JP2022130228A - Point group processing device and robot system - Google Patents

Abstract

To effectively reduce a point group obtained from a detector.SOLUTION: A point group processing device 160 includes: a point group acquisition unit 161 configured to acquire a point group, which is a plurality of point data, from a detector 110 that acquires, for a plurality of different detection points, point data indicating relative positions of detection points on the surface of an object existing in a room; a wall surface point group extraction unit 162 configured to extract a wall surface point group, which is a point group belonging to a wall surface of the room, by rectangular fitting applied to the acquired point group; and a partition point group extraction unit 163 configured to extract a parallel point group separated from the extracted wall surface point group by a predetermined distance or more and parallel to the wall surface point group, and extract a partition point group for partitioning the room from the wall surface point group and the parallel point group.SELECTED DRAWING: Figure 4

Description

The present invention relates to a point cloud processing device for processing a point cloud obtained from a detector that detects relative positions of detection points on the surface of an object, and a robot system provided with the point cloud processing device.

LiDAR (Laser Imaging Detection and Ranging) sensors, etc., are placed on the walls of a room or within a room using detectors that detect the relative positions of detection points corresponding to multiple locations on the surface of an object using light. Techniques exist for identifying objects that are For example, in Patent Document 1, a background point group, which is a point group belonging to a background area such as a floor surface or a wall surface, is extracted from a point group including a plurality of point data obtained from detectors arranged at the same point at a plurality of timings. is extracted, and an area where an object exists is extracted from the difference between the point cloud and the background point cloud.

JP 2019-219248 A

However, when objects such as furniture, home appliances, luggage, and foliage plants are arranged along the wall surface, point data cannot be obtained for the shape of the back side of the object where the light from the detector does not reach. Therefore, complicated processing is required, such as a step of moving the detector to detect the shape of the back side of the object from the side of the object.

SUMMARY OF THE INVENTION The present invention has been made in view of the above problems, and provides a point cloud processing device for quickly extracting objects placed in a room from a point cloud obtained from a detector, and a robot system equipped with the point cloud processing device. With the goal.

To achieve the above object, a point cloud processing apparatus according to one aspect of the present invention acquires point data indicating relative positions of detection points on a surface of an object existing in a room for a plurality of different detection points. a point cloud acquisition unit that acquires a point cloud that is a plurality of the point data from the detector, and a wall surface that extracts a wall surface point cloud that is a point cloud belonging to the wall surface of the room by rectangular fitting to the acquired point cloud. A point cloud extraction unit extracts a parallel point group separated from the extracted wall surface point group by a predetermined distance or more and parallel to the wall surface point group, and partitioning the room from the wall surface point group and the parallel point group. and a partitioning point group extracting unit for extracting the group.

In order to achieve the above object, another robot system of the present invention includes a robot that runs autonomously, and a robot system that is attached to the robot and detects relative positions of detection points on the surface of an object existing in a room. A detector that acquires point data indicating a plurality of different detection points, and a point group processing device that processes a point group obtained from the detector, wherein the point group processing device includes a plurality of A point cloud acquisition unit that acquires the point cloud that is the point data, and a wall surface point cloud extraction unit that extracts the wall surface point cloud that is the point cloud belonging to the wall surface of the room by rectangular fitting to the acquired point cloud, A partition point for extracting a parallel point group separated from the extracted wall surface point group by a predetermined distance or more and parallel to the wall surface point group, and extracting a partition point group for partitioning the room from the wall surface point group and the parallel point group. and a group extraction unit.

According to the present invention, it is possible to reduce the amount of information of the point cloud obtained from the detector and contribute to speeding up the process of extracting objects arranged in the room.

1 is a plan view showing an appearance of a robot system according to an embodiment; FIG. It is a bottom view showing the appearance of the robot system according to the embodiment. 1 is a perspective view showing an appearance of a robot system according to an embodiment; FIG. It is a block diagram which shows the functional structure of the point-group processing apparatus which concerns on embodiment. It is the figure which mapped the point cloud acquired from the detector which concerns on embodiment. It is a figure which shows the state which fitted the rectangle to the point cloud. FIG. 10 is a diagram showing an object point group remaining after removing a wall point group and a partition point group from the point group; FIG. 10 is a diagram showing a moving point cloud that is a difference between object point clouds at different times; FIG. 4 is a diagram showing motion vectors derived from moving point groups at different times;

BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, embodiments of a point group processing apparatus and a robot system according to the present invention will be described with reference to the drawings. It should be noted that the following embodiments are examples for explaining the present invention, and are not intended to limit the present invention. For example, the shapes, structures, materials, constituent elements, relative positional relationships, connection states, numerical values, formulas, contents of each step in the method, order of each step, etc. shown in the following embodiments are examples, and are described below. may include content not listed in In addition, geometric expressions such as parallel and orthogonal are sometimes used, but these expressions do not indicate mathematical rigor, and substantially allowable errors, deviations, and the like are included. Expressions such as "simultaneous" and "identical" also include substantially permissible ranges.

In addition, the drawings are schematic diagrams that have been appropriately emphasized, omitted, or adjusted in proportion in order to explain the present invention, and are different from the actual shape, positional relationship, and proportion.

Further, in some cases, a plurality of inventions will be collectively described as one embodiment below. Also, some of the contents described below are described as optional components related to the present invention.

FIG. 1 is a plan view showing the appearance of the robot system according to the embodiment. FIG. 2 is a bottom view showing the appearance of the robot system according to the embodiment. FIG. 3 is a perspective view showing the appearance of the robot system according to the embodiment. As shown in these figures, the robot system 100 is an autonomously traveling robot cleaner that autonomously travels in a room and cleans dust, dirt, etc. existing on the floor surface. 110 and a point cloud processing device 160 .

The robot 120 is equipped with various components and is a device that performs cleaning while autonomously running. cleaning unit 140 (see FIG. 2), suction unit 150 for sucking dust into robot 120, controller 170 for controlling drive unit 130, cleaning unit 140, and suction unit 150, and detector 110. It has different sensors.

A bumper that can be displaced with respect to the robot 120 is attached to the outer periphery of the robot 120 . Further, as shown in FIG. 2, the robot 120 is provided with a suction port 121 for sucking dust into the robot 120 .

The drive unit 130 is a device that causes the robot system 100 to travel based on instructions from the control device 170 . In the case of the present embodiment, one drive unit 130 is arranged on each of the left side and the right side of the center of the width direction of the robot 120 in plan view. The number of drive units 130 is not limited to two, and may be one or three or more.

The drive unit 130 has a wheel that travels on the floor surface, a travel motor that applies torque to the wheel, and a housing that houses the travel motor. Each wheel is housed in a recess formed in the bottom surface of the robot 120 and attached to the robot 120 so as to be rotatable.

The drive system of the robot system 100 of the present embodiment is of a two-wheel opposed type having casters 179 as auxiliary wheels. etc., the robot 120 can be made to run freely.

The cleaning unit 140 is a unit for sucking dust from the suction port 121, and includes a main brush arranged in the suction port 121, a side brush for collecting dust in the suction port 121, a brush driving motor for rotating each brush, and the like. It has

The suction unit 150 is arranged inside the robot 120 and has a fan case and an electric fan arranged inside the fan case. The electric fan sucks the air inside the dustbin unit 151 and discharges the air to the outside of the robot 120 , thereby sucking the dust from the suction port 121 and storing the dust in the dustbin unit 151 .

Examples of sensors other than the detector 110 provided in the robot 120 include the following sensors. The robot 120 may be provided with sensors other than these sensors, or may not be provided with at least one of these sensors.

The obstacle sensor 173 is a sensor that detects the presence or absence of an object existing in front of the robot 120 . In the case of this embodiment, an ultrasonic sensor is used as the obstacle sensor 173 . The obstacle sensor 173 has a transmitter 171 arranged in the front center of the robot 120, and receivers 172 arranged on both sides of the transmitter 171. The sensor 173 is transmitted from the transmitter 171 and reflected by obstacles. The distance and position of the obstacle can be detected by the receiving unit 172 receiving each of the ultrasonic waves that have returned.

The proximity sensor 174 detects the presence or absence of an object different from the object detected by the obstacle sensor 173, an object existing at a different distance from the object detected by the obstacle sensor 173, and the like. In the case of this embodiment, the proximity sensor 174 is an infrared sensor having a light-emitting portion and a light-receiving portion, and detects the presence or absence of an object based on whether the infrared rays reflected by the obstacle return.

A collision sensor (not shown) is a contact displacement sensor that is turned on when a bumper attached to the periphery of the robot 120 contacts an obstacle and is pushed into the robot 120 .

The floor surface sensors 176 are arranged at a plurality of locations on the bottom surface of the robot 120 and detect whether or not there is a floor surface. In the case of this embodiment, the floor surface sensor 176 is an infrared sensor having a light-emitting portion and a light-receiving portion. If not, it is detected as no floor.

The encoder is provided in the drive unit 130 and detects the respective rotation angles of the pair of wheels rotated by the driving motor. Information from the encoder can be used to calculate the travel distance, turning angle, speed, acceleration, angular velocity, and the like of the robot system 100 .

The acceleration sensor detects acceleration when robot system 100 runs, and the angular velocity sensor detects angular velocity when robot system 100 turns. Information detected by the acceleration sensor and the angular velocity sensor is used as information for correcting an error caused by wheel idling.

The detector 110 is attached to the robot 120 and detects a plurality of different point data indicating relative positions of detection points on surfaces of objects such as walls and objects existing in a room to be cleaned by the robot 120 . It is a so-called 3D sensor that acquires points. The type of the detector 110 is not particularly limited, and the laser light is rotated 360 degrees around the robot 120, and the reflected light is acquired at each predetermined angle to detect the object from the polar coordinates and the distance. A LiDAR sensor that detects the position of each detection point on the surface, and a ToF (Time of Flight) that resolves and acquires the distance to the object in pixel units by receiving light that is emitted and reflected by an image sensor with an image sensor. A camera etc. can be illustrated.

In this embodiment, the detector 110 is provided on the upper surface of the robot 120 so as to protrude upward, parallel to the floor on which the robot 120 runs, and at a predetermined height (for example, 20 cm) from the floor. A LiDAR sensor that acquires point data indicating the position of a detection point on the surface of an object in a set within ). Specifically, the detector 110 rotates laser light around the detector 110 in a horizontal plane at a predetermined height position, receives reflected light at each predetermined angle (for example, 5°), and measures the distance to the detection point. Point data indicating the position of the detection point is detected by two-dimensional polar coordinates based on the angle at the time of light reception.

FIG. 4 is a block diagram showing the functional configuration of the point cloud processing device. The point cloud processing device 160 is a device for processing a point cloud, which is point data indicating the positions of a plurality of detection points from the detector 110, and is a processing unit implemented by causing a processor to execute a program. 161 , a wall point group extraction unit 162 , and a partition point group extraction unit 163 . In the case of this embodiment, the point cloud processing device 160 further includes an object point cloud extraction unit 164 , a moving point cloud extraction unit 165 and a vector derivation unit 166 .

The point cloud acquisition unit 161 acquires a point cloud 200, which is point data indicating the positions of a plurality of detection points, from the detector 110. FIG. In the case of this embodiment, the detector 110 acquires, as a point group 200, a plurality of two-dimensional point data indicating positions of detection points within a plane at a predetermined height position. FIG. 5 is a diagram mapping the point cloud obtained from the detector.

Note that the robot 120 may move within the room, and the detector 110 may detect the position of the detection point at different positions. The point cloud acquisition unit 161 may update the point cloud 200 acquired using the feature position (for example, charging position of the robot 120) in the room as the reference position to absolute coordinates.

Further, when the point cloud acquisition unit 161 acquires point data that spreads three-dimensionally from the detector 110 as the point cloud 200 , the point cloud acquisition unit 161 extracts point data corresponding to a predetermined height position from the point cloud 200 to obtain the point cloud 200 . can be updated.

The wall surface point cloud extraction unit 162 extracts a wall surface point cloud, which is a point cloud belonging to the wall surface of the room, by rectangular fitting to the point cloud 200 acquired by the point cloud acquisition unit 161 from the detector 110 . Specifically, as shown in FIG.
Boundary box approximation is performed to determine a rectangle 210 (bounding box) that fits the outermost circumference of , and point data 201 existing on the rectangle 210 is extracted as a wall surface point group.

The partition point cloud extraction unit 163 extracts the wall surface point cloud extracted by the wall surface point cloud extraction unit 162, the wall surface point cloud spaced apart from one side of the rectangle 210 by a predetermined distance or more, or the parallel point cloud 202 parallel to one side of the rectangle 210. Then, based on the wall surface point group or the point group parallel to one side of the rectangle 210, the partition point group corresponding to the perimeter of the partition that partitions the room is extracted. When extracting the parallel point group 202, a threshold may be set for the number of points forming the point group. By limiting the point group to a parallel point group including a certain number of points, it is possible to prevent a point group that is not parallel to the wall surface point group from being erroneously recognized as a partition. Here, a "partition" is an object that partitions a room space, such as a large piece of furniture (without legs) placed in a room or a built-in piece of furniture such as a kitchen counter.

The object point cloud extraction unit 164 excludes the wall surface point cloud extracted by the wall surface point cloud extraction unit 162 and the partition point cloud extracted by the partition point cloud extraction unit 163 from the point cloud 200, and the points surrounded by the remaining residual point cloud. An object point group 203 indicating objects existing in the room is extracted from the area having a predetermined area or more among the areas where the room is located. FIG. 7 is a diagram showing an object point group remaining after removing the wall point group and the partition point group from the point group. In the case of this embodiment, the object point cloud extracting unit 164 associates the object point cloud 203 with the time when the detector 110 detects the point data (the time when the point cloud acquiring unit 161 acquires the point cloud, etc.). and store it in the storage device 180.

In the case of this embodiment, the predetermined area, which is the threshold value for extracting the object point cloud, can extract point data corresponding to at least the area of the human foot (one leg) at a predetermined height position on the horizontal plane as the object point cloud. set to a value. Therefore, the leg of a table, which has a larger area than a human leg, is extracted as an object point group.

The moving point cloud extraction unit 165 extracts the object point cloud 203 extracted by the wall point cloud extraction unit 162, the partition point cloud extraction unit 163, and the object point cloud extraction unit 164 based on the point cloud 200 acquired at the first time. The object point cloud 203 at the first time stored in the storage device 180 is acquired from the storage device 180, and the object points at the second time extracted based on the point cloud acquired at the second time after the first time A difference between the cloud 203 and the object point cloud 203 at the first time is derived, and the remaining object point cloud 203 is extracted as a moving point cloud 204 representing a moving object as shown in FIG. In the case of this embodiment, the moving point cloud extracting unit 165 extracts the extracted moving point cloud 204 at the time when the detector 110 detects the point data (the time when the point cloud acquiring unit 161 acquires the point cloud may be used). is stored in the storage device 180 in association with.

The vector deriving unit 166 extracts the moving point cloud 204 previously extracted by the moving point cloud extracting unit 165 and stored in the storage device 180 in association with the time information, and based on the moving point cloud 204 extracted at a later time. 9, the motion vector 220 of the moving object is derived. In the case of this embodiment, the vector derivation unit 166 successively derives the motion vector 220 based on the moving point group 204 stored in the past and the time information associated therewith each time the moving point group 204 is extracted. ing.

The control device 170 is a device that acquires the moving point cloud extracted by the moving point cloud extraction unit 165 and controls the drive unit 130 so that the robot 120 does not approach the moving object corresponding to the moving point cloud. In the case of this embodiment, the control device 170 acquires the motion vector 220 from the vector derivation unit 166, estimates the direction in which the moving object moves, and drives the robot 120 so that it does not approach the vicinity of the direction in which the moving object moves. It controls unit 130 .

The control device 170 controls movement and cleaning of the robot 120 by causing the processor to execute a program. Note that the processor that executes the program related to the control device 170 and the processor that executes the program related to the point cloud processing device 160 may be the same.

Control device 170 is connected to various sensors such as obstacle sensor 173 , proximity sensor 174 , collision sensor, floor sensor 176 , and detector 110 . Further, the control device 170 is connected to each drive source such as a running motor, a brush drive motor, and an electric fan, and controls rotation of the connected drive sources based on information from various sensors. Further, the control device 170 causes the robot 120 to travel according to the cleaning program stored in the storage device so as to clean the entire room as much as possible. The controller 170 creates a map using SLAM (Simultaneous Localization and Mapping) technology based on information from the detector 110 and the encoder, and controls traveling and cleaning while recognizing its own position.

According to the robot system 100 described above, the amount of point cloud information obtained from the detector 110 is reduced, and an object point cloud arranged in a room and a moving object such as a person moving in the room are quickly detected. be able to.

In particular, extracting the point cloud to be removed and reducing the point cloud in stages to reduce the amount of information in the point cloud makes it possible to effectively reduce the amount of information in the point cloud in a short period of time. .

According to the robot 120 equipped with the point cloud processing device 160 according to the embodiment, the robot 120 can move so as not to hinder the movement of the moving object and avoid the moving object.

Furthermore, by using the motion vector derived by the point cloud processing device 160, the motion of a moving object such as a person can be predicted, and the robot 120 can be controlled so as not to intersect with the motion of the moving object.

It should be noted that the present invention is not limited to the above embodiments. For example, another embodiment realized by arbitrarily combining the constituent elements described in this specification or omitting some of the constituent elements may be an embodiment of the present invention. The present invention also includes modifications obtained by making various modifications to the above-described embodiment within the scope of the gist of the present invention, that is, the meaning of the words described in the claims, which a person skilled in the art can think of. be

For example, although the robot system 100 whose planar shape is a Reuleaux triangle is illustrated, the shape of the robot system 100 is not particularly limited, and any shape such as a circle or a rectangle can be adopted.

Also, although the robot vacuum cleaner is exemplified, the robot 120 may be a guide robot, an autonomous mobile digital signage, an automatic guided vehicle, or the like.

Moreover, although the case where the detector 110, the point group processing device 160, etc. are mounted in the housing of one robot 120 is shown, part or all of the point group processing device 160 may be separate from the robot 120. The robot system 100 may be a robot system 100 that transmits and receives information through communication.

INDUSTRIAL APPLICABILITY The present invention can be used for autonomously moving robots such as robot cleaners.

100 Robot system 110 Detector 120 Robot 121 Suction port 130 Drive unit 140 Cleaning unit 150 Suction unit 151 Box unit 160 Point group processing device 161 Point group acquisition unit 162 Wall surface point group extraction unit 163 Partition point group extraction unit 164 Object point group extraction Unit 165 Moving point cloud extraction unit 166 Vector derivation unit 170 Control device 171 Transmission unit 172 Reception unit 173 Obstacle sensor 174 Proximity sensor 176 Floor sensor 179 Caster 180 Storage device 200 Point cloud 201 Point data 202 Parallel point cloud 203 Object point cloud 204 Moving point cloud 210 Rectangle 220 Vector

Claims (5)

a point cloud acquisition unit that acquires a plurality of point clouds, which are the point data, from a detector that acquires point data indicating relative positions of detection points on the surface of an object existing in the room for a plurality of different detection points; ,
a wall surface point cloud extracting unit for extracting a wall surface point cloud, which is a point cloud belonging to the wall surface of the room, by rectangular fitting to the obtained point cloud;
A partition point for extracting a parallel point group separated from the extracted wall surface point group by a predetermined distance or more and parallel to the wall surface point group, and extracting a partition point group for partitioning the room from the wall surface point group and the parallel point group. a group extractor;
A point cloud processing device. The wall surface point group and the partition point group are excluded from the point group, and the object point group indicating the objects existing in the room is extracted from the area having a predetermined area or more among the areas surrounded by the remaining point group. 2. The point group processing device according to claim 1, comprising an object point group extraction unit for extraction. Based on the point cloud acquired at the first time and the point cloud acquired at the second time after the first time, the wall point cloud extraction unit, the partition point cloud extraction unit, and the object point cloud extraction unit 3. The point cloud processing device according to claim 2, further comprising a moving point cloud extraction unit that derives the difference between the two object point clouds respectively extracted by and extracts the remaining object point cloud as a moving point cloud representing a moving object. . 4. The point group processing device according to claim 3, further comprising a vector derivation unit that derives a motion vector of a moving object based on the moving point groups at different times. an autonomous robot,
a detector that is attached to the robot and acquires point data indicating relative positions of detection points on the surface of an object existing in the room for a plurality of different detection points;
A point cloud processing device according to any one of claims 1 to 4,
A robot system with

Images