JP6758068B2 - Autonomous mobile robot - Google Patents

Description

The present invention relates to an autonomous mobile robot that autonomously moves in a moving space.

Conventionally, an autonomous flight robot capable of calculating a movement path from a self-position which is a current position to a predetermined movement target position and autonomously flying along the movement path has been proposed. For example, in Patent Document 1, a movement target position is set at a position separated by a predetermined distance from a moving object such as a thief, a movement path from a self-position to the movement target position is generated, and the movement path is followed. An autonomous flight robot that is controlled to fly is disclosed.

In the above-mentioned prior art, a plurality of candidate positions for moving target positions (moving candidate positions) are set at surrounding positions separated by a predetermined distance from the moving object so that the moving object can be photographed from various angles as much as possible. It is devised so that the position where you can shoot in the direction where you cannot shoot is easily set as the movement target position.

Japanese Unexamined Patent Publication No. 2014-119901

By the way, when a plurality of moving objects exist in the monitoring area, the observer wants to grasp the situation of the monitoring area from the captured image, or wants to recognize these moving objects in a short period of time. It is required to simultaneously shoot as many moving objects as possible at a recognizable resolution. However, in the above-mentioned prior art, since only a single moving object is considered and a scene in which a plurality of moving objects exist is not considered, it is not possible to appropriately move to a position satisfying such a requirement. It was.

Therefore, an object of the present invention is to make it possible to move a moving object to a position where as many moving objects as possible are simultaneously photographed with a recognizable resolution in a monitoring space where a plurality of moving objects exist.

One aspect of the present invention is an autonomous moving robot that autonomously moves in a moving space and acquires captured images of a plurality of moving objects by an imaging unit, the storage unit that stores the viewing angle of the imaging unit, and the storage unit. The minimum including the moving object position acquisition means for acquiring the moving object position of each moving object, the position estimating means for estimating the self-position of the autonomous moving robot, and the moving object position of the cluster which is a set of the moving objects. Clustering is performed so that the size of the circle is equal to or less than the upper limit size according to the viewing angle, an imaging cluster to be imaged is selected from the clusters according to a predetermined standard, and based on the cluster size of the imaging cluster and the viewing angle. A movement target position including the imaging cluster is set within the imaging range of the imaging unit, and a route search means for finding a movement route from the self-position to the movement target position and moving along the movement route. It is an autonomous mobile robot characterized by having a movement control means for controlling.

Here, it is preferable that the storage unit further stores the resolution of the imaging unit, and the path search means sets the viewing angle and the upper limit size according to the resolution.

Further, the route search means obtains a distance range from the imaging cluster according to the cluster size, the viewing angle, and the resolution of the imaging cluster, and sets the moving target position within the separation distance range from the imaging cluster. It is preferable to set.

Further, it is preferable that the route search means clusters the plurality of moving objects by setting the imaging range according to the predetermined separation distance and the viewing angle to the upper limit size.

Further, it is preferable that the route search means selects a priority imaging target from the plurality of moving objects, and selects the cluster including the moving object selected as the priority imaging target as an imaging cluster.

According to the present invention, in a monitoring space where a plurality of moving objects exist, it is possible to move as many moving objects as possible to a position where they are simultaneously photographed.

It is a figure which shows the structure of the autonomous flight robot in embodiment of this invention. It is a figure which shows the structure of the autonomous flight robot system in embodiment of this invention. It is a functional block diagram which shows the structure of the autonomous flight robot in embodiment of this invention. It is a flow chart of the route search process in embodiment of this invention. It is a figure explaining the calculation of the upper limit size in embodiment of this invention. It is a figure explaining clustering in embodiment of this invention. It is a figure explaining the separation distance range in embodiment of this invention. It is a figure explaining the generation of the movement path in embodiment of this invention. It is a figure explaining the calculation process of the local target in embodiment of this invention.

The autonomous flight robot 1 according to the embodiment of the present invention is a quad rotor type small unmanned helicopter as shown in the overview diagram of FIG. The scope of application of the present invention is not limited to the quad-rotor type small unmanned helicopter, but can be similarly applied to the single-rotor type small unmanned helicopter and the autonomously moving traveling type robot.

As shown in the system configuration diagram of FIG. 2, the autonomous flight robot 1 communicates with the external security center 100 and the management device 102, autonomously moves a plurality of moving objects M existing in the moving space, and images the images. It is configured. The moving object M is, for example, a person (thief or the like) or a vehicle that has invaded the monitoring area.

The security center 100 and the management device 102 are connected so as to be able to transmit information via an information communication network 110 such as the Internet. Further, the autonomous flight robot 1 and the management device 102 are connected so as to be able to transmit information by wireless communication or the like. The security center 100 communicates with the autonomous flight robot 1 via the management device 102, and receives an image captured by the moving object M captured by the autonomous flight robot 1. The security center 100 may have a function of performing image processing on the captured image and issuing a warning to an administrator or the like (not shown) who is monitoring an abnormality at the security center 100.

The management device 102 includes fixed moving object detection sensors 104 (104a, 104b ...) Installed on the ground, a wall surface, or the like, and detects the position of each moving object M. The moving object detection sensor 104 can be, for example, a laser sensor. The laser sensor two-dimensionally scans the laser at every angle of a certain angle sample interval to detect an object (obstacle) existing within the detection range in a horizontal plane at a certain height from the ground (or floor surface). The distance information of is acquired as a polar coordinate value. The laser sensor radially scans the exploration signal, which is the laser beam, receives the exploration signal reflected by the object and returns, calculates the distance to the object from the time difference between transmission and reception, and installs the laser sensor. The polar coordinate value of the position of the object is obtained from the coordinates of the above and the direction in which the search signal is transmitted and the calculated distance, and the three-dimensional Cartesian coordinate values (X _t , Y _t , Z _t ) are obtained from the polar coordinate value. The management device 102 transmits the position of each moving object M obtained by the moving object detection sensor 104 to the autonomous flying robot 1 by assigning a moving object ID which is an identifier of each moving object M. When the autonomous flight robot 1 receives the position of each moving object M, it calculates its own movement path based on the position and moves along the movement path. The management device 102 tracks the moving object M over the detection ranges of the plurality of laser sensors by verifying the identity of the moving objects M existing in the regions where the detection ranges of the laser sensors overlap. That is, even if the moving object M existing in the detection range of the laser sensor 104a moves to the detection range of the laser sensor 104b, the management device 102 can determine that the moving object M is the same object.

Hereinafter, the configuration and function of the autonomous flight robot 1 will be described with reference to the overview diagram of FIG. 1 and the functional block diagram of FIG.

As shown in FIG. 1, the autonomous flight robot 1 has four rotors (propellers) 2 (2a to 2d) on one plane. Each rotor 2 is rotated by using a motor 4 (4a to 4d) driven by a battery (secondary battery: not shown). Generally, in a single rotor type helicopter, the azimuth is maintained by canceling the anti-torque generated by the main rotor with the moment generated by the tail rotor. On the other hand, in a quad rotor type helicopter such as the autonomous flight robot 1, anti-torque is canceled by using a rotor 2 that rotates in different directions in the front-rear direction and the left-right direction. Then, by controlling the rotation speeds (fa to fd) of each rotor 2, various movements and postures of the aircraft can be adjusted. For example, when it is desired to rotate the machine body in the yaw direction, the rotation speeds of the front and rear rotors 2a and 2c and the left and right rotors 2d and 2b may be different.

The image pickup unit 3 is composed of, for example, an optical system such as a lens and a two-dimensional image sensor having a two-dimensional array element such as a CCD or CMOS having a predetermined number of pixels (for example, a resolution of 640 × 480 pixels), and is an image captured in a moving space. Is a so-called color camera that acquires images at predetermined time intervals. In the present embodiment, the imaging unit 3 is installed in the housing portion so that its optical axis images the front direction of the autonomous flying robot 1, and is obliquely downward from the horizontal plane (XY plane) by a predetermined depression angle θ. It is installed so that the space is imaged at a viewing angle φ (angle of view). The acquired captured image is output to the control unit 7 described later, stored in the storage unit 8 by the control unit 7, or transmitted to the management device 102 via the communication unit 9 described later.

The distance detection sensor 5 is a sensor that detects the distance between an obstacle existing around the autonomous flight robot 1 and the autonomous flight robot 1 and acquires the relative position of the obstacle existing within the sensor detection range. is there. In the present embodiment, a microwave sensor is provided as the distance detection sensor 5. The microwave sensor emits microwaves into space and detects the reflected waves to detect an object around the autonomous flight robot 1 and obtain the distance to the object. The distance detection sensor 5 can be provided, for example, toward the front of the autonomous flight robot 1 and used to measure the distance to the moving object M. Further, the distance detection sensor 5 can be provided, for example, downward on the lower part of the autonomous flight robot 1 and used to measure the distance (altitude) from the ground. Further, the distance detection sensor 5 can be provided, for example, toward the optical axis direction of the image pickup unit 3 and can be used to measure the distance to the moving object M which is the image pickup target of the image pickup unit 3.

The position detection sensor 6 is a sensor for acquiring the current position of the autonomous flight robot 1. The position detection sensor 6 receives radio waves (navigation signals) transmitted from a navigation satellite (artificial satellite) such as GNSS (Global Navigation Satellite System), for example. The position detection sensor 6 receives navigation signals transmitted from a plurality of navigation satellites and inputs them to the control unit 7. The position detection sensor 6 may use other sensors such as a laser scanner, a gyro sensor, an electronic compass, and a barometric pressure sensor to acquire information for obtaining a self-position by a known technique.

The communication unit 9 is a communication module for wireless communication with the management device 102 by, for example, a wireless LAN or a mobile phone line. In the present embodiment, the captured image acquired by the imaging unit 3 is transmitted to the management device 102 by the communication unit 9, and the captured image is transmitted from the management device 102 to the security center 100, whereby a security guard or the like remotely invades. Allows you to monitor people. Further, the communication unit 9 receives the position (coordinates: X _t , Y _t , Z _t ) of each moving object M from the management device 102, thereby enabling the setting of the moving route as described later. That is, in the present embodiment, the communication unit 9 functions as the moving object position acquisition means of the present invention.

The storage unit 8 is an information storage device such as a ROM (Read Only Memory), a RAM (Random Access Memory), and an HDD (Hard Disk Drive). The storage unit 8 stores various programs and various data, and inputs and outputs these information to and from the control unit 7. The various data include information used for each process of the control unit 7, such as moving object position 81, cluster information 82, voxel information 83, and imaging condition information 84, output values of each sensor, captured images, and the like.

The moving object position 81 is information in which the moving object ID, which is an identifier of each moving object M received from the management device 102, is associated with the position information (coordinates: X _t , Y _t , Z _t ) of the moving object M. Is. In the present embodiment, the moving object position 81 is set to the foot position of each moving object M, that is, the position where the moving object M is in contact with the ground (floor surface). When the control unit 7 receives the moving object ID and the position of each moving object M via the communication unit 9, the control unit 7 stores the moving object ID and the position as the moving object position 81 in the storage unit 8.

The cluster information 82 is information about a cluster which is a set of moving objects M, and is a cluster ID which is an identifier of the cluster, a moving object ID of the moving objects M constituting the cluster, and a cluster which is a value indicating the size of the cluster. This is information that associates the size with the cluster position that represents the position of the cluster in the moving space. In the present embodiment, the smallest circle including the moving object position 81 of each moving object M constituting the cluster is obtained, the radius of the circle is set as the cluster size, and the coordinates (X,) of the center position of the circle on the ground surface are obtained. It is assumed that Y) is set as the cluster position and calculated by the route search means 73 as described later.

The voxel information 83 is information representing the structure of obstacles in the moving space by dividing the moving space into a plurality of voxels as the voxel space, which is set in advance by an administrator or the like and stored in the storage unit 8, and will be described later. This is the information updated by the position estimation means 71 so as to be performed. In the present embodiment, the moving space is equally divided into voxels of a predetermined size (for example, 15 cm × 15 cm × 15 cm), the voxel ID which is the identifier of each voxel, and the position (three-dimensional coordinates) of the voxels in the moving space. , The voxel attribute and the voxel cost value are associated with each other and stored as voxel information 83. The voxel corresponding to the no-fly zone defines its voxel attribute as "occupied voxel" to be a space in which the autonomous flight robot 1 cannot move. The no-fly zone includes, for example, an area corresponding to an obstacle such as a building, an area outside the site where flight is permitted, and an area higher than the altitude at which flight is permitted. Then, the voxel attribute of the voxel located in the space existing near the occupied voxel is defined as "proximity voxel", and the voxel attribute of the voxel located in the other freely flyable area is defined as "free voxel".

A voxel cost value indicating the degree of occupancy is associatedly registered with each voxel so that it can be used when a movement route is generated by the route search means 73 described later. The voxel cost value takes the maximum value C _max in the occupied voxel, and is set so as to become smaller as the distance from the occupied voxel increases. For example, the voxel cost value of a certain voxel (evaluation voxel) is preferably calculated by the formula of voxel cost value = C _max × exp {−λ × R}. Here, λ is a parameter obtained by experiment, and R is the distance from the occupied voxel closest to the evaluation voxel. Then, a voxel having a voxel cost value equal to or higher than a predetermined threshold value is referred to as a "proximity voxel". Further, a voxel whose voxel cost value is smaller than the threshold value is set as a "free voxel", and a voxel cost value of a voxel regarded as a free voxel is set to 0.

The imaging condition information 84 is information representing the field of view of the imaging unit 3, and in the present embodiment, the viewing angle φ and the resolution of the imaging unit 3 are stored in the storage unit 8 as the imaging condition information 84. That is, the imaging condition information 84 is a value appropriately set by the administrator or the like according to the performance of the imaging unit 3 installed in the autonomous flight robot 1. In another embodiment equipped with the imaging unit 3 having a zoom function capable of changing the viewing angle φ and a resolution changing function capable of changing the resolution to be imaged, the current viewing angle φ and the resolution are appropriately referred to. The information may be acquired and stored in the storage unit 8 as the imaging condition information 84.

The control unit 7 is composed of a computer equipped with a CPU or the like, and has a position estimation means 71, a speed estimation means 72, and a route search means as a series of processes for performing position estimation processing, speed estimation processing, route search processing, and route tracking control. 73, movement control means 74 is included.

The position estimation means 71 performs a position estimation process for estimating the current position (self-position) of the autonomous flight robot 1 in the moving space based on the output of the position detection sensor 6.

Specifically, the coordinates of the self-position from the latitude / longitude estimated by a well-known technique based on the navigation signals from a plurality of navigation satellites obtained from the position detection sensor 6 and the altitude obtained from the distance detection sensor 5 ( X _s , Y _s , Z _s ) are calculated. Further, the attitude YAW is obtained as the self-position by receiving the output from the position detection sensor 6 such as the electronic compass and the gyro sensor. The self-position estimation method is not limited to this, and the current position of the autonomous flight robot 1 may be estimated by using another method.

The position estimating means 71 has an estimated self-position (coordinates: X _s , Y _s , Z _s and attitude YAW) and a position of a moving object M received from the management device 102 (coordinates: X _t , Y _t , Z _t ). Is output to the route search means 73. Further, the position estimation means 71 outputs the estimated self-position to the speed estimation means 72 and the movement control means 74.

The position estimation means 71 performs a process of updating the voxel information 83 based on the position of the moving object M. Specifically, a cylindrical model having a size substantially the same as the size of the moving object M predetermined in the voxel space based on the voxel information 83 of the storage unit 8 with the position of the moving object M as the center of gravity (for example, the moving object to be monitored). Assuming that M is an intruder, the voxel information 83 is updated by arranging a cylindrical model with a bottom radius of 0.3 m and a height of 1.7 m) and setting a voxel that interferes with the cylindrical model as an occupied voxel. To do. As will be described later, the autonomous flight robot 1 is flight-controlled so as not to move to the occupied voxel. However, by updating the voxel information 83 based on the position of the moving object M as described above, the autonomous flight robot 1 And contact with the moving object M can be avoided.

Since the speed estimation means 72 is used for movement control in the movement control means 74, which will be described later, the speed estimation means 72 performs a speed estimation process for estimating the current flight speed (v _x , v _y , v _z , v _yaw ) of the autonomous flight robot 1. .. In the present embodiment, the flight speed is obtained from the time change of the self-position (coordinates: X _s , Y _s , Z _s and attitude YAW) estimated by the position estimation means 71. At this time, it is preferable to estimate the flight speed by using an extended Kalman filter in consideration of the influence of measurement error and the like. In addition to this, a speed estimation method using the Doppler effect in GNSS may be used.

The route search means 73 uses the self-position estimated by the position estimation means 71, the moving object position 81 stored in the storage unit 8, and various information to obtain a movement route from the self-position to a predetermined movement target position. Performs the route search process to be calculated. Hereinafter, an example of the flow of the route search process executed by the route search means 73 according to the autonomous flight robot 1 of the present embodiment will be described in detail with reference to FIG.

In the route search process, first, a process of determining the upper limit size of the cluster in clustering described later is performed (ST1). The route search means 73 calculates the upper limit size of the cluster so that the moving object M can be recognized based on the viewing angle φ and the resolution of the imaging condition information 84. In the present embodiment, first, the maximum distance L _max from the moving object M is obtained based on the resolution. That is, the higher the resolution, the more the moving object M can be recognized even if it is farther away from the moving object M, and conversely, the lower the resolution, the more the moving object M cannot be recognized unless it approaches the moving object M. Find the maximum separation distance L _max so that it becomes large. The maximum separation distance L _max may be obtained by experimentally determining the relationship between the distance and the resolution in advance and depending on the resolution to be applied. Then, the upper limit size r _max is obtained from r _max = {L _max × sin (φ / 2)} / {sin (φ / 2) +1} by using the obtained maximum separation distance L _max and the viewing angle φ. That is, as shown in FIG. 5, the moving object M ₀ that is assumed to exist at the position of the maximum distance L _max from the autonomous flying robot 1 is set as the moving object that exists at the farthest end position of the cluster. The upper limit size r _max is obtained as the radius of the circle whose farthest end position is the position of the moving object M ₀ and which is in contact with the imaging range determined by the viewing angle φ. In the present embodiment, the upper limit size is calculated based on the viewing angle φ and the resolution of the imaging unit 3, but the present invention is not limited to this, and a fixed value determined in advance by an experiment is set and the fixed value is used. You may.

Next, in the route search process, clustering for generating a cluster composed of a combination of a plurality of moving objects M is executed by using the upper limit size r _max obtained in ST1 and the moving object position 81 (ST2). In this embodiment, clustering is performed using a known technique, the Euclidean clustering method. At this time, the Euclidean clustering distance is set so that the upper limit size r _max is the maximum radius of the generated cluster. Many clustering methods have been proposed so far, and known methods may be appropriately adopted. This makes it possible to generate a cluster consisting of a plurality of moving objects M existing at positions included in a circle having a radius of the upper limit size r _max or less. When the route search means 73 generates clusters by clustering, the route search means 73 stores the information of each cluster as cluster information 82 in the storage unit 8. FIG. 6 is a view when a part of the moving space is looked down from directly above, and is a diagram showing the result of clustering the moving objects M1 to M6 existing in the moving space with the upper limit size r _max . In the figure, moving objects M1 to M4 are clustered as cluster C1, and moving objects M5 and M6 are clustered as cluster C2. It is assumed that the cluster size, which is the radius of the clusters C1 and C2, is clustered so as to be equal to or less than the upper limit size r _max .

Next, in the route search process, a process of selecting an imaging cluster, which is a cluster to be imaged by the autonomous flight robot 1, is performed from the clusters generated in ST2 (ST3). In the present embodiment, the cluster closest to the self-position obtained by the position estimation means 71 is selected as the imaging cluster. Here, assuming that the cluster C1 of FIG. 6 is set as the selected cluster, it will be described below.

Next, in the route search process, the cluster size of the imaging cluster selected in ST3 is read from the cluster information 82, and the separation distance range indicating the range of the separation distance from the imaging cluster based on the cluster size and the imaging condition information 84. (ST4). Here, the separation distance is a relative distance between the autonomous flight robot 1 and the imaging cluster that should be maintained in the horizontal plane when flying following the selected cluster. In the present embodiment, as shown in FIG. 7, when the cluster size of the imaging cluster C ₁ is r ₁ , the minimum separation distance L _min , which is the smallest separation distance from the imaging cluster C ₁ , is set to L _min = Obtained by r ₁ / {sin (φ / 2)}. That is, the minimum separation distance L _min is obtained as the shortest distance including the imaging cluster within the imaging range of the imaging unit 3. Then, the distance range from the imaging cluster is obtained by using the minimum distance L _min and the maximum distance L _max obtained in ST1.

Next, in the route search process, the separation distance within the separation distance range obtained in ST4 is set (ST5). In the present embodiment, the separation distance is selected according to a preset policy. For example, if a policy of approaching the moving object M as much as possible is set in order to photograph the details of the moving object M, the minimum separation distance L _min , which is the smallest value in the separation distance range, is set as the separation distance. On the other hand, if the policy of keeping the distance from the moving object M as much as possible is set so as to easily avoid the attack from the moving object M, the maximum separation distance L _max , which is the largest value in the separation distance range, is set as the separation distance. .. In addition to this, the policy may be changed according to the current situation. For example, the larger the average moving speed of the moving objects M constituting the imaging cluster, the larger the separation distance may be set within the separation distance range. In either case, if the distance is within the distance range obtained in ST4, the position where the image pickup cluster is included in the image pickup range of the image pickup unit 3 can be obtained.

Next, in the route search process, the movement candidate position is set at a peripheral position separated from the imaging cluster by the separation distance using the separation distance set in ST5 and the center position of the imaging cluster read from the cluster information 82. (ST6). That is, a plurality of movement candidate positions are set in the surrounding area separated by the distance from the center position of the imaging cluster. In the present embodiment, the movement candidate position is a circular position whose radius is the separation distance set in ST5 from the center position of the imaging cluster, and the flight altitude (for example, 3 m) predetermined by the administrator or the like. ) Areas are set so that they are evenly spaced.

Next, in the route search process, each of the movement candidate positions set in ST6 is evaluated, and the movement candidate position having the highest evaluation value is set as the movement target position (ST7). At this time, first, in order to prevent the autonomous flight robot 1 from coming into contact with an obstacle, the lowest evaluation is made so that the movement target position is not set at the position included in the occupied voxel or the proximity voxel with reference to the voxel information 83. Make it a value. Further, the smaller the distance from the self-position of the autonomous flight robot 1 to each movement candidate position, the higher the evaluation value. Further, the smaller the distance from the previously set movement target position to the movement candidate position, the higher the evaluation value.

Next, in the route search process, a movement route generation process for calculating the movement route of the autonomous flight robot 1 is performed using the self-position estimated by the position estimation means 71 and the movement target position set in ST7 (ST8). ). Specifically, in the movement route generation process, first, the voxel information 82 is read from the storage unit 8, and the center of each voxel of the free voxel and the proximity voxel, which are divided movable spaces, is set as a node, and the node is adjacent to the node. A graph structure is generated using a line segment connecting nodes as an edge. At this time, it is assumed that the distance cost obtained based on the distance between adjacent nodes is set as the edge weight. After generating a graph structure for all movable voxels based on the voxel information 82 in this way, the movement from the voxel node in which the autonomous flight robot 1 exists to the voxel node in which the movement target position exists. Generate a route. Various route generation methods can be applied to the movement route generation method, but in the present embodiment, the movement route is searched by using the A * (Aster) route search method. At this time, the voxel cost and the distance cost stored as the voxel information 82 are used as the movement cost in the A * route search method. The data of the movement route generated by the route search means 73 is a set data of waypoints (x, y, z), and this information is temporarily stored in the storage unit 8. FIG. 8 is a diagram schematically showing an outline from setting a movement candidate position to generating a movement route. In the figure, a plurality of movement candidate positions Pc are set at equal intervals at a position separated from the center position O1 of the imaging cluster C1 by the separation distance R1 set in ST5. Then, Po is set as the movement target position from these movement candidate positions Pc, and the movement path W0 from the self position to the movement target position Po is generated.

The movement control means 74 uses the movement route calculated by the route search means 73, the self-position estimated by the position estimation means 71, and the flight speed estimated by the speed estimation means 72, so that the autonomous flight robot 1 can use the movement control means 74. The route tracking control is performed so as to fly along the movement route calculated by the route search means 73. Specifically, the control command value, which is the flight control value at each time, is obtained using the movement path, the self-position, and the flight speed, the motor 4 is controlled based on the control command value, and the rotation speed of the rotor 2 is determined. Control.

In the route tracking control, first, a process of calculating the nearest position (hereinafter, referred to as “local target”) to be targeted by the autonomous flight robot 1 at each time is performed. FIG. 9 is a diagram illustrating the calculation of the local target. In calculating the local target, the movement control means 74 reads the movement route generated by the route search means 73 from the storage unit 8, and the waypoint Wp1 that the autonomous flight robot 1 is aiming at at the current time and the way that has passed last time. The straight line W connecting the two points with the point Wp0 is obtained. Then, the movement control means 74 calculates the intersection points Lp'and Lp between the obtained straight line W and the sphere S centered on the self-position of the autonomous flight robot 1, and sets the intersection point Lp close to the target waypoint Wp1 as a local target. Ask as. In this way, by controlling the flight so that the autonomous flight robot 1 moves toward the local target at each time, the local target always moves on the movement path toward the movement target position Po, and the autonomous flight robot 1 always moves. Will fly along the movement path.

Next, in the route tracking control, the process of calculating the control command values u _x , u _y , u _z , u _ψ for each direction of the X, Y, Z, and yaw angles so as to fly toward the calculated local target is performed. Do. At this time, a control command value is obtained so that the difference between the current self position and the local target position becomes small. Speed Specifically, the control command value XYZ axis _{u = (u x, u y} , u z) , the self-position r = obtained by the position estimation means _{71 (X s, Y s,} Z s) and It is obtained by PID control using the velocity v = (v _x , v _y , v _z ) estimated by the estimation means 72. Each control command value of the XYZ-axis direction _{u = (u x, u y} , u z), the local target r '= (x, y, z) when the control command value is, u = Kp (r' -R) + Kd · v + Ki · e is calculated. Here, Kp, Kd, Ki is that the gain of the PID control, _{respectively, e = (e x, e} y, e z) is the integral value of the error. On the other hand, for the control command value u _ψ in the yaw angle direction, ψ'is the target angle, ψ is the attitude (angle) of the autonomous flying robot 1 estimated by the position estimation means 71, and v _yaw is the angular velocity estimated by the speed estimation means 72. Then, it is obtained by PD control such as the equation u _ψ = Kp (ψ'−ψ) + Kd · v _yaw . In the present embodiment, the target angle ψ'is set as an angle facing the direction of the center position O1 of the imaging cluster C1.

In this way, the control unit 7 sequentially repeats each process of the position estimation means 71, the speed estimation means 72, the route search means 73, and the movement control means 74 described above. As a result, the autonomous flight robot 1 of the present embodiment sequentially sets and updates the movement target position Po with respect to a position separated by the distance from the imaging cluster C1, and sequentially sets and updates the movement route each time. By going, it is possible to follow and move to the imaging cluster. In particular, the autonomous flight robot 1 in the present embodiment sets the position including the imaging cluster within the imaging range of the imaging unit 3 as the moving target position Po based on the cluster size of the imaging cluster and the viewing angle φ. .. As a result, in a moving space where a plurality of moving objects exist, it is possible to follow and move at a position where many moving objects M are simultaneously photographed.

Further, the autonomous flight robot 1 in the present embodiment has an upper limit of the cluster so that the moving object M can be recognized by the path search process (ST1) based on the viewing angle φ and the resolution of the imaging condition information 84. The size is calculated. As a result, even if a plurality of moving objects M are widely dispersed and located in the moving space, the size of the cluster generated by the clustering can be set to a size that the moving objects M can recognize. With a resolution that allows a plurality of moving objects M constituting an imaging cluster to be recognized, it is possible to follow and move while simultaneously photographing as many moving objects as possible.

Further, the autonomous flight robot 1 in the present embodiment has a separation distance range indicating a range of separation distance from the imaging cluster according to the cluster size, viewing angle φ, and resolution of the imaging cluster in the route search process (ST4). Is obtained, and the moving target position is set at a position within the separation distance range from the imaging cluster. As a result, a plurality of moving objects M constituting the imaging cluster can be recognized at a position where the imaging cluster selected from the clusters having various cluster sizes can be included in the imaging range of the imaging unit 3. Depending on the resolution, it is possible to accurately determine the position where as many moving objects as possible can be photographed at the same time.

By the way, the present invention is not limited to the above-described embodiment, and may be further implemented in various different embodiments within the scope of the technical idea described in the claims. Moreover, the effect described in the embodiment is not limited to this.

In the route search process in the above embodiment, the upper limit size of the cluster is determined in ST1, the separation distance range in which the cluster (imaging cluster) is included in the imaging range is obtained in ST4, and the separation distance is determined in ST5. The separation distance is set as the distance value within the range. However, not limited to this, the separation distance is set in advance as a fixed value in the storage unit 8 by the administrator or the like, and the movement candidate is moved to a position separated by the set separation distance from the position (for example, the center position) of the selected cluster. The position may be set. In this case, the route search means 73 has the center position of the cluster at a position separated from the self position by a preset separation distance in the route search process (ST1), and the imaging unit 3 obtained based on the viewing angle φ. The maximum cluster size that fits within the imageable range shall be calculated as the upper limit size. Then, the imaging cluster is selected from the clusters obtained by clustering (ST2) using the upper limit size (ST3), and the movement candidate position is set at a position separated by the separation distance set from the imaging cluster (ST3). ST6). In this way, even if the separation distance is a fixed value, it is possible to follow and move at a position where many moving objects M are simultaneously photographed in a moving space where a plurality of moving objects exist.

Further, in the route search process in the above embodiment, when the imaging cluster is selected in ST3, the cluster closest to the self-position is selected as the imaging cluster. However, the present invention is not limited to this, and a moving object M to be a priority imaging target is selected from a plurality of moving objects M, and a cluster including the moving object M selected as the priority imaging target is selected as an imaging cluster. Good. For example, the moving object M first detected by the moving object detection sensor 104 is set as the priority imaging target, and the cluster including the priority imaging target is selected as the imaging cluster. Alternatively, a moving object M having a predetermined image feature (for example, a moving object M possessing a weapon or the like or a moving object M wearing a mask) is detected from the detected moving objects M, and the moving object M is concerned. M may be set as the priority imaging target, and the cluster including the priority imaging target may be selected as the imaging cluster.

Further, in the above embodiment, the moving object M is detected by using the moving object detection sensor 104 connected to the management device 102. However, the present invention is not limited to this, and an image analysis means (moving object position acquisition means) for estimating the position of the moving object M by image analysis of the captured image acquired by the imaging unit 3 may be provided. For example, each frame of the captured image is image-processed to extract an image region of the moving object M. At this time, an optical flow method, which is a known technique (see JP-A-2006-146551), a classifier by boosting learning (for example, a face detection method using an AdaBoost-based classifier using the Har-like feature), etc. Is used to extract the image area (person area) of the moving object M. Next, the distance between the moving object M and the autonomous flying robot 1 is estimated based on the position of the extracted image region. Specifically, a correspondence table between the y-coordinate position (in the captured image) of the crown of the image region of the extracted moving object M and the distance is prepared in advance for each flight altitude, and the current flight altitude and the moving object M are created. The distance to the autonomous flight robot 1 is estimated by comparing the y-coordinate position of the top of the head with the correspondence table. However, the distance may be calculated from the size of the head of the extracted moving object M. That is, a correspondence table between the size of the head and the distance is created in advance, and the distance to the autonomous flight robot 1 is estimated by comparing the size of the head of the extracted moving object M with the correspondence table. May be good.

Further, the autonomous flight robot 1 is equipped with a distance image sensor as an imaging unit 3 instead of a color camera, and a moving object M is extracted by a known moving object extraction technique using the distance image acquired from the distance image sensor. Alternatively, an image analysis means (moving object position acquisition means) for estimating the position of the moving object M from the distance value between the extracted moving object M and the autonomous flying robot 1 and its own position may be provided. Further, the autonomous flight robot 1 may be equipped with a laser scanner, and the position estimation means 71 (moving object position acquisition means) may estimate the position of the moving object M using the output value of the laser scanner and its own position. ..

Further, in the above embodiment, the control unit 7 performs a series of processes of position estimation processing, speed estimation processing, movement route generation processing, and route tracking control. However, the present invention is not limited to this, and a control PC (not shown) may be prepared and the PC may be allowed to perform a series of these processes. That is, the autonomous flight robot 1 receives the control command value obtained by the position estimation process, the speed estimation process, the movement route generation process, and the route tracking control performed by the PC from the PC by wireless communication or wired communication, and controls the control. By controlling the rotation speed of the motor 4 based on the command value, the robot may fly to a target position. In this way, by sharing the above series of processes using the external PC, the CPU processing load of the autonomous flight robot 1 can be reduced, and eventually the battery consumption can be suppressed.

1 ... Autonomous flight robot 2 ... Rotor 3 ... Imaging unit 4 ... Motor 5 ... Distance detection sensor 6 ... Position detection sensor 7 ... Control unit 8 ... Storage unit 9・ ・ ・ Communication unit 71 ・ ・ ・ Position estimation means 72 ・ ・ ・ Velocity estimation means 73 ・ ・ ・ Route search means 74 ・ ・ ・ Movement control means 81 ・ ・ ・ Moving object position 82 ・ ・ ・ Cluster information 83 ・ ・ ・Voxel information 84 ・ ・ ・ Imaging condition information 100 ・ ・ ・ Security center 102 ・ ・ ・ Management device 104 ・ ・ ・ Moving object detection sensor 110 ・ ・ ・ Information communication network M ・ ・ ・ Moving object

Claims (5)

An autonomous mobile robot that autonomously moves in a moving space and acquires images of a plurality of moving objects by an imaging unit.
A storage unit that stores the viewing angle of the imaging unit,
A moving object position acquisition means for acquiring the moving object position of each moving object, and
A position estimating means for estimating the self-position of the autonomous mobile robot, and
Clustering is performed so that the size of the smallest circle including the position of the moving object in the cluster, which is a collection of the moving objects, is equal to or less than the upper limit size according to the viewing angle, and imaging is performed from the cluster according to a predetermined standard. A cluster is selected, a moving target position including the imaging cluster is set within the imaging range of the imaging unit based on the cluster size of the imaging cluster and the viewing angle, and the movement target position is reached from the self-position. A route search means for finding a movement route and
A movement control means that controls movement along the movement path, and
An autonomous mobile robot characterized by having. The storage unit further stores the resolution of the imaging unit.
The autonomous mobile robot according to claim 1, wherein the route search means sets the upper limit size according to the viewing angle and the resolution. The route search means obtains a distance range from the imaging cluster according to the cluster size, the viewing angle, and the resolution of the imaging cluster, and sets the movement target position within the separation distance range from the imaging cluster. The autonomous mobile robot according to claim 2. The autonomous mobile robot according to claim 1, wherein the route search means sets a predetermined separation distance and an imageable range according to the viewing angle to the upper limit size to cluster the plurality of moving objects. The path search means according to claim 1 to 4, wherein a priority imaging target is selected from the plurality of moving objects, and the cluster including the selected moving object as the priority imaging target is selected as an imaging cluster. The autonomous mobile robot according to any one item.

Images