Robot and attitude control method of robot

• US 7,657,345 B2
• Filed: 08/23/2004
• Issued: 02/02/2010
• Est. Priority Date: 08/25/2003
• Status: Active Grant

• Current Assignee: Advanced Telecommunications Research Institute International, Sony Corporation (Sony Group Corp.)
• Original Assignee: Advanced Telecommunications Research Institute International, Sony Corporation (Sony Group Corp.)
• Inventors: Endo, Gen, Cheng, Gordon, Kawato, Mitsuo, Nakanishi, Jun, Morimoto, Jun
• Primary Examiner(s): Black; Thomas G
• Assistant Examiner(s): Behncke; Christine M
• Application Number:
  US10/922,907
• Publication Number:
  US 20050113973A1
• Time in Patent Office: 1,989 Days
• Field of Search: 700/245, 700/247, 700/249, 700/250, 700/19, 700/52, 318/568.2, 318/567, 318/568.1, 318/568.18, 318/568.17, 318/568.23, 901/8, 901/20, 901/50, 901/46, 901/48, 706/2, 706/10, 706/904, 706/912, 601/5
• US Class Current: 700/249
• CPC Class Codes: B25J 13/08 by means of sensing devices, e.g. viewing or touching devices
• CPC Class Codes: B25J 9/161 Hardware, e.g. neural networks, fuzzy logic, interfaces, processor
• CPC Class Codes: B62D 57/032 with alternately or sequentially lifted supporting base and legs; with alternately or sequentially lifted feet or skid B62D57/024 takes precedence

Abstract

The motion of the movable sections of the robot is taken for a periodic motion so that the attitude of the robot can be stably controlled in a broad sense of the word by regulating the transfer of the movable sections. More specifically, one or more than one phase generators are used for the robot system and one of the plurality of controllers is selected depending on the generated phase. Then, the controller controls the drive of the movable sections according to continuous phase information. Additionally, the actual phase is estimated from the physical system and the frequency and the phase of the phase generator are regulated by using the estimated value, while the physical phase and the phase generator of the robot system are subjected to mutual entrainment so that consequently, it is possible to control the motion of the robot by effectively using the dynamics of the robot.

Claims

1. A robot having a plurality of movable sections, the robot comprising:

phase signal generating means for generating a phase signal relating to a periodic motion of at least part of the movable sections;

control means for generating a control signal for the movable sections according to the phase signal;

drive means for driving the movable sections according to the control signal;

state quantity detecting means for detecting the state quantities of the movable sections driven by the drive means; and

estimation means for estimating the phase or the angular frequency of the periodic motion of the movable sections according to the state quantities;

the phase signal generating means being adapted to update the phase signal according to the phase or the angular frequency estimated by the estimation means, wherein

the control means includes a plurality of controllers having different control rules and selects an appropriate controller according to the phase signal and the selected controller generates a control signal for the movable sections according to the phase signal;

wherein

the robot includes movable sections regarded as doing a periodic motion and those not regarded as doing a periodic motion, the robot further comprising

second control means for generating a control signal according to a control rule for the movable sections not regarded as doing a periodic motion; and

coordinating means for coordinating the control quantities from the control means and the second control means in order to realize a coordinated motion for the entire robot.

2. The robot according to claim 1, wherein the movable sections have one or more than one joints and the state quantity detecting means detects the angle and the angular velocity of the joint or each of the joints as state quantity, while the estimation means estimates the phase or the angular frequency relating to the periodic motion of the joint or each of the joints according to the angle and the angular velocity of the joint or each of the joints.

3. The robot according to claim 1, wherein the robot has a plurality of movable legs as the movable sections and the state quantity detecting means detects the floor reaction force received by each of the soles of the movable legs as state quantity, while the estimation means estimates the phase or the angular frequency of the periodic motion of each of the movable sections according to the reaction force of the floor received by the sole of each of the movable legs.

4. The robot according to claim 1, wherein the robot has a plurality of movable legs as the movable sections and the state quantity detecting means detects the friction force received by each of the soles of the movable legs from the floor in the moving direction or in a direction perpendicular to the moving direction as state quantity, while the estimation means estimates the phase or the angular frequency of the periodic motion of each of the movable sections according to the friction force received by the sole of each of the movable legs from the floor.

5. The robot according to claim 1, wherein the robot has a trunk section and the state quantity detecting means detects the gradient of the trunk section as state quantity, while the estimation means estimates the phase or the angular frequency of the periodic motion of each of the movable sections according to the gradient of the trunk section.

6. The robot according to claim 1, wherein the state quantity detecting means detects the acceleration of the robot or each of the movable sections as state quantity and the estimation means estimates the phase or the angular frequency of the periodic motion of each of the movable sections according to the acceleration.

7. The robot according to claim 1, wherein the estimation means resets the phase relating to the periodic motion of the joint or each of the joints in response to the occurrence of a predetermined event.

8. The robot according to claim 1, wherein the estimation means learns a training signal relating to the state quantity detected by the state quantity detecting means and the frequency of each of the movable sections and estimates the phase or the angular frequency according to the learnt training signal.

9. The robot according to claim 1, wherein the estimation means is constituted by a neural network adapted to output the phase or the angular frequency relating to the periodic motion of each of the movable sections according to the state quantity detected by the state quantity detecting means.

10. The robot according to claim 1, wherein

the robot includes a plurality of movable sections regarded as doing a periodic motion and the control means provides each of the movable sections regarded as doing a periodic motion with a controller,

the robot further comprising coordinating means for coordinating the control quantities from the controllers of the movable sections in order to realize a coordinated motion for the entire robot.

11. An attitude control method of a robot having a plurality of movable sections, the method comprising:

a phase signal generating step of generating a phase signal relating to a periodic motion of at least part of the movable sections;

a control step of generating a control signal for the movable sections according to the phase signal;

a drive step of driving the movable sections according to the control signal;

a state quantity detecting step of detecting the state quantities of the movable sections driven in the drive step; and

an estimation step of estimating the phase or the angular frequency of the periodic motion of the movable sections according to the state quantities;

the phase signal generating step being adapted to update the phase signal according to the phase or the angular frequency estimated in the estimation step, wherein

the control step including a plurality of control sub-based on different control rules, selects an appropriate control sub-step according to the phase signal and generates a control signal for the movable sections according to the phase signal from the selected control sub-step;

wherein

the robot includes movable sections regarded as doing a periodic motion and those not regarded as doing a periodic motion, the robot further comprising

a second control step of generating a control signal according to a control rule for the movable sections not regarded as doing a periodic motion; and

a coordinating step of for coordinating the control quantities from the control step and the second control step in order to realize a coordinated motion for the entire robot.

12. The method according to claim 11, wherein the movable sections have one or more than one joints and the state quantity detecting step is adapted to detect the angle and the angular velocity of the joint or each of the joints as state quantity, while the estimation step is adapted to estimate the phase or the angular frequency relating to the periodic motion of the joint or each of the joints according to the angle and the angular velocity of the joint or each of joints.

13. The method according to claim 11, wherein the robot has a plurality of movable legs as the movable sections and the state quantity detecting step is adapted to detect the floor reaction force received by each of the soles of the movable legs as state quantity, while the estimation step is adapted to estimate the phase or the angular frequency of the periodic motion of each of the movable sections according to the reaction force of the floor received by the sole of each of the movable legs.

14. The method according to claim 11, wherein the robot has a plurality of movable legs as the movable sections and the state quantity detecting step is adapted to detect the friction force received by each of the soles of the movable legs from the floor in the moving direction or in a direction perpendicular to the moving direction as state quantity, while the estimation step is adapted to estimate the phase or the angular frequency of the periodic motion of each of the movable sections according to the friction force received by the sole of each of the movable legs from the floor.

15. The method according to claim 11, wherein the robot has a trunk section and the state quantity detecting step is adapted to detect the gradient of the trunk section as state quantity, while the estimation step is adapted to estimate the phase or the angular frequency of the periodic motion of each of the movable sections according to the gradient of the trunk section.

16. The method according to claim 11, wherein the state quantity detecting step is adapted to detect the acceleration of the robot or each of the movable sections as state quantity and the estimation step is adapted to estimate the phase or the angular frequency of the periodic motion of each of the movable sections according to the acceleration.

17. The method according to claim 11, wherein the estimation step is adapted to reset the phase relating to the periodic motion of the joint or each of the joints in response to the occurrence of a predetermined event.

18. The method according to claim 11, wherein the estimation step is adapted to learn a training signal relating to the state quantity detected in the state quantity detecting step and the period of each of the movable sections and estimate the phase or the angular frequency according to the learnt training signal.

19. The method according to claim 11, wherein the estimation step is realized by a neural network adapted to output the phase or the angular frequency relating to the periodic motion of each of the movable sections according to the state quantity detected in the state quantity detecting step.

20. The method according to claim 11, wherein

the robot includes a plurality of movable sections regarded as doing a periodic motion and the control step includes control sub-steps that correspond to respective movable sections regarded as doing a periodic motion,

the method further comprising a coordinating step of coordinating the control quantities from the control sub-steps for the respective movable sections in order to realize a coordinated motion for the entire robot.