determines the said sub specific 2nd virtual value by weighting each said 2nd virtual value and the matching rate of each said 2nd virtual value and said 1st virtual value. The autonomous mobile body control apparatus of any one of thru | or 3.   The autonomous mobile control apparatus according to claim 1, wherein the probabilistic method is a particle filter based on a SLAM algorithm.   The autonomous mobile body control device according to any one of claims 1 to 5, wherein the autonomous mobile body is an industrial vehicle that travels autonomously.

Images