JPH05265548A - Environment learning method for mobile robot - Google Patents

Abstract

PURPOSE:To provide the environment learning method for mobile robot accurate ly measuring a mobile distance, uniting local environment information, and constituting accurate knowledge for the entire environment. CONSTITUTION:At least two mobile robots of robots A and B having selective movability, a sensor detecting a surrounding state and a radio communication equipment are located within the fluid of mutual vision. The robot B in a place 17 informs the robot A of movement information from a place 15 to a place 16, and the robot A unites the environment information on a room 29 collected at the place 15 and the received movement information to store the result. Then, the robot A at the place 16 measures the moving distance of the robot B from the place 17 to the place 18, reporting the movement information and the environment information on a relative position of the place 18 to the place 15 and a room 29 to the robot B based on the position information of the places 16 and 17. Thus, the robot B unites the environment information on a room 30 and the environment information on the room 29 with the movement information, learning the environment information on the entire moving path.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a learning system for a robot having a moving ability to learn information about an environment including a region where the robot moves.

[0002]

2. Description of the Related Art First, a robot will be described. In this specification, a robot refers to a computer including a special input device and a special output device. Here, the special input device is an input device having a function of detecting a situation around the computer, such as a television camera or a sonar sensor. The special output device is an output device that can exert a physical action on an object outside the computer, such as a manipulator or a wheel driven by an electric motor. Further, a storage device, which is one of the main components of the computer, stores a program that enables the robot to automatically perform a series of special operations using the input device and the output device. And Here, the series of special operations refers to all of the series of operations performed by the robot from the time when a given task is given to the time when the task is achieved. (Tasks will be described later.) In addition, among the robots described above, a robot equipped with at least wheels or tracks driven by an electric motor or the like as an output device and capable of moving by this is particularly It is called a mobile robot. In the following description, the robot is always a mobile robot.

Robots are often given certain work goals. For example, in the case of an industrial robot, the operator instructs the robot to move the part C from the point A to the point B. Such a work target given to the robot is called a task. It should be noted that giving a task to a robot is an anthropomorphic expression of the robot, and does not correctly represent the operation actually performed by the robot operator. However, for the sake of simplicity, we will use the anthropomorphic representation of the robot in this way. For example, the information stored in the storage device of the robot is called the knowledge of the robot, and the robot learning is to increase the knowledge based on new information input from the input device.

Now, some tasks are difficult to achieve directly, and can be achieved only by executing secondary work stepwise. A robot that, when given such a task, devises and executes the necessary secondary work and its work procedure is called an intelligent robot. The intelligent robot is necessary for, for example, a task of moving a part between two points separated by a distance in a factory to carry a part.
Since there are various obstacles in the factory, in many cases,
It is impossible for a robot to move linearly between these two points. Therefore, when such a task is given, the robot first devises a plurality of moving routes connecting the starting point and the target point based on the knowledge of obstacles existing in the factory, and selects the shortest of them. We have to identify the thing and devise an action procedure to move along the shortest path. In this case, what is important in the process of the intelligent robot devising a moving route is knowledge about obstacles in the factory.

As described above, when an intelligent robot having a moving ability tries to accomplish a task requiring the moving ability, it is necessary to have knowledge about obstacles existing in the surroundings in advance. In many cases, the environment surrounding a robot changes over time. Therefore, it is desirable that the robot itself collects knowledge about this environment by searching its surroundings and appropriately. Collecting information about the environment surrounding a robot by itself is called robot environment learning. Further, the robot is an intelligent robot hereinafter.

In the conventional method for realizing the environment learning of the robot, the robot is equipped with a sensor such as a television camera and the information about the local environment is collected by the robot, and these local information are integrated to obtain the whole environment. It was about getting knowledge about the environment.

[0007]

The above-described conventional environment learning method for a mobile robot has a problem in that it is difficult to accurately know the relative positional relationship between local environments. For example, in the case of a robot that moves by wheels, there is a method to determine the moving distance and moving direction of the robot based on the diameter and the number of rotations of the wheel, and to integrate local information by this method. According to
When the robot travels a long distance, it is difficult to obtain correct knowledge about the entire environment, because the errors resulting from the inaccuracies in the wheel diameters accumulate.

Although the above point, that is, the information about the local environment can be obtained correctly, it is difficult to integrate them to form accurate knowledge about the entire environment. This is a conventional method for environment learning of a robot. Is the problem.

It is an object of the present invention to provide an environment learning method for a robot, in which the robot correctly integrates information about the local environment obtained in environment learning to form accurate knowledge about the entire environment. is there.

[0010]

An environment learning method for a mobile robot according to the present invention has at least a moving ability capable of selecting a direction and a sensor for detecting a surrounding situation, and at least enables communication between the two. Arranging a plurality of robots including the first and second robots in positions within mutual fields of view, and the first robot measures movement distance of the second robot that moves and movement information including the measurement values. To the second robot, the second robot adds the movement information received from the first robot to the stored environment information around the position before the movement, and stores the movement environment information. The robot measures the movement information of the moving first robot and notifies the measurement value to the first robot, and the first robot uses the stored environment information around the position before the movement as the second robot. Received from Movement information including a motion distance added, second
The steps of storing as moving environment information including the relative positional relationship with the position of the robot before movement are mutually performed.

[0011]

The first robot accurately measures the movement information including the movement direction and the movement distance of the second robot by triangulation or the like and notifies the second robot of the movement information stored in the second robot. The environment information around the position and the received movement information are added and stored, and then the second robot notifies the first robot of the movement information. The information and the received movement information are added to form totally integrated movement environment information including a relative positional relationship with the position of the second robot before movement.

[0012]

Embodiments of the present invention will now be described with reference to the drawings.

FIG. 1 is a block diagram of an embodiment of a robot to which the robot environment learning method of the present invention is applied.

This robot includes a television camera 1, a support base 2 thereof, a computer 3, a wireless communication device 4, and wheels 5.
And a drive mechanism 6 for the wheel 5, a frame 7, and signal cables 8, 9, 10. The TV camera 1 is installed on the support base 2 so as to be oriented in an arbitrary direction. The wireless communication device 4 communicates information with other robots. The wheels 5 are fixed to a frame 7, and the drive mechanism 6 moves the robot in an arbitrary direction designated by the computer 3. The signal cable 8 is a signal path for transmitting a shooting instruction of the computer 3 to the television camera 1, and transmitting the image signal generated by the television camera 1 and the rotation direction and angle of the television camera 1 to the computer 3. The signal cable 9 is used for the computer 3 and the wireless communication device 4.
It is a signal path to and. The signal cable 10 is a signal path for transmitting the movement information output by the computer 3 to the drive mechanism 6. The computer 3 includes a storage device, and stores a program that realizes the operation of the mobile robot, and controls the television camera 1, the wireless communication device 4, and the drive mechanism 6 by the program.

Next, the operation of this embodiment will be described.

FIG. 2 is a diagram showing a positional relationship when the moving distance is measured using the two robots shown in FIG. 1, and FIG. 3 is shown in FIG.
It is a figure which shows the positional relationship at the time of environment learning of the robot shown in FIG.

First, the measurement of the moving distance will be described.

Robot B remains stationary at location 13 while robot A moves from location 11 to location 12. Robot B
Before the robot A starts to move, the places 13 and 11 are
The distance 19 between and is measured by shifting the camera position by the TV camera 1 or by triangulation with two cameras. Next, similarly, the distance 20 between the place 12 and the place 13 after the movement of the robot A is measured, and the angle 21 is also measured to calculate the movement distance 22 of the robot A.

Next, an environment learning method by the robots A and B shown in FIG. 3 will be described. The robot A observes the indoor environment of the room 29 before moving and learns environmental information such as the position of obstacles. Then, the robot A moves to the place 16. During this time, the robot B is stopped at the location 17, the locations 15 and 16 of the robot A are measured, the distances are calculated, and the value is notified to the robot A. Robot A
Based on the movement information received from the robot B and the environmental information about the room 29 collected at the location 15, the location 16
And the relative position of robot B's location 17 is recognized.

Next, the robot B moves to the place 18 after notifying the movement information of the robot A at the place 17. Robot A is located at location 16 during this time and observes robot B at locations 17 and 18. This observation result and place 16
The robot A calculates the relative position of the place 18 with respect to the room 29 based on the information about the positions of 17 and 17, and notifies the robot B together with the environmental information about the room 29. The robot B observes the indoor environment in the room 30, and recognizes and stores the relative positional relationship between the rooms 29 and 30 based on the observation result and the information received from the robot A. Through the above procedure, the robot B can integrate the local environment information of the rooms 29 and 30 and correctly learn the environment information of the entire movement route including these rooms.

[0021]

As described above, according to the present invention, two robots are used to accurately measure and notify each other of their respective movement distances, so that each robot can locally detect the entire movement path. There is an effect that environmental information can be accurately integrated and configured.

[Brief description of drawings]

FIG. 1 is a block diagram of an embodiment of a mobile robot to which a mobile robot environment learning method of the present invention is applied.

FIG. 2 is a diagram showing a positional relationship when a moving distance is measured using the two mobile robots shown in FIG.

FIG. 3 is a diagram showing a positional relationship during environment learning of the mobile robot shown in FIG.

[Explanation of Codes] 1 TV camera 2 Support base 3 Computer 4 Wireless communication device 5 Wheels 6 Drive mechanism 7 Frames 8, 9, 10 Signal cable 11, 12, 13, 14, 15, 16, 17, 18
Location 19, 20, 22, 23 Distance 21 Angle A, B Robot

Claims (1)

[Claims] 1. An environment learning method for a mobile robot according to the present invention has at least first and second communication capabilities capable of transmitting and receiving communication, having a direction-selectable movement ability and a sensor for detecting a surrounding condition. Locating a plurality of robots including other robots at positions within mutual fields of view, the first robot measures the movement distance of the second robot that moves, and the movement information including the measurement value is transmitted to the second robot. Notifying the robot, the second robot adds the movement information received from the first robot to the stored environment information around the position before the movement, and stores the movement environment information, and the second robot moves Measuring the movement information of the first robot, and notifying the measurement value to the first robot, and the first robot stores the environmental information around the position before movement stored therein,
A movement characterized by performing mutual steps of adding movement information including a movement distance received from the second robot and storing as movement environment information including a relative positional relationship with a position of the second robot before movement. Robot environment learning method.