Network-based robot system and method for action execution of robot

• US 8,265,789 B2
• Filed: 12/02/2008
• Issued: 09/11/2012
• Est. Priority Date: 12/03/2007
• Status: Active Grant

• Current Assignee: Electronics and Telecommunications Research Institute
• Original Assignee: Electronics and Telecommunications Research Institute
• Inventors: Lee, Seung-Ik, Kim, Sung Hoon
• Primary Examiner(s): Black, Thomas G.
• Assistant Examiner(s): LOUIE, WAE LENNY
• Application Number:
  US12/326,618
• Publication Number:
  US 20090143909A1
• Time in Patent Office: 1,379 Days
• Field of Search: 700/245, 700/254, 700/250, 700/259, 700/264, 901/50, 318/567, 318/568.1, 318/568.11
• US Class Current: 700/245
• CPC Class Codes: B25J 9/1661 characterised by task planning, object-oriented languages
• CPC Class Codes: G05B 2219/40394 Combine offline with online information to generate robot actions
• CPC Class Codes: G05B 2219/40395 Compose movement with primitive movement segments from database

Abstract

A network-based robot system includes an activity creating device creating activities in hierarchical structure, and a robot downloading the activity tree over a communication network and executing an activity selected in the activity tree. The activities describe action execution procedures to be run by the robot and are arranged in an activity tree.

Claims

1. A network-based robot system comprising:

an activity creating device creating activities in hierarchical structure, each activity describing action execution procedures to be carried out and being arranged in an activity tree;

a robot downloading the activity tree over a communication network and executing an activity selected in the activity tree,

wherein the robot comprises;

an activity storage storing the activity trees provided through the communication network;

an activity manager searching for, in response to an activity execution instruction, an activity tree having an activity corresponding to the execution instruction from the activity storage;

an activity interpreter interpreting the activity tree searched by the activity manager;

an activity executor selectively executing activities in the interpreted activity tree; and

a plurality of primitive actions to be taken by the robot.

2. The network-based robot system of claim 1, wherein the activity is comprised of plural modes, each mode representing what action of the robot is taken by the activity at a moment, the activity including:

start mode starting when the activity is executed;

goal mode being reached when a predetermined goal has finished by an execution of the activity; and

history storing previous execution records of the activity,

wherein each of the modes includes a mode transition decision routine for the transition decision of the mode, a mode execution time, an entry action at the mode execution time, an exit action for the mode transition, and a mode action for executing the mode.

3. The network-based robot system of claim 2, wherein each of the entry action, the exit action, and the mode action is expressed in a manner that the action is completed when a plurality of activities or primitive actions is executed concurrently and are all finished.

4. The network-based robot system of claim 2, wherein each of the entry action, the exit action, and the mode action is expressed in a manner that the action is completed when one of activities or primitive actions is finished.

5. The network-based robot system of claim 2, wherein each of the entry action, the exit action, and the mode action is expressed in a manner that a single activity is executed when the action executes one single action.

6. The network-based robot system of claim 2, wherein each of the entry action, the exit action, and the mode action is expressed in a manner that a single primitive action is executed when the action executes one single action.

7. The network-based robot system of claim 2, wherein the start mode is executed depending on execution records in the history.

8. The network-base robot system of claim 2, wherein the robot further comprises:

one or more sensors mounted on the robot collecting surrounding conditions of the robot at a predetermined time interval; and

a world model storing the surrounding conditions to reflect changes in the surrounding conditions,

wherein the activity executor decides whether the mode in the activity should transit based on the world model, when the mode in the activity is executed.

9. The network-based robot system of claim 2, wherein the activity executor calls out a child activity in the activity tree or the primitive action when the activity is executed.

10. A method for action execution of a robot in a network-based robot system, comprising:

creating activities in hierarchical structure, each activity describing action execution procedures to be run by the robot and being arranged in an activity tree;

downloading the activity tree via communication network to storing the activity tree in the robot;

searching for, in response an execution instruction of the robot, the activity tree including a target activity corresponding to the execution instruction;

interpreting the searched activity tree; and

executing the target activity in the interpreted activity tree,

wherein executing the target activity includes;

setting a root activity in the activity tree as an execution activity;

executing the execution activity;

if the execution activity calls out a lower activity in the activity tree when it is executed, setting the called lower activity as the execution activity to execute the execution activity; and

if the lower activity is not called out, executing a preset primitive action to be taken primitively by the robot.

11. The method of claim 10, wherein the activity is comprised of plural modes, each mode representing what action is taken by the activity;

and

wherein executing the execution activity executes the modes within the execution activity.

12. The method of claim 11, further comprising:

updating changes in surrounding conditions of the robot;

deciding whether a current mode taken by the execution activity needs to transit any of the modes when the execution activity enters the current mode based on surrounding conditions;

deciding whether the current mode has been an activated mode in a previous execution if the current mode does not need the mode transition;

executing entry action within the current mode and then mode action if the current mode is a non-activated mode in the previous execution;

executing mode action within the current mode and the primitive action if the current mode is an activated mode; and

executing exit action within the current mode and transiting the current mode to any of the modes if the current mode needs the mode transition.

13. The method of claim 12, wherein deciding whether the current mode has been an activated mode is performed based on mode execution time set within the current mode.

14. The method of claim 12, wherein transiting the mode to any of the modes includes initializing mode execution time within the current mode and then transiting the current mode to any of the modes.