Self-position estimation device

Abstract

To provide a self-position estimation device capable of stably performing self-position estimation.SOLUTION: A self-position estimation device comprises a self-position estimation unit and a determination unit. The self-position estimation unit estimates the self-position of a mobile body moving autonomously on the basis of two-dimensional map information including a plurality of two-dimensional maps in a height direction or a traveling direction within a movement range of the mobile body and sensor information obtained by detecting the periphery of the mobile body. The determination unit determines one of the plurality of two-dimensional maps used by the self-position estimation unit for estimation and a detection area of the sensor information on the basis of predetermined conditions.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present disclosure relates to a self-position estimation device that estimates the self-position of a moving body that moves autonomously. [Background technology]

[0002] A moving body that moves autonomously to a destination while grasping the surrounding environment is called an autonomous moving body. An autonomous moving body holds in advance environmental map information that is used when determining whether or not the autonomous moving body can move in a specific space, observes the surroundings while moving, and generates a moving route from the current location to the destination while avoiding obstacles in the vicinity. The autonomous moving body then travels to the destination along the generated moving route. When traveling along the moving route, the autonomous moving body autonomously travels through points (e.g., destination, waypoint) included in the moving route while matching the environmental map information with sensor information that detects the surroundings to estimate its own position.

[0003] The autonomous mobile body moves autonomously while estimating its position using sensor information, even in places where people or obstacles are present, and at this time utilizes a self-position estimation function that determines a certain degree of reliability or degree of agreement.

[0004] Conventionally, a technique has been disclosed in which the position and attitude of a moving object are estimated based on information acquired by a plurality of sensors, and the estimated position and attitude are corrected when the reliability is low (see, for example, Patent Document 1). Also disclosed is a technique in which the position of a moving object is estimated using information acquired by a first sensor and a first reliability is acquired regarding the estimation result, the position of a moving object is estimated using information acquired by a second sensor (information different from the information acquired by the first sensor) and a second reliability is acquired regarding the estimation result, and the self-position of the moving object is acquired using the estimation result with the higher reliability of the first reliability or the second reliability (see, for example, Patent Document 2). [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Patent Publication No. 2022-42630 [Patent Document 2] Patent Publication No. 2021-18638 Summary of the Invention [Problem to be solved by the invention]

[0006] In an autonomous mobile body, when people, obstacles, or temporary installations are present in the moving space, their information is also used for matching during self-location estimation, or a situation may arise in which information that can be acquired by a sensor cannot be acquired due to an obstacle or the like. In such a situation, the environmental map information does not match the information acquired by the sensor, making it impossible to estimate the self-location. In addition, the environmental map includes places that are difficult for the sensor to detect accurately due to transparent or reflective objects such as glass.

[0007] In order to address these factors that make self-location estimation difficult, the challenge is to stably estimate the self-location without increasing the amount and types of detection by sensors as much as possible.

[0008] The present disclosure has been made to solve such problems, and has an object to provide a self-location estimation device that can stably estimate its own location. [Means for solving the problem]

[0009] In order to solve the above problems, the self-location estimation device according to the present disclosure includes a self-location estimation unit that estimates the self-location of the autonomously moving body based on two-dimensional map information including multiple two-dimensional maps in the height direction or travel direction within the movement range of the autonomously moving body and sensor information that detects the periphery of the moving body, and a determination unit that determines one of the multiple two-dimensional maps used by the self-location estimation unit for estimation and the detection area of ​​the sensor information based on predetermined conditions. Effect of the Invention

[0010] According to the present disclosure, it is possible to stably estimate the self-location. [Brief description of the drawings]

[0011] [Figure 1] 1 is a block diagram showing an example of a configuration of a self-location estimation device according to a first embodiment. [Diagram 2] 1 is an external view of a moving body according to a first embodiment. [Diagram 3] 3A and 3B are diagrams for explaining the angle of view and the detection area in the height direction of a sensor provided in the moving object according to the first embodiment. [Figure 4] 3A and 3B are diagrams for explaining the angle of view and the detection area in the height direction of a sensor provided in the moving object according to the first embodiment. [Diagram 5] FIG. 2 is a diagram for explaining the relationship between the self-position (center) of a moving object according to the first embodiment and point cloud data acquired by a sensor. [Figure 6] FIG. 2 is a diagram for explaining the relationship between the self-position (left side) of a moving object according to the first embodiment and point cloud data acquired by a sensor. [Figure 7] FIG. 2 is a diagram for explaining the relationship between the self-position (right side) of a moving object according to the first embodiment and point cloud data acquired by a sensor. [Figure 8] 1 is a perspective view showing an example of a travel area of ​​a moving body according to a first embodiment. [Figure 9] FIG. 2 is a diagram showing a two-dimensional map at a height α in accordance with the first embodiment. [Figure 10] FIG. 2 is a diagram showing a two-dimensional map at a height β in accordance with the first embodiment. [Figure 11] FIG. 2 is a diagram showing a two-dimensional map at a height γ in accordance with the first embodiment. [Figure 12] 1 is a diagram showing a state in which a moving body according to the first embodiment travels on the left side of a travel area A. FIG. [Figure 13] 2 is a diagram for explaining point cloud data acquired by a sensor when a moving body according to the first embodiment travels on the left side of a travel area A. FIG. [Figure 14] 1 is a diagram in which point cloud data acquired by a sensor when a moving object travels on the left side of a travel area A is matched with a two-dimensional map at a height α in accordance with the first embodiment. [Figure 15] 1 is a diagram in which point cloud data acquired by a sensor when a moving object travels on the left side of a travel area A is matched with the two-dimensional map at height β in accordance with the first embodiment. [Figure 16] 1 is a diagram in which point cloud data acquired by a sensor when a moving object travels on the left side of a travel area A is matched with the two-dimensional map at height γ in accordance with the first embodiment. [Figure 17] 2 is a diagram for explaining point cloud data acquired by a sensor when a moving body according to the first embodiment travels on the right side of a travel area A. FIG. [Figure 18] 1 is a diagram in which point cloud data acquired by a sensor when a moving object travels on the right side of a travel area A is matched with the two-dimensional map at height α in accordance with the first embodiment. [Figure 19] 1 is a diagram in which point cloud data acquired by a sensor when a moving object travels on the right side of a travel area A is matched with the two-dimensional map at height β in accordance with the first embodiment. [Figure 20] 1 is a diagram in which point cloud data acquired by a sensor when a moving object travels on the right side of a travel area A is matched with the two-dimensional map at height γ in accordance with the first embodiment. [Figure 21] 4A to 4C are diagrams for explaining point cloud data acquired by a sensor when a movable cart according to the first embodiment is located next to a wall. [Figure 22] 4A to 4C are diagrams for explaining point cloud data acquired by a sensor when a movable cart is not present in accordance with the first embodiment. [Diagram 23] 4 is a diagram showing a state in which a moving body according to the first embodiment travels in a travel area B. FIG. [Figure 24] FIG. 2 is a diagram for explaining point cloud data acquired by a sensor when the moving body according to the first embodiment travels in a travel area B (without a movable cart). [Diagram 25]FIG. 13 is a diagram in which point cloud data acquired by a sensor when a moving object travels through a travel area B (without a movable cart) is matched with a two-dimensional map at a height α according to the first embodiment. [Figure 26] FIG. 13 is a diagram in which point cloud data acquired by a sensor when a moving object travels through a travel area B (without a movable cart) is matched with a two-dimensional map at a height β according to the first embodiment. [Figure 27] FIG. 13 is a diagram in which point cloud data acquired by a sensor when a moving object travels in a travel area B (without a movable cart) is matched with a two-dimensional map at a height γ in accordance with the first embodiment. [Figure 28] FIG. 2 is a diagram for explaining point cloud data acquired by a sensor when a moving body according to the first embodiment travels in a travel area B (where a movable cart is present). [Figure 29] FIG. 13 is a diagram in which point cloud data acquired by a sensor when a moving object travels through a travel area B (with a movable cart) is matched with a two-dimensional map at a height α according to the first embodiment. [Diagram 30] FIG. 13 is a diagram in which point cloud data acquired by a sensor when a moving object travels through a travel area B (with a movable cart) is matched with a two-dimensional map at a height β according to the first embodiment. [Diagram 31] FIG. 13 is a diagram in which point cloud data acquired by a sensor when a moving object travels through a travel area B (with a movable cart) is matched with a two-dimensional map at a height γ according to the first embodiment. [Diagram 32] FIG. 4 is a diagram showing an example of a selection map according to the first embodiment. [Diagram 33] FIG. 11 is a block diagram showing an example of the configuration of a self-location estimation device according to a second embodiment. [Diagram 34] FIG. 11 is a diagram showing an example of a variance value variation map according to the second embodiment. [Diagram 35] FIG. 11 is a diagram showing a state in which a moving object according to the second embodiment travels in a travel area C (without people). [Diagram 36]FIG. 13 is a diagram in which point cloud data acquired by a sensor when a moving object travels through a travel area C (without people) is matched with a two-dimensional map at a height α according to the second embodiment. [Figure 37] FIG. 13 is a diagram in which point cloud data acquired by a sensor when a moving object travels through a travel area C (without people) is matched with a two-dimensional map at a height β according to the second embodiment. [Figure 38] FIG. 13 is a diagram in which point cloud data acquired by a sensor when a moving object travels through a travel area C (without people) is matched with a two-dimensional map at a height γ according to the second embodiment. [Figure 39] FIG. 11 is a diagram showing a state in which a moving object according to the second embodiment travels in a travel area C (where people are present). [Diagram 40] FIG. 13 is a diagram in which point cloud data acquired by a sensor when a moving object travels through a travel area C (where people are present) is matched with a two-dimensional map at a height α according to the second embodiment. [Diagram 41] FIG. 13 is a diagram in which point cloud data acquired by a sensor when a moving object travels through a travel area C (where people are present) is matched with a two-dimensional map at a height β according to the second embodiment. [Diagram 42] FIG. 13 is a diagram in which point cloud data acquired by a sensor when a moving object travels through a travel area C (where people are present) is matched with a two-dimensional map at a height γ according to the second embodiment. [Diagram 43] FIG. 11 is a block diagram showing an example of the configuration of a self-location estimation device according to embodiment 3. [Diagram 44] FIG. 13 is a diagram showing a state in which a moving object according to the third embodiment travels in a travel area C (where people are present). [Diagram 45] FIG. 13 is a diagram in which point cloud data acquired by a sensor when a moving object travels through a travel area C (where people are present) is matched with a two-dimensional map at a height α according to the third embodiment. [Figure 46] FIG. 13 is a diagram in which point cloud data acquired by a sensor when a moving object travels through a travel area C (where people are present) is matched with a two-dimensional map at a height β according to the third embodiment. [Figure 47]FIG. 13 is a diagram in which point cloud data acquired by a sensor when a moving object travels through a travel area C (where people are present) is matched with a two-dimensional map at a height γ according to the second embodiment. [Figure 48] FIG. 13 is a diagram in which point cloud data (excluding unused point clouds) acquired by a sensor when a moving object travels through a travel area C (where people are present) is matched with a two-dimensional map at height α in accordance with embodiment 3. [Figure 49] FIG. 13 is a diagram in which point cloud data (excluding unused point clouds) acquired by a sensor when a moving object travels through a travel area C (where people are present) is matched with a two-dimensional map at height β in accordance with embodiment 3. [Figure 50] FIG. 13 is a diagram in which point cloud data (excluding unused point clouds) acquired by a sensor when a moving object travels through a travel area C (where people are present) is matched with a two-dimensional map at a height γ in accordance with the second embodiment. [Figure 51] FIG. 11 is a block diagram showing an example of the configuration of a self-location estimation device according to a fourth embodiment. [Figure 52] FIG. 13 is a diagram showing an example of a composite map according to the fourth embodiment. [Figure 53] 1 is a diagram illustrating an example of a hardware configuration of a self-location estimation device according to first to fourth embodiments. [Figure 54] 1 is a diagram illustrating an example of a hardware configuration of a self-location estimation device according to first to fourth embodiments. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0012] <Embodiment 1> FIG. 1 is a block diagram showing an example of the configuration of a self-location estimation device 1 according to the first embodiment.

[0013] The self-location estimation device 1 includes two-dimensional map information 2, a map information switching unit 3, a sensor information switching unit 4, a self-location estimation unit 5, and a determination unit 6. The self-location estimation device 1, a sensor 7, and a movement control unit 8 are provided in a moving body 9 (see FIG. 2). The moving body 9 corresponds to the above-mentioned autonomous moving body.

[0014] The two-dimensional map information 2 is information extracted from the three-dimensional map information of the movement range of the moving body 9, and includes multiple two-dimensional maps in multiple planes (horizontal planes) with different height directions of the moving body 9, or multiple two-dimensional maps in multiple planes (vertical planes) with different directions of travel of the moving body 9.

[0015] The map information switching unit 3 switches to one of the multiple 2D maps included in the 2D map information 2 according to the determination result of the determination unit 6. The map information switching unit 3 outputs the switched 2D map to the self-position estimation unit 5.

[0016] The sensor information switching unit 4 switches the detection area of ​​the sensor information acquired by the sensor 7. The sensor information switching unit 4 outputs the sensor information in the switched detection area to the self-position estimation unit 5. The detection area of ​​the sensor information will be described later in detail.

[0017] The self-location estimation unit 5 estimates the self-location of the moving object 9 based on the two-dimensional map input from the map information switching unit 3 and the sensor information in the detection area input from the sensor information switching unit 4. Specifically, the self-location estimation unit 5 estimates the self-location of the moving object 9 by matching the point cloud data included in the sensor information with the two-dimensional map.

[0018] The determination unit 6 (decision unit) determines a combination of a two-dimensional map and a detection area used when the self-location estimation unit 5 estimates its own location, based on the self-location of the moving object 9 estimated by the self-location estimation unit 5. The determination unit 6 outputs the determination result to each of the map information switching unit 3 and the sensor information switching unit 4.

[0019] The movement control unit 8 controls the movement of the moving object 9 based on the self-position estimated by the self-position estimation unit 5.

[0020] 2 is an external view of the moving body 9. The right side in the figure is the front side of the moving body 9.

[0021] A motor is attached to each of the left and right wheels of the moving body 9. The moving body 9 moves forward and backward by the driving force of each motor, and turns right and left by utilizing the difference in the rotation speed of the left and right wheels.

[0022] In addition, 3D LIDAR (Light Detection And Ranging) is mounted on the front and rear of the moving object 9 as a sensor 7 for detecting the surroundings. The 3D LIDAR is a device that irradiates laser light and measures the distance to an object and the shape of the object based on the information of the reflected light.

[0023] FIG. 3 is a front view of the moving body 9. As shown in FIG. 3, the sensor 7 can measure a specific angle of view range and distance range. The angle of view range in the height direction is defined by the angle between the upper angle of view and the lower angle of view. The distance range is the range from the moving body 9 to the detection limit. A height α distance boundary, a height β distance boundary, and a height γ distance boundary are set at three positions in the distance range. The height β distance boundary means that on the moving body 9 side from the position where the height β distance boundary is set, anything above height β is outside the angle of view range and cannot be measured by the sensor 7. The same is true for the height α distance boundary and the height γ distance boundary. In this way, the angle of view range in the height direction becomes narrower as it approaches the moving body 9.

[0024] 4 is a top view of the moving object 9. As shown in FIG. 4, the detection limit, the height α distance boundary, the height β distance boundary, and the height γ distance boundary each extend in an arc shape. Outside the detection limit (a predetermined distance or more from the sensor 7), the sensor 7 cannot measure.

[0025] In the first embodiment, the sensors 7 are mounted on the front and rear of the moving object 9, but there are blind spots, such as directly to the side of the moving object 9, because the moving object 9 casts its own shadow. As a countermeasure to this, by increasing the number of sensors 7 mounted, it is possible to eliminate the blind spots and expand the detection range (angle of view range and detection range).

[0026] 5 to 7 are diagrams showing the relationship between the self-position of the moving object 9 in the passage and the point cloud data acquired by the sensor 7. Fig. 5 shows a case where the self-position of the moving object 9 is in the center, Fig. 6 shows a case where the self-position of the moving object 9 is on the left side, and Fig. 7 shows a case where the self-position of the moving object 9 is on the right side.

[0027] 3D LIDAR cannot measure transparent objects such as glass, and therefore cannot obtain point cloud data related to transparent objects. Also, point cloud data for reflective objects may exist in the wrong position. Therefore, when the mobile object 9 travels through a narrow space such as a passageway, the point cloud data of walls and the like that is matched with the 2D map will differ depending on the position of the mobile object 9. As shown in Figures 5 to 7, it can be seen that the point cloud data obtained by the sensor 7 differs depending on the position of the mobile object 9.

[0028] Fig. 8 is a perspective view showing an example of a passageway which is a travel area (movement range) of a moving object 9. In the passageway shown in Fig. 8, a two-dimensional map at height α is shown in Fig. 9, a two-dimensional map at height β is shown in Fig. 10, and a two-dimensional map at height γ is shown in Fig. 11. In the two-dimensional maps shown in Figs. 9 to 11, the window glass areas are treated as if there were no walls.

[0029] Fig. 12 is a diagram showing a state in which the moving object 9 moves on the left side of the moving area A shown in Fig. 8. As shown in Fig. 12, the moving object 9 has a window glass on its left side.

[0030] Fig. 13 is a diagram showing point cloud data acquired by the sensor 7 when the moving object 9 travels on the left side of the travel area A. Fig. 14 is a diagram showing a two-dimensional map at height α matched with point cloud data acquired by the sensor 7 when the moving object 9 travels on the left side of the travel area A. Fig. 15 is a diagram showing a two-dimensional map at height β matched with point cloud data acquired by the sensor 7 when the moving object 9 travels on the left side of the travel area A. Fig. 16 is a diagram showing a two-dimensional map at height γ matched with point cloud data acquired by the sensor 7 when the moving object 9 travels on the left side of the travel area A.

[0031] FIG. 17 is a diagram showing point cloud data acquired by the sensor 7 when the moving object 9 travels on the right side of the travel area A. FIG. 18 is a diagram showing a two-dimensional map at height α matched with point cloud data acquired by the sensor 7 when the moving object 9 travels on the right side of the travel area A. FIG. 19 is a diagram showing a two-dimensional map at height β matched with point cloud data acquired by the sensor 7 when the moving object 9 travels on the right side of the travel area A. FIG. 20 is a diagram showing a two-dimensional map at height γ matched with point cloud data acquired by the sensor 7 when the moving object 9 travels on the right side of the travel area A.

[0032] In the travel area A, when the moving object 9 travels on the left side (Fig. 13) and on the right side (Fig. 17), the point cloud data at height γ matches the wall included in the 2D map (Figs. 16 and 20), but at height β, the height matches the height of the window glass, so the sensor 7 acquires less point cloud data (Figs. 15 and 19). Also, at height α, the sensor 7 cannot acquire point cloud data in the vicinity beside the moving object 9 because the passage is a narrow space, and the amount of point cloud data available for matching is small (Figs. 14 and 18). Therefore, in the travel area A, it is understood that self-location estimation can be performed by matching the 2D map at height γ with the point cloud data at height γ (detection area).

[0033] 21 is a diagram showing point cloud data acquired by the sensor 7 when the movable cart 11 is located next to a wall. The movable cart 11 refers to a cart that does not move autonomously, such as a dolly for transporting luggage.

[0034] When the movable cart 11 is near a wall as shown in Fig. 21, the moving body 9 recognizes the movable cart 11 as an obstacle and travels along a route that avoids the movable cart 11. At this time, the moving body 9 travels along a position shifted toward the center compared to the route when the movable cart 11 is not present (see Fig. 22). In addition, the self-position estimation unit 5 estimates the self-position of the moving body 9 by using the point cloud data (Fig. 21) corresponding to the movable cart 11.

[0035] FIG. 23 is a diagram showing a state in which the moving object 9 travels through the travel area B shown in FIG.

[0036] FIG. 24 is a diagram showing point cloud data acquired by the sensor 7 when the moving object 9 travels in the travel area B. In FIG. 24, there is no movable cart 11 in the travel area B. FIG. 25 is a diagram showing point cloud data acquired by the sensor 7 when the moving object 9 travels in the travel area B shown in FIG. 24 matched with a two-dimensional map at height α. FIG. 26 is a diagram showing point cloud data acquired by the sensor 7 when the moving object 9 travels in the travel area B shown in FIG. 24 matched with a two-dimensional map at height β. FIG. 27 is a diagram showing point cloud data acquired by the sensor 7 when the moving object 9 travels in the travel area B shown in FIG. 24 matched with a two-dimensional map at height γ.

[0037] FIG. 28 is a diagram showing point cloud data acquired by the sensor 7 when the moving object 9 travels in the travel area B. In FIG. 28, a movable cart 11 exists in the travel area B. FIG. 29 is a diagram showing a two-dimensional map at height α matched with point cloud data acquired by the sensor 7 when the moving object 9 travels in the travel area B shown in FIG. 28. FIG. 30 is a diagram showing a two-dimensional map at height β matched with point cloud data acquired by the sensor 7 when the moving object 9 travels in the travel area B shown in FIG. 28. FIG. 31 is a diagram showing a two-dimensional map at height γ matched with point cloud data acquired by the sensor 7 when the moving object 9 travels in the travel area B shown in FIG. 28.

[0038] In the traveling area B, when there is no obstacle such as the movable cart 11, the point cloud data at height γ matches the two-dimensional map (FIG. 27). On the other hand, in the traveling area B, when the movable cart 11 is present, the point cloud data at height γ includes point cloud data corresponding to the movable cart 11, making it difficult to perform matching for clearly estimating the self-position. In addition, as shown in FIG. 28, a window glass is provided on the side of the moving body 9 opposite to the side where the movable cart 11 is present, so that matching between the two-dimensional map and the point cloud data at height γ is almost impossible.

[0039] In the travel area B, since point cloud data can be obtained stably for heights α and β, it is desirable to perform matching using the 2D map and point cloud data at height β, where there is more point cloud data. However, if the moving object 9 travels along the left side of the wall, matching at height β becomes impossible, so it is better to move the path of the moving object 9 from the center to the right.

[0040] Since the building shapes and environmental conditions vary depending on the location where the moving object 9 is traveling, a selection map (FIG. 32) indicating which map should be selected depending on the location may be stored in advance. In this case, the determination unit 6 determines a combination of the two-dimensional map and the detection range of the sensor information used when the self-location estimation unit 5 estimates its own location, based on the self-location of the moving object 9 estimated by the self-location estimation unit 5 and the selection map.

[0041] Furthermore, although the above describes the case where a horizontal two-dimensional map is used, if the ceiling has features, a vertical two-dimensional map may be used.

[0042] As described above, according to the first embodiment, it is possible to stably estimate the self-location.

[0043] <Embodiment 2> 33 is a block diagram showing an example of the configuration of the self-location estimation device 12 according to embodiment 2. The self-location estimation device 12 is characterized by including an extraction unit 13, a variance value variation map 14, and a determination unit 15. Other configurations are similar to those of the self-location estimation device 1 according to embodiment 1 (FIG. 1), and therefore detailed description thereof will be omitted here.

[0044] Similar to the determination unit 6 of the self-location estimation device 1 according to the first embodiment, the extraction unit 13 determines a combination of the two-dimensional map and the detection area of ​​the sensor information based on the self-location estimated by the self-location estimation unit 5.

[0045] For example, when the self-position estimation unit 5 estimates the self-position by a self-position estimation algorithm using a particle filter, variance value information for scattering particles is acquired in advance every time the mobile object 9 travels, and a variance value variation map 14 (FIG. 34) of the information is stored for each height. In this way, the variance value variation map 14 includes variance values ​​based on the probability distribution for the self-position of the mobile object 9 in the two-dimensional map for each height.

[0046] In the two-dimensional map according to the self-position estimated by the self-position estimation unit 5, if there is no person or obstacle, the particles will scatter as usual, so the variance value is small and does not change significantly.

[0047] Fig. 35 is a diagram showing a state in which the moving object 9 travels in the travel area C shown in Fig. 8. No people are present in the travel area C.

[0048] Fig. 36 is a diagram in which a two-dimensional map at height α is matched with point cloud data acquired by the sensor 7 when the moving object 9 travels through the travel area C shown in Fig. 35. Fig. 37 is a diagram in which a two-dimensional map at height β is matched with point cloud data acquired by the sensor 7 when the moving object 9 travels through the travel area C shown in Fig. 35. Fig. 38 is a diagram in which a two-dimensional map at height γ is matched with point cloud data acquired by the sensor 7 when the moving object 9 travels through the travel area C shown in Fig. 35.

[0049] Fig. 39 is a diagram showing a state in which the moving object 9 travels in the travel area C shown in Fig. 8. In the travel area C, a person is present.

[0050] Fig. 40 is a diagram in which a two-dimensional map at height α is matched with point cloud data acquired by the sensor 7 when the moving object 9 travels through the travel area C shown in Fig. 39. Fig. 41 is a diagram in which a two-dimensional map at height β is matched with point cloud data acquired by the sensor 7 when the moving object 9 travels through the travel area C shown in Fig. 39. Fig. 42 is a diagram in which a two-dimensional map at height γ is matched with point cloud data acquired by the sensor 7 when the moving object 9 travels through the travel area C shown in Fig. 39.

[0051] When no people are present in the travel area C, it is easy to match the 2D map and the point cloud data at height γ (Fig. 38). On the other hand, when people are present in front of and behind the moving object 9, the point cloud data is occluded by people at heights β and γ and is significantly different from normal (when no people are present), so the variance values ​​are also different (Figs. 41 and 42).

[0052] At height α, the variance is close to normal in both cases where no person is present (FIG. 36) and where a person is present (FIG. 40), because the point cloud data is not occluded by a person. Therefore, the determination unit 15 determines the combination of the 2D map at height α and the detection area of ​​the sensor information.

[0053] As described above, according to the second embodiment, the detection area of ​​the two-dimensional map and the sensor information is switched using not only the self-location but also the variance value, so that it is possible to respond to dynamically changing events.

[0054] Furthermore, since traveling in a dynamically changing environment is often dangerous and an error may occur in estimating the self-position, the moving body 9 may be stopped when the change in the variance value is equal to or greater than a predetermined range (when the difference in the variance values ​​is greater than a predetermined threshold). Alternatively, the moving body 9 may be temporarily stopped when the change in the variance value is equal to or greater than the predetermined range, and then restarted when the change in the variance value falls below the predetermined range.

[0055] <Embodiment 3> 43 is a block diagram showing an example of a configuration of a self-location estimation device 17 according to embodiment 3. The self-location estimation device 17 is characterized by including a person position determination unit 18, a person position point cloud removal unit 19, and a determination unit 20. Other configurations are similar to those of the self-location estimation device 1 according to embodiment 1 (FIG. 1), and therefore detailed description thereof will be omitted here.

[0056] The human position determination unit 18 acquires an image of the periphery of the moving object 9 captured by the camera 21, and performs image processing to determine the position of a person present in the periphery of the moving object 9. The camera 21 is mounted on the moving object 9. Note that the person is not limited to a person, and may be a dynamic object other than a person.

[0057] The human position point cloud removal unit 19 removes point cloud data corresponding to the positions of people determined by the human position determination unit 18.

[0058] The self-position estimation unit 5 estimates the self-position of the moving object 9 based on the point cloud data other than the point cloud removed by the human position point cloud removal unit 19 and the two-dimensional map.

[0059] Fig. 44 is a diagram showing a state in which the moving object 9 travels in the travel area C shown in Fig. 8. In the travel area C, a person is present.

[0060] Fig. 45 is a diagram in which a two-dimensional map at height α is matched with point cloud data acquired by the sensor 7 when the moving object 9 travels through the travel area C shown in Fig. 44. Fig. 46 is a diagram in which a two-dimensional map at height β is matched with point cloud data acquired by the sensor 7 when the moving object 9 travels through the travel area C shown in Fig. 44. Fig. 47 is a diagram in which a two-dimensional map at height γ is matched with point cloud data acquired by the sensor 7 when the moving object 9 travels through the travel area C shown in Fig. 44.

[0061] As shown in Fig. 45, at height α, the two-dimensional map and the point cloud data match. In this case, the human position point cloud removal unit 19 does not perform removal processing because there is no point cloud data corresponding to the human position (Fig. 48). The self-position estimation unit 5 estimates the self-position of the moving object 9 in the same manner as in the first embodiment.

[0062] On the other hand, as shown in Fig. 46 and Fig. 47, there are some places where the two-dimensional map and the point cloud data do not match at height β and height γ. In this case, the human position point cloud removal unit 19 judges the point cloud data in the direction where the person exists as noise, and removes the point cloud data judged as noise (point cloud data within the non-used point cloud range) from the used point cloud (Fig. 49 and Fig. 50). The self-position estimation unit 5 estimates the self-position of the moving object 9 by matching the used point cloud with the two-dimensional map.

[0063] As described above, according to the third embodiment, it is possible to estimate the self-position of the moving object 9 excluding dynamic objects such as people.

[0064] Since the range in which a dynamic object such as a person exists can be predicted to some extent, the range photographed by the camera 21 may be set to a predetermined range from the moving object 9. This makes it possible to reduce the amount of information acquired from the camera 21.

[0065] <Fourth embodiment> 51 is a block diagram showing an example of a configuration of the self-location estimation device 22 according to embodiment 4. The self-location estimation device 22 is characterized by including two-dimensional map information 23 and a determination unit 24. Other configurations are similar to those of the self-location estimation device 1 according to embodiment 1 (FIG. 1), and therefore detailed description thereof will be omitted here.

[0066] The two-dimensional map information 23 is a composite map that combines two-dimensional maps at each height shown in Figures 9 to 11, and includes an optimum map according to the position of the moving object 9. Figure 52 is a diagram showing an example of the composite map.

[0067] The determination unit 24 estimates the direction of travel of the moving body 9 based on the self-position of the moving body 9 estimated by the self-position estimation unit 5 and odometry information 25 obtained from the rotation of the wheels of the moving body 9, and determines the detection area of ​​the sensor information based on the estimated direction of travel.

[0068] As described above, according to the fourth embodiment, it is possible to stably estimate the self-location without switching between two-dimensional maps as in the first to third embodiments.

[0069] <Hardware configuration> The functions of the map information switching unit 3, the sensor information switching unit 4, the self-location estimation unit 5, and the determination unit 6 in the self-location estimation device 1 described in the first embodiment are realized by a processing circuit. That is, the self-location estimation device 1 includes a processing circuit for switching two-dimensional maps, switching detection areas of sensor information, estimating the self-location of the moving object 9, and determining the detection areas of the two-dimensional maps and the sensor information based on the estimated self-location. The processing circuit may be dedicated hardware, or may be a processor (also called a CPU, central processing unit, processing unit, arithmetic unit, microprocessor, microcomputer, or DSP (Digital Signal Processor)) that executes a program stored in a memory.

[0070] When the processing circuit is a dedicated hardware, the processing circuit 30 corresponds to, for example, a single circuit, a composite circuit, a programmed processor, a parallel programmed processor, an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or a combination of these, as shown in Fig. 53. Each function of the map information switching unit 3, the sensor information switching unit 4, the self-position estimation unit 5, and the determination unit 6 may be realized by the processing circuit 30, or each function may be realized collectively by one processing circuit 30.

[0071] When the processing circuit 30 is the processor 40 shown in FIG. 54, the functions of the map information switching unit 3, the sensor information switching unit 4, the self-location estimation unit 5, and the determination unit 6 are realized by software, firmware, or a combination of software and firmware. The software or firmware is written as a program and stored in the memory 41. The processor 40 realizes each function by reading and executing the program recorded in the memory 41. That is, the self-location estimation device 1 includes a memory 41 for storing a program that results in the execution of a step of switching a two-dimensional map, a step of switching a detection area of ​​the sensor information, a step of estimating the self-location of the moving object 9, and a step of determining the detection area of ​​the two-dimensional map and the sensor information based on the estimated self-location. In addition, it can be said that these programs cause a computer to execute the procedures or methods of the map information switching unit 3, the sensor information switching unit 4, the self-location estimation unit 5, and the determination unit 6. Here, the memory may be, for example, a non-volatile or volatile semiconductor memory such as RAM (Random Access Memory), ROM (Read Only Memory), flash memory, EPROM (Erasable Programmable Read Only Memory), EEPROM (Electrically Erasable Programmable Read Only Memory), a magnetic disk, a flexible disk, an optical disk, a compact disk, a DVD (Digital Versatile Disc), or any storage medium that will be used in the future.

[0072] It should be noted that some of the functions of the map information switching unit 3, the sensor information switching unit 4, the self-position estimation unit 5, and the judgment unit 6 may be realized by dedicated hardware, and other functions may be realized by software or firmware.

[0073] Thus, the processing circuitry may implement the functions described above through hardware, software, firmware, or a combination of these.

[0074] The above describes the hardware configuration of the self-location estimation device 1 according to embodiment 1, but the same applies to the hardware configurations of the self-location estimation device 12 according to embodiment 2 (FIG. 33), the self-location estimation device 17 according to embodiment 3 (FIG. 43), and the self-location estimation device 22 according to embodiment 4 (FIG. 51).

[0075] Within the scope of the present disclosure, the embodiments can be freely combined, and each embodiment can be modified or omitted as appropriate.

[0076] <Additional Notes> Various aspects of the present disclosure are summarized below as appendices.

[0077] (Appendix 1) A self-location estimation unit that estimates a self-location of a moving body based on two-dimensional map information including a plurality of two-dimensional maps in a height direction or a traveling direction in a moving range of the autonomous moving body and sensor information that detects the periphery of the moving body; a determination unit that determines one of the plurality of two-dimensional maps and a detection area of ​​the sensor information to be used when the self-location estimation unit estimates the self-location based on a predetermined condition; A self-location estimation device comprising: (Appendix 2) The self-location estimation device according to claim 1, wherein the determination unit holds a plurality of combinations of one of the plurality of two-dimensional maps and the detection area of ​​the sensor information, and determines one combination from the plurality of combinations based on the predetermined condition. (Appendix 3) A variance value based on a probability distribution for the self-position of the moving object on each of the plurality of two-dimensional maps is stored in advance; The self-location estimation device according to claim 1 or 2, wherein the determination unit determines one of the plurality of two-dimensional maps and a detection area of ​​the sensor information based on a difference between a variance value based on the sensor information acquired when the moving body moves autonomously and the variance value stored in advance. (Appendix 4) The self-location estimation device according to claim 3, wherein if the difference between the variance values ​​in all of the plurality of two-dimensional maps is greater than a predetermined threshold, the autonomous movement of the moving body is temporarily stopped, and then the moving body is restarted after the difference becomes equal to or less than the threshold. (Appendix 5) 5. The self-location estimation device according to claim 1, further comprising a selection map indicating a selection of the two-dimensional map according to each position within a movement range of the moving body. (Appendix 6) The self-location estimation device according to any one of claims 1 to 5, wherein the determination unit determines a detection area of ​​the sensor information based on rotation information of wheels of the moving object. (Appendix 7) The self-location estimation device according to any one of claims 1 to 6, wherein the two-dimensional map information includes a composite map that combines two or more of the two-dimensional maps. (Appendix 8) The self-location estimation device according to any one of claims 1 to 7, wherein the self-location estimation unit estimates the location of the moving body based on the sensor information detected within a predetermined distance from the moving body. (Appendix 9) The self-location estimation device according to claim 2, wherein the multiple combinations are in the forward and backward directions of the moving body. (Appendix 10) 10. The self-location estimation device according to any one of claims 1 to 9, wherein the two-dimensional map in the height direction is a two-dimensional map in the horizontal direction, and the two-dimensional map in the traveling direction is a two-dimensional map in the vertical direction. (Appendix 11) The self-location estimation device according to any one of claims 1 to 10, wherein the detection area is an area excluding an area in which a person or an obstacle is present. [Explanation of symbols]

[0078] 1 self-location estimation device, 2 two-dimensional map information, 3 map information switching unit, 4 sensor information switching unit, 5 self-location estimation unit, 6 determination unit, 7 sensor, 8 movement control unit, 9 moving body, 10 person, 11 movable cart, 12 self-location estimation device, 13 extraction unit, 14 variance value variation map, 15 determination unit, 16 fixed base, 17 self-location estimation device, 18 person position determination unit, 19 person position point cloud removal unit, 20 determination unit, 21 camera, 22 self-location estimation device, 23 two-dimensional map information, 24 determination unit, 25 odometry information, 30 processing circuit, 40 processor, 41 memory.

Claims

1. a self-location estimation unit that estimates a self-location of the autonomously moving body based on two-dimensional map information including a plurality of two-dimensional maps in a height direction or a travel direction in a movement range of the autonomously moving body and sensor information that detects the periphery of the autonomously moving body; a determination unit that determines one of the plurality of two-dimensional maps and a detection area of ​​the sensor information to be used when the self-location estimation unit performs estimation based on a predetermined condition; A self-location estimation device comprising:

2. The self-location estimation device according to claim 1 , wherein the determination unit holds a plurality of combinations of one of the plurality of two-dimensional maps and the detection area of ​​the sensor information, and determines one combination from the plurality of combinations based on the predetermined condition.

3. A variance value based on a probability distribution for the self-position of the moving object on each of the plurality of two-dimensional maps is stored in advance; 2. The self-location estimation device according to claim 1, wherein the determination unit determines one of the plurality of two-dimensional maps and a detection area of ​​the sensor information based on a difference between a variance value based on the sensor information acquired when the moving body moves autonomously and the variance value stored in advance.

4. 4. The self-location estimation device according to claim 3, wherein if a difference between each of the variance values ​​in all of the plurality of two-dimensional maps is greater than a predetermined threshold, the autonomous movement of the moving body is temporarily stopped, and then the moving body is restarted after the difference becomes equal to or less than the threshold.

5. The self-location estimation device according to claim 1 , further comprising a selection map indicating a selection of the two-dimensional map according to each position within a movement range of the mobile object.

6. The self-location estimation device according to claim 1 , wherein the determination unit determines the detection area of ​​the sensor information based on rotation information of wheels of the moving object.

7. The self-location estimation device according to claim 1 , wherein the two-dimensional map information includes a composite map that combines two or more of the two-dimensional maps.

8. The self-location estimation device according to claim 1 , wherein the self-location estimation unit estimates the location of the moving object based on the sensor information detected within a predetermined distance from the moving object.

9. The self-location estimation device according to claim 2 , wherein the plurality of combinations are in forward and backward directions of the moving body.

10. The self-location estimation device according to claim 1 , wherein the two-dimensional map in the height direction is a two-dimensional map in the horizontal direction, and the two-dimensional map in the traveling direction is a two-dimensional map in the vertical direction.

11. The self-location estimation device according to claim 1 , wherein the detection area is an area excluding an area in which a person or an obstacle is present.

Images