Learning Capture Points for Humanoid Push Recovery

• US 20090132087A1
• Filed: 11/19/2008
• Published: 05/21/2009
• Est. Priority Date: 11/21/2007
• Status: Active Grant

• Current Assignee: Florida Institute For Human and Machine Cognition, Honda Motor Co., Ltd. (Honda Motor Company)
• Original Assignee: Florida Institute For Human and Machine Cognition, Honda Motor Co., Ltd. (Honda Motor Company)
• Inventors: Canas, Fabian, Rebula, John, Goswami, Ambarish, Pratt, Jerry
• Granted Patent: US 8,195,332 B2
• Time in Patent Office: Days
• US Class Current: 700/258
• 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

A system and method is disclosed for controlling a robot having at least two legs, the robot subjected to an event such as a push that requires the robot to take a step to prevent a fall. In one embodiment, a current capture point is determined, where the current capture point indicates a location on a ground surface that is the current best estimate of a stepping location for avoiding a fall and for reaching a stopped state. The robot is controlled to take a step toward the current capture point. After taking the step, if the robot fails to reach a stopped state without taking any additional steps, an updated current capture point is determined based on the state of the robot after taking the step. The current capture points can be stored in a capture point memory and initialized based on a model of the robot.

Claims

1. A method for controlling a robot having at least two legs, the robot subjected to an event that requires the robot to take a step to prevent a fall, the method comprising:

determining a current capture point, the current capture point indicating a location on a ground surface;

controlling the robot to take a capture step toward the current capture point; and

responsive to the robot failing to reach a stopped state without taking any additional steps, determining an updated current capture point based on a state of the robot after taking the capture step.

2. The method of claim 1, wherein the current capture point is determined based on data retrieved from a capture point memory, wherein the data is retrieved using a current state of the robot, and further comprising:

responsive to the robot failing to reach a stopped state without taking any additional steps, updating the capture point memory based on the updated current capture point.

3. The method of claim 2, wherein the capture point memory comprises a multidimensional array storing current capture point data.

4. The method of claim 2, wherein the capture point memory comprises a neural network storing current capture point data.

5. The method of claim 2, further comprising:

determining the current state of the robot based on sensors at a plurality of positions on the robot, the sensors for measuring position or velocity.

6. The method of claim 2, further comprising:

determining a representation of the current state of the robot based on a model of the robot, the representation for retrieving data from the capture point memory.

7. The method of claim 1, wherein the current capture point is initialized based on a model of the robot.

8. The method of claim 1, wherein the updated capture point is determined based on a residual stopping energy associated with the capture step.

9. The method of claim 8, wherein the residual stopping energy is determined based on a position and a velocity of a support foot relative to a center of mass of the robot after the robot has taken the capture step.

10. The method of claim 1, wherein a stopped state comprises a state in which the robot has not fallen and in which a center of mass velocity of the robot, an angular momentum about a center of mass of the robot, and a center of mass acceleration of the robot are all less than given thresholds.

11. A system for controlling a robot having at least two legs, the robot subjected to an event that requires the robot to take a step to prevent a fall, the system comprising:

a computer-readable storage medium storing executable computer program modules configured for;

determining a current capture point, the current capture point indicating a location on a ground surface;

controlling the robot to take a capture step toward the current capture point; and

responsive to the robot failing to reach a stopped state without taking any additional steps, determining an updated current capture point based on a state of the robot after taking the capture step.

12. The system of claim 11, wherein the current capture point is determined based on data retrieved from a capture point memory, wherein the data is retrieved using a current state of the robot, and wherein the modules are further configured for:

responsive to the robot failing to reach a stopped state without taking any additional steps, updating the capture point memory based on the updated current capture point.

13. The system of claim 12, wherein the capture point memory comprises a multidimensional array storing current capture point data.

14. The system of claim 12, wherein the capture point memory comprises a neural network storing current capture point data.

15. The system of claim 12, wherein the modules are further configured for:

determining the current state of the robot based on sensors at a plurality of positions on the robot, the sensors for measuring position or velocity.

16. The system of claim 12, wherein the modules are further configured for:

determining a representation of the current state of the robot based on a model of the robot, the representation for retrieving data from the capture point memory.

17. The system of claim 11, wherein the current capture point is initialized based on a model of the robot.

18. The system of claim 11, wherein the updated capture point is determined based on a residual stopping energy associated with the capture step.

19. The system of claim 18, wherein the residual stopping energy is determined based on a position and a velocity of a support foot relative to a center of mass of the robot after the robot has taken the capture step.

20. The system of claim 11, wherein a stopped state comprises a state in which the robot has not fallen and in which a center of mass velocity of the robot, an angular momentum about a center of mass of the robot, and a center of mass acceleration of the robot are all less than given