- Number
- 9579796
- Published
- 2017-02-28
- Filed
- 2013-09-25
- Assignee
- Disney Enterprises, Inc.
- Inventors
- Nagarajan; Umashankar et al.
- CPC
- B25J9/1671; B25J9/1605
- Verdict
- Medium Notable software
- Source
- Google Patents · FreePatentsOnline
The keeper's note
Robotics control technique reducing a humanoid robot's model for faster task-specific computation — locomotion/control policy.
Abstract
The disclosure provides an approach for automatically determining task-specific robot model reductions. In one embodiment, a simplification application determines a smallest order statespace model whose stabilizing controller also stabilizes a full-order robot model. The simplification application may determine such a model via an iterative procedure in which the reduced order is initialized to the number of unstable poles of the open-loop full-order system and, while the closed loop full-order system with the balanced reduced order system's stabilizing controller is unstable, fractional balanced reduction is applied to generate abalanced reduced system. If one or more unstable closed-loop poles exist in the full-order system with the stabilizing controller of the newly-generated balanced reduced system, the reduced order is incremented by one, and fractional balanced reduction repeated, until no unstable closed-loop poles remain. In another embodiment, the model reduction is made task-specific by formulating the full model with task-specific outputs.
Background
BRIEF DESCRIPTION OF THE DRAWINGS(1) So that the manner in which the above recited aspects are attained and canbe understood in detail, a more particular description of aspects of the invention, briefly summarized above, may be had by reference to the appended drawings.(2) The appended drawings illustrate only typical aspects of this invention and are therefore not to be considered limiting of its scope, for the invention may admit to other equally effective aspects.(3) FIG. 1 illustrates an approach for controlling a humanoid robot using a task-specificsimplified model, according to an embodiment of the invention.(4) FIG. 2 illustrates a method for determining a task-specific simplified robot model, according to an embodiment ofthe invention.(5) FIG. 3 illustrates a method for determining a minimum stable balanced reduced model and associated stabilizing controller, according to an embodiment of the invention.(6) FIG. 4 illustrates a method for controlling a robot to perform a motion using a simplified dynamics model, according to an embodiment of the invention.(7) FIG. 5 depicts a block diagram of a system in which an embodiment may be implemented.DETAILED DESCRIPTION(8) Embodiments disclosed herein provide techniques for automatically determining task-specific robot model reductions. As used herein, a model includes one or more equations with mass and dynamic properties of a robot, having joint torques as inputs and motion of the robot as output. Techniques disclosed he
Claims
1. A computer-implemented method for simplifying a model of a robot, comprising: receiving the robot model and a controller, wherein the received robot model includes one or more equationswith mass and dynamic properties of the robot used to simulate motion of the robot, and wherein the received controller is configured to compute inputs to the received robot modelfor achieving one or more control objectives; receiving task-specific output corresponding to a task or motion to be performed; and performing, via one or more processors, a search on an order of the received robot model to obtain a minimum stable reduced order robot model and a first stabilizing controller associated therewith given the received robot model, the received controller, and the received task-specific output, wherein the first stabilizing controller also stabilizes the received robot model, and wherein the search includes: initializing a reduced order to a number of unstable poles of an open-loop system of the received robot model, reducing the received robot model to a balanced reduced order robot model, obtaining a second stabilizing controller associated with the balanced reduced order robot model and capable of controlling the received robot model, if a closed loop of the received robot model with the second stabilizing controller is unstable, incrementing the reduced order and repeating the steps of reducing the received robot model and obtaining the second stabilizing controller, and if the closed loop of the received robot model with the second stabilizing controller is stable, taking the balanced reduced order robot model to be the minimum stable reduced order robot model and the second stabilizing controller to be the first stabilizing controller.
10. A non-transitory computer-readable storage media storing instructions, which when executed by a computer system, perform operations for simplifying a model of a robot, the operations comprising: receiving the robot model and a controller, wherein the received robot model includes one or more equations with mass and dynamic properties of the robot used to simulate motion of the robot, and wherein the received controller is configured to compute inputs to the received robot model for achieving one or more control objectives; receiving task-specific output corresponding to a taskor motion to be performed; and performing a search on an order of the received robot model to obtain a minimum stable reduced order robot model and a first stabilizing controller associated therewith given the received robot model, the received controller, and the received task-specific output, wherein the first stabilizing controller also stabilizes the received robot model, and wherein the search includes: initializing a reduced order to a number of unstable poles of an open-loop system of the received robot model, reducing the received robot model to a balanced reduced order robot model, obtaining a second stabilizing controller associated with the balanced reduced order robot model and capable of controlling the received robot model, if a closed loop of the received robot model with the second stabilizing controller is unstable, incrementing the reduced order and repeating the stepsof reducing the received robot model and obtaining the second stabilizing controller, andif the closed loop of the received robot model with the second stabilizing controller is stable, taking the balanced reduced order robot model to be the minimum stable reduced order robot model and the second stabilizing controller to be the first stabilizing controller.
19. A system, comprising: a processor; and a memory, wherein the memory includes an application program configured to perform operations for simplifying a model of a robot, the operations comprising: receiving the robot model and a controller, wherein the received robot model includes one or more equations with mass and dynamic properties of the robot used to simulate motion of the robot, and wherein the received controller is configured to compute inputs to the received robot model for achieving one or more control objectives, receiving task-specific output corresponding to a task or motion to be performed, and performing a search on an order of the received robot model to obtain a minimum stable reduced order robot model and a first stabilizing controller associated therewith given the receivedrobot model, the received controller, and the received task-specific output, wherein the first stabilizing controller also stabilizes the received robot model, and wherein the search includes: initializing a reduced order to a number of unstable poles of an open-loop system of the received robot model; reducing the received robot model to a balanced reducedorder robot model; obtaining a second stabilizing controller associated with the balancedreduced order robot model and capable of controlling the received robot model; if a closed loop of the received robot model with the second stabilizing controller is unstable, incrementing the reduced order and repeating the steps of reducing the received robot model and obtaining the second stabilizing controller; and if the closed loop of the received robot model with the second stabilizing controller is stable, taking the balanced reduced order robot model to be the minimum stable reduced order robot model and the second stabilizing controller to be the first stabilizing controller.