JP7641551B1 - Map creation system and automated driving system - Google Patents

Abstract

[Problem] To provide a map creation system that can easily create wide-area maps of various heights. [Solution] The map creation system 10 includes a first three-dimensional point cloud data acquisition unit 11 that measures the surroundings and acquires three-dimensional point cloud data D1, and a wide-area map creation unit 12 that projects three-dimensional point cloud data D1a within a predetermined range Hr in the height direction out of the three-dimensional point cloud data D1 acquired by the first three-dimensional point cloud data acquisition unit 11 onto a horizontal plane to create a two-dimensional wide-area map M1. [Selected Figure] Figure 1

Description

Applicable under Article 30, Paragraph 2 of the Patent Act Exhibition name: CEATEC 2023 Venue: Makuhari Messe (2-1 Nakase, Mihama-ku, Chiba City, Chiba Prefecture) Opening date: October 17th to October 20th, 2023

The present invention relates to a map creation system and an automated driving system.

Patent document 1 discloses an invention for a mobile object control device that enables a mobile object with wheels to travel autonomously.

In the mobile body control device described in Patent Document 1, before the mobile body starts autonomous driving, an environmental map (wide area map) is created based on SLAM (Simultaneous Localization and Mapping) technology by driving the mobile body within a driving area. Then, the mobile body drives autonomously based on the environmental map and the surrounding situation (two-dimensional point cloud data) recognized by a two-dimensional LiDAR (Light Detection And Ranging) mounted on the mobile body.

JP 2022-13243 A

Two-dimensional LiDAR recognizes the surrounding situation in two dimensions (plane). The invention described in Patent Document 1 compares an environmental map with the surrounding situation recognized by the two-dimensional LiDAR, so it is necessary to prepare an environmental map in advance that matches the measurement position of the two-dimensional LiDAR.

However, in the invention described in Patent Document 1, there is a demand for multiple moving bodies to travel in the same space. In this case, if the installation position (height) of the two-dimensional LiDAR mounted on the moving bodies differs depending on the moving body, an environmental map (wide area map) that matches the installation position (height) of the two-dimensional LiDAR mounted on each moving body is required. In this case, it is necessary to create an environmental map (wide area map) according to each installation position (height) before each moving body can travel autonomously, which requires excessive effort and time.

The present invention aims to provide a map creation system that can easily create wide-area maps at various heights.

The present invention comprises a first three-dimensional point cloud data acquisition unit that measures the surroundings and acquires three-dimensional point cloud data, and a wide-area map creation unit that projects, in the height direction , three-dimensional point cloud data within a predetermined range in the height direction from the three-dimensional point cloud data acquired by the first three-dimensional point cloud data acquisition unit onto any horizontal plane , to create a two-dimensional wide-area map.

The present invention makes it easy to create wide-area maps at various heights.

FIG. 1 is a block diagram of a map creation system and an automatic driving system according to an embodiment of the present invention. FIG. 2 is a side view of a cart in the map creation system according to the embodiment of the present invention. FIG. 3 is a top view of a cart in a map creation system according to an embodiment of the present invention. FIG. 4 is a diagram showing a specific section R of a construction site. FIG. 5 is a side view of an automatic vehicle according to an embodiment of the present invention. FIG. 6 is a flowchart showing a flow for creating a wide-area map according to an embodiment of the present invention. FIG. 7(A) is a diagram showing a portion of the data when three-dimensional point cloud data is extracted on one plane, and FIG. 7(B) is a diagram showing a portion of the data when three-dimensional point cloud data is extracted within a specified range and compressed. FIG. 8 is a flowchart showing a flow when an automatic vehicle according to an embodiment of the present invention is caused to automatically travel. FIG. 9 is a side view of an automated vehicle according to a modified example, and a diagram for explaining an extraction range of measured three-dimensional point cloud data. FIG. 10 is a flowchart showing a flow for creating a wide-area map according to a modified example. FIG. 11 is a flowchart showing a flow when an automatic vehicle according to the modified example is caused to automatically travel.

The following describes a map creation system 10 and an automated driving system 100 according to an embodiment of the present invention, with reference to the drawings.

The autonomous driving system 100 of this embodiment is used to control the autonomous driving of an autonomous vehicle 1 used, for example, at construction sites, logistics sites, production plants, restaurants, or medical facilities. The autonomous driving system 100 includes a map creation system 10 and an autonomous vehicle 1. The autonomous driving system 100 automatically drives the autonomous vehicle 1 using a wide-area map M1 created by the map creation system 10 and a surrounding map M2 created by a controller 20 mounted on the autonomous vehicle 1. In the following embodiment, the map creation system 10 and the autonomous driving system 100 are used at a construction site.

First, the map creation system 10 will be described with reference to Figures 1 to 4. Figure 1 is a block diagram of the automated driving system 100.

The map creation system 10 of this embodiment creates a wide-area map M1 required for the automated vehicle 1 to automatically navigate to a destination at a construction site.

As shown in FIG. 1, the map creation system 10 includes a first three-dimensional point cloud data acquisition unit 11 and a computer C. The computer C includes a wide-area map creation unit 12, a memory unit 13, and a tilt correction unit 14. The first three-dimensional point cloud data acquisition unit 11 and the computer C are mounted on a cart 15 (mobile body) (see FIG. 2 and FIG. 3). Note that the wide-area map creation unit 12 and the tilt correction unit 14 are virtual units that represent the functions of the computer C, and do not represent physical entities.

In this embodiment, the wide-area map creation unit 12, the memory unit 13, and the tilt correction unit 14 are provided inside one computer C mounted on the dolly 15, but they do not have to be provided inside the same computer. In other words, the computer C is not limited to being composed of one computer, but may be composed of multiple computers. In addition, some or all of the wide-area map creation unit 12, the memory unit 13, and the tilt correction unit 14 may be provided in a computer installed in a location different from the dolly 15.

The first three-dimensional point cloud data acquisition unit 11 is composed of a three-dimensional LiDAR and is mounted on a cart 15 (see FIG. 2). In the following description, the three-dimensional LiDAR will be given the same reference number "11" as the first three-dimensional point cloud data acquisition unit 11.

As shown in Figures 2 and 3, the three-dimensional LiDAR 11 measures the surroundings of the cart 15 to obtain three-dimensional point cloud data. Specifically, the three-dimensional LiDAR 11 measures the distance and height to the object (such as a wall or pillar) by irradiating the object with a laser light while scanning and measuring the reflected light. The data obtained by the three-dimensional LiDAR 11 is a collection of three-dimensional points (hereinafter, the collection of data obtained by the three-dimensional LiDAR 11 is referred to as "three-dimensional point cloud data D1"). In this embodiment, a three-dimensional LiDAR 11 with a measurement distance of several tens of meters to about 200 m is used. In addition, as shown in Figures 2 and 3, the three-dimensional LiDAR 11 performs measurements within a range of a predetermined angle in the vertical and horizontal directions with the three-dimensional LiDAR 11 as the center.

When creating a wide-area map M1 using the map creation system 10, for example, a worker moves thoroughly around an area R in a construction site where an automated vehicle 1 is to automatically travel, pushing a cart 15 as shown in FIG. 4. During this time, the 3D LiDAR 11 constantly measures the area around the cart 15 and acquires 3D point cloud data D1.

The wide area map creation unit 12 creates a wide area map M1 based on the three-dimensional point cloud data D1 measured by the three-dimensional LiDAR 11. In this embodiment, the wide area map M1 is created based on SLAM (Simultaneous Localization and Mapping) technology. Note that the wide area map M1 in this embodiment refers to a two-dimensional map of the entire section R including pillars, walls, obstacles, etc. A specific method for creating the wide area map M1 will be described in detail later.

The storage unit 13 stores the three-dimensional point cloud data D1 measured by the three-dimensional LiDAR 11, the wide-area map M1 created by the wide-area map creation unit 12, and the like. The storage unit 13 is composed of various storage devices such as a hard disk mounted on the computer C, or storage media such as a memory card or USB memory connectable to the computer C. The storage unit 13 may be mounted on the cart 15, or may be provided in a computer or the like installed in a location separate from the cart 15.

The tilt correction unit 14 extracts point cloud data of horizontal references (e.g., floors and ceilings) from the three-dimensional point cloud data D1 measured by the three-dimensional LiDAR 11, calculates the tilt of the extracted point cloud data, and corrects the tilt of the three-dimensional point cloud data D1 to be horizontal. Alternatively, a tilt angle sensor (not shown) may be provided on the cart 15, and the tilt angle and tilt direction of the cart 15 may be detected by the tilt angle sensor, and the tilt of the measured three-dimensional point cloud data D1 may be corrected based on the detected tilt angle and tilt direction.

Next, we will explain the autonomous vehicle 1.

The autonomous vehicle 1 of this embodiment is, for example, an autonomous transport cart. As shown in FIG. 5, the autonomous vehicle 1 includes a vehicle body 2, four wheels 3 arranged on the front, rear, left and right sides of the vehicle body 2, a battery 4 that supplies power to a motor (not shown) that drives the wheels 3, a two-dimensional point cloud data acquisition unit 5, and a controller 20 that controls the travel of the autonomous vehicle 1. A loading platform 6 on which luggage is placed is installed on the vehicle body 2. Note that the autonomous vehicle 1 does not necessarily have to be an autonomous vehicle with a transport function.

As shown in FIG. 1, the controller 20 has a surrounding area map creation unit 21, a memory unit 22, a control unit 23, and a transmission/reception unit 24. Note that the surrounding area map creation unit 21 and the control unit 23 are virtual units that represent the functions of the controller 20, and do not represent physical entities.

In this embodiment, the two-dimensional point cloud data acquisition unit 5 is configured with a two-dimensional LiDAR and is mounted on the autonomous vehicle 1. In the following, the two-dimensional LiDAR mounted on the autonomous vehicle 1 will be described with the same reference number "5" as the two-dimensional point cloud data acquisition unit 5. The two-dimensional LiDAR 5 measures the surroundings of the autonomous vehicle 1 and acquires two-dimensional point cloud data D2.

The surrounding area map creation unit 21 creates a map of the surroundings of the autonomous vehicle 1 (hereinafter referred to as the "surrounding area map M2") based on the two-dimensional point cloud data D2 acquired by the two-dimensional LiDAR 5. The specific method of creating the surrounding area map M2 will be described in detail later.

The storage unit 22 is composed of a storage medium such as a readable/writable hard disk or SSD. The storage unit 22 stores the two-dimensional point cloud data D2 measured by the two-dimensional LiDAR 5 and the wide-area map M1 created by the map creation system 10, as well as various programs for controlling the driving operation of the autonomous vehicle 1. Furthermore, the storage unit 22 stores various data such as the destination of the autonomous vehicle 1 and the route to the destination.

The memory unit 22 may be configured as a storage medium (e.g., a memory card) that is detachable from the controller 20. Alternatively, the memory unit 22 may be configured as a storage medium installed in a location different from the controller 20. Furthermore, these may be used in combination.

The control unit 23 controls the traveling operation of the automated vehicle 1 based on the wide-area map M1 stored in the memory unit 22 and the surrounding area map M2 created by the surrounding area map creation unit 21. Specifically, when a destination is input to the automated vehicle 1, the control unit 23 calculates a route to the destination while taking into consideration obstacle avoidance. The control unit 23 then estimates the current position of the automated vehicle 1 based on the surrounding area map M2, which is created as needed while the automated vehicle 1 is traveling, and the wide-area map M1 stored in the memory unit 22, and drives the wheels 3 with a motor (not shown) to automatically drive the automated vehicle 1 to the destination.

The transmission/reception unit 24 communicates data between the controller 20 and an external computer. Through the transmission/reception unit 24, the controller 20 can transmit, for example, information about the current position of the automated driving vehicle 1 and the driving route to the external computer, and can receive information such as a wide-area map M1 and destination information from the external computer.

In this way, the autonomous vehicle 1 can automatically navigate to a destination based on the wide-area map M1 created by the map creation system 10 and the surrounding area map M2 created based on the two-dimensional point cloud data D2 acquired by the two-dimensional LiDAR 5 of the autonomous vehicle 1.

The two-dimensional point cloud data D2 acquired by the two-dimensional LiDAR 5 of the autonomous vehicle 1 is two-dimensional (on a horizontal plane) point cloud data. In contrast, the three-dimensional point cloud data D1 measured by the three-dimensional LiDAR 11 in the map creation system 10 is three-dimensional (solid) point cloud data. However, the three-dimensional point cloud data D1 is a collection of sparse point data. For this reason, for example, if only plane data at the same height as the two-dimensional point cloud data D2 is extracted from the three-dimensional point cloud data D1 in order to compare it with the two-dimensional point cloud data D2, the map will be composed of sparse points, as shown in FIG. 7(A), in other words, it will be a map with a small amount of information (low accuracy). If the autonomous vehicle 1 is driven automatically using such a map, for example, there is a risk that the autonomous vehicle 1 will not be able to estimate its own position.

Furthermore, when using the automated driving system 100, multiple automated vehicles 1 may be driven within the area R. In this case, if the installation position (height H in FIG. 5) of the two-dimensional LiDAR 5 mounted on each automated vehicle 1 differs depending on the automated vehicle 1, a wide-area map M1 that matches the installation position (height H) of the two-dimensional LiDAR mounted on each automated vehicle 1 is required. For this reason, before each automated vehicle 1 is driven automatically, it becomes necessary to create a wide-area map M1 that matches the installation position (height H) of each two-dimensional LiDAR 5, which requires more effort and time.

Therefore, the map creation system 10 of this embodiment employs a method for creating a wide-area map M1 that is highly accurate and can easily correspond to the installation position (height) of the two-dimensional LiDAR 5, using three-dimensional point cloud data D1 measured by the three-dimensional LiDAR 11. Below, a specific method for creating the wide-area map M1 will be described with reference to the flowchart shown in FIG. 6.

In step S11, three-dimensional point cloud data D1 is acquired. Specifically, while the cart 15 is moving, the three-dimensional LiDAR 11 measures the surroundings of the cart 15 to acquire the three-dimensional point cloud data D1.

In step S12, tilt correction is performed. Specifically, point cloud data of horizontal reference points (e.g., floor surface or ceiling) is extracted from the three-dimensional point cloud data D1 measured by the three-dimensional LiDAR 11, the tilt of the extracted point cloud data is calculated, and the tilt of the three-dimensional point cloud data D1 is corrected to be horizontal. Note that if tilt correction is not required, the processing of step S13 may be omitted.

In step S13, the height (height H) of the installation position of the two-dimensional LiDAR 5 mounted on the autonomous vehicle 1 is input. Specifically, the height H from the floor surface of the two-dimensional point cloud data D2 measured by the two-dimensional LiDAR 5 mounted on the autonomous vehicle 1 is input to computer C.

In step S14, three-dimensional point cloud data D1a within a predetermined range Hr is extracted. Specifically, computer C (wide area map creation unit 12) extracts three-dimensional point cloud data D1a within a predetermined range Hr in the height direction from the three-dimensional point cloud data D1 acquired by the three-dimensional LiDAR 11. In this embodiment, the range Hr is set to a range of, for example, about ±20 cm, centered on the height H of the plane (the installation position of the two-dimensional LiDAR 5) measured by the two-dimensional LiDAR 5 mounted on the autonomous vehicle 1 (see FIG. 2).

In step S15, the three-dimensional point cloud data D1a is compressed. Specifically, computer C (wide area map creation unit 12) projects the three-dimensional point cloud data D1a within range Hr onto a horizontal plane to create a wide area map M1; in other words, the point cloud data within range Hr is overlaid on a single plane to create a two-dimensional wide area map M1. This reduces the gaps between data (see FIG. 7B) compared to when only data from one plane of the three-dimensional point cloud data D1 is extracted (see FIG. 7A). As a result, the data density increases, making it possible to create a wide area map M1 with high accuracy.

In addition, when autonomous vehicles 1 with different installation heights H of the two-dimensional LiDAR 5 are operated autonomously, a wide-area map M1 corresponding to the autonomous vehicle 1 can be easily created simply by changing the input of the installation height H of the two-dimensional LiDAR 5 in step S12 according to the installation height of the two-dimensional LiDAR 5 of each autonomous vehicle 1.

Next, the driving control of the autonomous vehicle 1 in the autonomous driving system 100 will be described with reference to the flowchart shown in FIG. 8. The process shown in the flowchart in FIG. 8 is executed based on a program pre-stored in the memory unit 22 of the controller 20. The controller 20 repeats the process shown in FIG. 8 several times per second.

The process shown in FIG. 8 is performed when the route to the destination of the autonomous vehicle 1 is set in the controller 20 and the autonomous vehicle 1 is traveling autonomously.

In step S21, two-dimensional point cloud data D2 is acquired. Specifically, two-dimensional point cloud data D2 of the surroundings of the autonomous vehicle 1 is acquired by a two-dimensional LiDAR 5 mounted on the autonomous vehicle 1.

In step S22, the self-position is estimated. Specifically, the controller 20 (periphery map creation unit 21) creates a peripheral map M2 of the surroundings of the autonomous vehicle 1 from the two-dimensional point cloud data D2 acquired by the two-dimensional LiDAR 5. Next, the controller 20 (control unit 23) compares the wide-area map M1 previously stored in the memory unit 22 with the created peripheral map M2 to estimate the current position (self-position) of the autonomous vehicle 1.

In step S23, the presence or absence of an obstacle is determined. Specifically, the controller 20 (control unit 23) determines whether or not an obstacle has appeared on the route to the set destination based on the surrounding area map M2 created in step S22 or the two-dimensional point cloud data D2. For example, at a construction site, materials or equipment that did not exist when the wide-area map M1 was created may be placed on the route set when the automated vehicle 1 is traveling. The controller 20 (control unit 23) detects obstacles that have appeared in this way when the automated vehicle 1 is traveling. If an obstacle is not detected on the route to the set destination, the process proceeds to step S24. If an obstacle is detected on the route to the set destination, the process proceeds to step S25. Note that the detection of an obstacle may be performed by comparing the wide-area map M1 with the surrounding area map M2.

In step S24, it is determined whether the destination has been reached. Specifically, the controller 20 (control unit 23) determines whether the current position of the automated vehicle 1 estimated in step S22 is the destination. If the automated vehicle 1 has reached the destination, the control (automatic driving) is terminated. In contrast, if the automated vehicle 1 has not reached the destination, the process returns to step S21.

Next, step S25 will be described. As described above, if it is determined in step S23 that an obstacle exists on the route to the set destination, the process proceeds to step S25. In step S25, the route is reset. Specifically, if the controller 20 (control unit 23) detects an obstacle on the set route, the controller 20 searches for a route to reach the destination while avoiding the obstacle of the automated driving body 1, and resets the route. Then, the process returns to step S21, and the process from step S21 onwards is executed again.

As described above, in the map creation system 10 of this embodiment, the three-dimensional point cloud data D1a within a predetermined height range Hr is projected onto a horizontal plane to create a wide-area map M1, so that a highly accurate wide-area map M1 can be created. And in the automated driving system 100 of this embodiment, the highly accurate wide-area map M1 is used when the automated driving vehicle 1 is driven automatically, so that the estimation accuracy of the self-position of the automated driving vehicle 1 can be improved, and the accuracy of the automated driving can also be improved.

Furthermore, with the map creation system 10, it is possible to easily create a wide-area map M1 that corresponds to the height H of the installation position of the two-dimensional LiDAR 5 of the autonomous vehicle 1, simply by changing the input of the height H of the installation position of the two-dimensional LiDAR 5. In other words, by using the map creation system 10, wide-area maps M1 of various heights can be easily created, so that the effort and time required to recreate the wide-area map M1 can be significantly reduced when using an autonomous vehicle 1 with a different installation position of the two-dimensional LiDAR 5.

Next, a modified example of the automated driving system 100 will be described. In the automated driving system 100, the automated vehicle 1 is equipped with a two-dimensional LiDAR 5, but the automated driving system 100 according to the modified example differs in that the automated vehicle 1 is equipped with a three-dimensional LiDAR 50 (second three-dimensional point cloud data acquisition unit) (see FIG. 9). In addition, in the automated driving system 100, the wide-area map M1 is created according to the height H of the installation position of the two-dimensional LiDAR 5, but the automated driving system 100 according to the modified example differs in that the wide-area map M11 can be created at any height. Note that, in the following, the same configurations and the same processes are given the same numbers and descriptions are omitted as appropriate.

First, the method for creating the wide-area map M11 in the modified example will be described with reference to the flowchart shown in FIG. 10. Note that only steps S113, S114, and S115, which differ from the flowchart shown in FIG. 6, will be described here.

In this modified example, the three-dimensional point cloud data D1 measured by the three-dimensional LiDAR 11 mounted on the cart 15 is also used. In step S113, the height H1 (see FIG. 9) is input. Specifically, the height H1 from the floor surface is input to the computer C. The height H1 is an arbitrary height and does not need to correspond to the installation position of the three-dimensional LiDAR 50 mounted on the autonomous vehicle 1. In this modified example, the height H1 is set to about 2 m, for example (see FIG. 9).

Then, in step S114, three-dimensional point cloud data D1b of a predetermined range Hr is extracted. Specifically, computer C (wide area map creation unit 12) extracts three-dimensional point cloud data D1b within a predetermined range Hr1 in the height direction from the three-dimensional point cloud data D1 acquired by the three-dimensional LiDAR 11. In this modified example, the predetermined range Hr1 from which the three-dimensional point cloud data D1 is extracted is, for example, a range of about height H1 ± 50 cm (see Figure 9).

At construction sites, materials and equipment are sometimes temporarily placed on the floor (see Figure 9). If a wide-area map M11 is created in this state, the materials and equipment will naturally be reflected in the wide-area map M11. However, when the autonomous vehicle 1 is driven automatically, the materials and equipment that were placed on the floor may have moved. In this case, there may be a discrepancy between the wide-area map M11 created in advance and the surrounding area map M2 acquired by the autonomous vehicle 1 while it is driving automatically, and the controller 20 (control unit 23) may not be able to estimate the current position of the autonomous vehicle 1.

Therefore, by setting the height H1 of the extraction range of the three-dimensional point cloud data D1 to about 2 m (range Hr1 is about 1.5 m to 2.5 m in height) as in this modified example, the wide-area map M11 can be created using point cloud data in a range that does not include materials or equipment temporarily placed on the floor. In other words, the wide-area map M11 can be created without being affected by materials or equipment.

In step S115, the three-dimensional point cloud data D1b is compressed. The specific method is the same as in step S15, so the explanation is omitted.

Next, the driving control of the automated driving vehicle 1 in this modified example will be described with reference to the flowchart shown in FIG. 11. The process shown in the flowchart in FIG. 11 is executed based on a program pre-stored in the memory unit 22 of the controller 20. The controller 20 repeats the process shown in FIG. 11 several times per second.

In this modified example, the surrounding map M12 is created based on the three-dimensional point cloud data D3 measured by the three-dimensional LiDAR 50 mounted on the autonomous vehicle 1, and is created in the same manner as the wide-area map M11 (flowchart in FIG. 10). Although not shown, to briefly explain, the predetermined range Hr1 in the height direction of the three-dimensional point cloud data D3 acquired by the three-dimensional LiDAR 50 is set to be the same as the extraction range of the three-dimensional point cloud data D1 (height H1 ± 50 cm). When creating the surrounding map M12, once the height H1 is input in step S112 in FIG. 10, there is no need to input the height H1 (step S112) thereafter.

In this way, by creating the wide-area map M11 and the surrounding area map M12 under the same conditions, the accuracy of estimating the self-position can be improved. Therefore, optimally, it is desirable that the 3D LiDAR 50 mounted on the autonomous vehicle 1 is the same model as the 3D LiDAR 11 mounted on the bogie 15.

Also, in the automated driving system 100 of this modified example, the extraction range of the data of the three-dimensional point cloud data D1, D3, i.e., the height of the wide-area map M11 and surrounding area map M12 to be created, can be changed as appropriate. For example, by changing the extraction range (heights H, H1) of the data of the three-dimensional point cloud data D1, D3 according to the height of materials and equipment placed on the floor surface and the progress of construction, it is possible to create the wide-area map M11 and surrounding area map M12 of a range that is not affected by these factors as appropriate. This makes it possible to improve the estimation accuracy of the self-position of the automated driving body 1, and also improve the accuracy of automated driving.

The above map creation system 10 and automated driving system 100 provide the following advantages:

In the map creation system 10, from the three-dimensional point cloud data D1 acquired by the three-dimensional LiDAR 11, three-dimensional point cloud data D1a within a predetermined range Hr in the height direction is extracted, and this three-dimensional point cloud data D1a is compressed to create a two-dimensional wide-area map M1. This makes it possible to create a highly accurate two-dimensional wide-area map M1 even when using the three-dimensional point cloud data D1.

In addition, with the map creation system 10, two-dimensional wide-area maps M1 of different heights can be easily created simply by changing the range Hr (input of height H) as appropriate. In other words, by using the map creation system 10, two-dimensional wide-area maps M1 of various heights can be easily created. For example, when using multiple automated vehicles 1 with different installation positions of the two-dimensional LiDAR 5, the effort and time required to create a wide-area map M1 corresponding to each automated vehicle 1 can be significantly reduced.

Furthermore, in the autonomous driving system 100 of this embodiment, when the autonomous vehicle 1 is driven autonomously, the high-precision wide-area map M1 created by the map creation system 10 is used, thereby improving the estimation accuracy of the autonomous vehicle 1's own position and improving the accuracy of autonomous driving.

In addition, when a 3D LiDAR 50 is mounted on the autonomous vehicle 1, the data extraction range (heights H, H1) of the 3D point cloud data D1, D3 can be changed according to the height of materials and equipment placed on the floor and the progress of the construction work, making it possible to appropriately create a wide-area map M11 and a surrounding map M12 of a range that is not affected by these. This can further improve the estimation accuracy of the autonomous vehicle 1's self-position, and also further improve the accuracy of autonomous driving.

Although the embodiments of the present invention have been described above, the above embodiments merely show some of the application examples of the present invention, and are not intended to limit the technical scope of the present invention to the specific configurations of the above embodiments.

In the above embodiment, the method of acquiring three-dimensional point cloud data has been described using three-dimensional LiDAR 11, 50, but the method of acquiring three-dimensional point cloud data is not limited to this. For example, three-dimensional point cloud data may be acquired from VisualSLAM technology using an imaging device such as a camera, or three-dimensional point cloud data may be acquired by moving a two-dimensional LiDAR three-dimensionally.

In addition, in the above embodiment, the surrounding map M2 is created from the two-dimensional point cloud data D2 and processing is then carried out, but the two-dimensional point cloud data D2 may be used as is.

In the above embodiment, an example was described in which three-dimensional point cloud data D1 is acquired by a three-dimensional LiDAR 11 mounted on a cart 15 pushed by a worker, but this is not limited to the above. For example, the three-dimensional LiDAR 11 may be mounted on a remotely controllable cart, and the cart may be remotely operated to acquire the three-dimensional point cloud data D1. In addition, a worker may move around holding the three-dimensional LiDAR 11 in his/her hand, or may acquire the three-dimensional point cloud data D1 by mounting the three-dimensional LiDAR 11 on a helmet or wearing it on the arm and walking within the section R.

In addition, in the above embodiment, an example was described in which the three-dimensional point cloud data D1 is acquired while the three-dimensional LiDAR11 is moved, but the three-dimensional point cloud data D1 does not necessarily have to be acquired while the three-dimensional LiDAR11 is moved. For example, three-dimensional point cloud data D1 acquired at multiple points may be overlaid, and a wide-area map M1 may be created based on this overlaid three-dimensional point cloud data D1. Also, three-dimensional point cloud data may be acquired using images captured at fixed points by an imaging device such as a camera.

100 Autonomous driving system 10 Map creation system 1 Autonomous driving body 5 Two-dimensional LiDAR (two-dimensional point cloud data acquisition unit)
11 3D LiDAR (first 3D point cloud data acquisition unit)
12 Wide area map creation unit 13 Memory unit 14 Inclination correction unit 15 Cart (moving body)
20 Controller 21 Surrounding area map creation unit 22 Memory unit 23 Control unit 50 Three-dimensional LiDAR (second three-dimensional point cloud data acquisition unit)
C Computer

Claims (3)

a first three-dimensional point cloud data acquisition unit that measures the surroundings and acquires three-dimensional point cloud data;
a wide-area map creation unit that projects three-dimensional point cloud data within a predetermined range in the height direction onto any horizontal plane , from the three-dimensional point cloud data acquired by the first three-dimensional point cloud data acquisition unit, in the height direction to create a two-dimensional wide-area map. An autonomous vehicle that automatically drives to a destination;
A storage unit that stores the wide-area map obtained by the map creation system according to claim 1;
a two-dimensional point cloud data acquisition unit mounted on the automated driving body and configured to measure the surroundings of the automated driving body and acquire two-dimensional point cloud data on a predetermined horizontal plane;
a control unit that controls a traveling operation of the automated driving vehicle based on the wide-area map stored in the storage unit and the two-dimensional point cloud data acquired by the two-dimensional point cloud data acquisition unit,
The wide-area map creation unit is an automated driving system that creates the wide-area map from three-dimensional point cloud data within a predetermined range including the horizontal plane measured by the two-dimensional point cloud data acquisition unit. An autonomous vehicle that automatically travels to a destination;
A storage unit that stores the wide-area map obtained by the map creation system according to claim 1;
A second three-dimensional point cloud data acquisition unit that is mounted on the automated driving body and measures the surroundings of the automated driving body to acquire three-dimensional point cloud data;
a surroundings map creation unit that creates a two-dimensional surroundings map of the surroundings of the autonomous vehicle by projecting, in a height direction , three-dimensional point cloud data within a predetermined range in the height direction, among the three-dimensional point cloud data acquired by the second three-dimensional point cloud data acquisition unit, onto any one plane in the horizontal direction ;
An autonomous driving system comprising: a control unit that controls the driving operation of the autonomous driving body based on the wide-area map stored in the memory unit and the surrounding area map created by the surrounding area map creation unit.

Images