## Comparative experiments on task space control with redundancy resolution (2005)

Venue: | in IEEE International Conference on Intelligent Robots and Systems |

Citations: | 19 - 11 self |

### BibTeX

@INPROCEEDINGS{Nakanishi05comparativeexperiments,

author = {Jun Nakanishi and Michael Mistry and Stefan Schaal},

title = {Comparative experiments on task space control with redundancy resolution},

booktitle = {in IEEE International Conference on Intelligent Robots and Systems},

year = {2005},

pages = {1575--1582},

publisher = {IEEE}

}

### OpenURL

### Abstract

Abstract — Understanding the principles of motor coordination with redundant degrees of freedom still remains a challenging problem, particularly for new research in highly redundant robots like humanoids. Even after more than a decade of research, task space control with redundacy resolution still remains an incompletely understood theoretical topic, and also lacks a larger body of thorough experimental investigation on complex robotic systems. This paper presents our first steps towards the development of a working redundancy resolution algorithm which is robust against modeling errors and unforeseen disturbances arising from contact forces. To gain a better understanding of the pros and cons of different approaches to redundancy resolution, we focus on a comparative empirical evaluation. First, we review several redundancy resolution schemes at the velocity, acceleration and torque levels presented in the literature in a common notational framework and also introduce some new variants of these previous approaches. Second, we present experimental comparisons of these approaches on a seven-degree-of-freedom anthropomorphic robot arm. Surprisingly, one of our simplest algorithms empirically demonstrates the best performance, despite, from a theoretical point, the algorithm does not share the same beauty as some of the other methods. Finally, we discuss practical properties of these control algorithms, particularly in light of inevitable modeling errors of the robot dynamics. Index Terms — Redundancy resolution, Inverse kinematics, Task space control, Null space optimization, Dynamical decoupling

### Citations

329 | Robot Learning From Demonstration
- Atkeson, Schaal
- 1997
(Show Context)
Citation Context ...ching for a static target; for example, tasks which require control of position, velocity and/or acceleration of the task coordinates, as is the case when tracking a moving target or balancing a pole =-=[10]-=-. From our point of view, even after more than a decade of research, inverse kinematics control remains a surprisingly incompletely understood theoretical topic, and also lacks a larger body of thorou... |

322 |
A unified approach to motion and force control of robot manipulators: the operational space formulation
- Khatib
- 1987
(Show Context)
Citation Context ...zation in order to determine the inverse kinematics transformation [1]. Within this general framework, kinematic redundancy can be resolved at the velocity [2], acceleration [3]–[6] and torque levels =-=[7]-=-, [8], where the desired joint velocities, accelerations, and torques are computed, respectively, for a desired endeffector velocity, acceleration, and force. It should be noted, however, that many of... |

161 |
Automatic supervisory control of the configuration and behavior of multibody mechanisms
- Liégeois
- 1977
(Show Context)
Citation Context ...solution offers a wide variety of techniques. A typical approach makes use of the Jacobian pseudoinverse with local null-space optimization in order to determine the inverse kinematics transformation =-=[1]-=-. Within this general framework, kinematic redundancy can be resolved at the velocity [2], acceleration [3]–[6] and torque levels [7], [8], where the desired joint velocities, accelerations, and torqu... |

94 |
Model-Based Control of a Robot Manipulator
- An, Atkeson, et al.
- 1988
(Show Context)
Citation Context ...x. As CAD data only captures the complex dynamics of idealized rigid bodies, and not the hydraulics routing throughout the arm and the hydraulic actuators, we resorted to parameter estimation methods =-=[13]-=- to obtain the necessary inertial and center of mass parameters of the robot arm. Interestingly, the base parameter set obtained from this least squares estimation procedure resulted in a non-positive... |

80 | Learning inverse kinematics
- D’Souza, Vijayakumar, et al.
- 2001
(Show Context)
Citation Context ...redundancy resolution. In our previous studies, we have addressed computational efficiency in redundancy resolution algorithms and inverse kinematics learning based on velocity based approaches [11], =-=[12]-=-. From a more algorithmic point of view, we are interested in developing a robust inverse dynamics controller with redundancy resolution that satisfies the following requirements: 1) avoids high-gain ... |

67 |
Redundancy resolution of manipulators through torque optimization
- Hollerbach, Suh
- 1987
(Show Context)
Citation Context ...ocal null-space optimization in order to determine the inverse kinematics transformation [1]. Within this general framework, kinematic redundancy can be resolved at the velocity [2], acceleration [3]–=-=[6]-=- and torque levels [7], [8], where the desired joint velocities, accelerations, and torques are computed, respectively, for a desired endeffector velocity, acceleration, and force. It should be noted,... |

52 |
Taskpriority based redundancy control of robot manipulators
- Nakamura, Hanafusa, et al.
- 1987
(Show Context)
Citation Context ...n pseudoinverse with local null-space optimization in order to determine the inverse kinematics transformation [1]. Within this general framework, kinematic redundancy can be resolved at the velocity =-=[2]-=-, acceleration [3]–[6] and torque levels [7], [8], where the desired joint velocities, accelerations, and torques are computed, respectively, for a desired endeffector velocity, acceleration, and forc... |

32 | Inverse kinematics for humanoid robots
- Tevatia, Schaal
(Show Context)
Citation Context ...s for redundancy resolution. In our previous studies, we have addressed computational efficiency in redundancy resolution algorithms and inverse kinematics learning based on velocity based approaches =-=[11]-=-, [12]. From a more algorithmic point of view, we are interested in developing a robust inverse dynamics controller with redundancy resolution that satisfies the following requirements: 1) avoids high... |

30 |
Dynamic control of redundant manipulators
- Hsu, Hauser, et al.
- 1989
(Show Context)
Citation Context ...th local null-space optimization in order to determine the inverse kinematics transformation [1]. Within this general framework, kinematic redundancy can be resolved at the velocity [2], acceleration =-=[3]-=-–[6] and torque levels [7], [8], where the desired joint velocities, accelerations, and torques are computed, respectively, for a desired endeffector velocity, acceleration, and force. It should be no... |

24 | A unifying methodology for the control of robotic systems
- Peters, Mistry, et al.
- 2005
(Show Context)
Citation Context ...udy suggests that some of the redundacy resolution algorithms discussed in this paper can be derived from a generalized Gauss’ principle with different metrics for a cost function in a unified manner =-=[16]-=-. We hope that this result provides insight into further understanding of the properties of these redundancy resolution algotithms. Future work will address further experiments with the suggested vari... |

8 |
Load independence of the dynamically consistent inverse of the jacobian matrix
- Featherstone, Khatib
- 1997
(Show Context)
Citation Context ...n in order to determine the inverse kinematics transformation [1]. Within this general framework, kinematic redundancy can be resolved at the velocity [2], acceleration [3]–[6] and torque levels [7], =-=[8]-=-, where the desired joint velocities, accelerations, and torques are computed, respectively, for a desired endeffector velocity, acceleration, and force. It should be noted, however, that many of the ... |

3 | Issues in acceleration resolution of robot redundancy - Luca, Oriolo - 1991 |

3 |
Quasioptimal control of space redundant manipulators
- Senda
- 1999
(Show Context)
Citation Context ...o the null space of the Jacobian J. Note that by equating (6) and the analytical differentiation of (5), sufficient conditions for these two resolution schemes to be consistent are derived in [4] and =-=[5]-=-, respectively, as and ξ 2 = ˙ J † J( ˙q − ξ 1)+ ˙ ξ 1 ξ 2 = ˙ J T (J † ) T ( ˙q − ξ 1)+ ˙ ξ 1 where ˙ J † = d dt (J† ) for notational convenience. Note that it can be shown that (7) and (8) are equiv... |

3 |
Verification of the Positive Definiteness of the Inertial Matrix of Manipulators Using
- Yoshida, Khalil
- 2000
(Show Context)
Citation Context ...stent base parameter set. If this physically inconsistent model were used for inverse dynamics control, the performance of the robot would be degraded and sometimes could even cause instabilities. In =-=[14]-=-, a sufficient condition is discussed to test whether the inertia matrix is positive definite, and to check whether a given parameter set is physically consistent. However, this method does not provid... |

3 |
Physically Consistent Nonlinear Parameter Identification of Rigid Body Dynamics,” (in preparation
- Schaal, Cory, et al.
(Show Context)
Citation Context ...r this purpose, we derived a novel nonlinear parameter estimation method, which ensures physically consistent inertia parameters by using a reparameterization with the help of Cholesky decompositions =-=[15]-=-. Without this nonlinear parameter identification method, most of the above controllers would not be implementable in a stable fashion. V. EXPERIMENTAL EVALUATIONS The purpose of this paper is to impl... |

1 |
A natural resolution of bernstein’s degrees-of-freedom problem in case of multi-joint reaching
- Arimoto
- 2004
(Show Context)
Citation Context ...ements, like pointing, reaching, or circular drawing movements. Moreover, stability properties of most algorithms are only incompletely understood. A noteworthy exception is a recent study by Arimoto =-=[9]-=-, which proposes a provably stable redundancy resolution algorithm for reaching movements based on the Jacobian transpose with joint velocity damping. One of the advantages of this approach is that it... |