The Optimal Movement Planning for the Reconfigurable Brachiating Robot with Multi-Degree-of-Freedom
- Paper number
IAC-05-E2.P.06
- Author
Mr. Noriaki Tawada, Keio University, Japan
- Year
2005
- Abstract
In recent years, the necessity for the work support technology in space is increasing with progress of construction of the International Space Station. On the basis of such a background, the Reconfigurable Brachiating Robot (RBR) was developed for the purpose of supporting an astronaut’s work. RBR is the quadruped robot, which has four redundant (seven degrees-of-freedom) arms connected to the center hub. Therefore, RBR is the system with multi-degree-of-freedom (total of 28). However, operation at a present stage is limited to the most fundamental walking on a fixed point, and a possibility of RBR system is not fully examined. For this reason, the purpose of this research is to propose a method of the Optimal Movement Planning (OMP) for RBR. By introducing pyramid-type hierarchical architecture, the algorithm of OMP can be described in following three levels.
- High level — Deciding optimal configuration from the task and failure mode. Examples of configuration are two arms as hands for carrying an object and the rest as walking legs, cooperation after separation, or a series connection by reconfiguration, etc.
- Medium level — The Optimal path planning of the end effector or a gait generation.
- Low level — Calculating the optimal joint angles from the given path.
This paper mainly shows the result about the low-level algorithm of OMP using proposed optimization technique explained below, and the whole OMP including the medium and high level is summarized briefly.
In use of a manipulator, the avoidance of the bad condition such as singular point, joint limits or approach with an obstacle is important. Generally, bad condition avoidance can be divided into the following two types: (A) formulizing an optimization problem using the relation of forward kinematics; (B) treating the problem mathematically using the relation of inverse kinematics. The technique proposed in this paper is based on (A) and a penalty function is introduced instead of an inequality constraint to express a flexible boundary. By including penalty functions, the objective function represents manipulator’s bad condition totally and quantitatively. Moreover, when the manipulator is not in bad condition, the convergence calculation using the pseudoinverse matrix and the Newton-Raphson method, which is the conventional technique of (B), realizes high-speed processing. The proposed technique is applied to one of the redundant arms of RBR and simulation results show effectiveness quantitatively.
- Abstract document
- Manuscript document
IAC-05-E2.P.06.pdf (🔒 authorized access only).
To get the manuscript, please contact IAF Secretariat.