Critical element study on autonomous position control of articulated-arm type manipulator

摘要

An articulated-arm type manipulator can be operated effectively in a restricted space due to its flexibility and it can be attractive for a wide range of in-vessel maintenance such as viewing, inspection and limiter handling in fusion experimental reactors. In case of the in-vessel maintenance using a flexible manipulator, it is quite essential to develop an autonomous control method for compensating a deflection of manipulator so as to minimize the maintenance time with high precision. For this purpose, a new position control method using a combination of neural network predictor with a rigid inverse kinematics is being developed. The key features of this method are to simplify a kinematics modeling of flexible manipulator, to enable quick position compensation in stead of ordinary large matrix compensation, and to be applicable to a wide variety of manipulator characteristics. A sub-scaled model of flexible manipulator with 4 joints has been fabricated for a benchmark experiments of the autonomous position control. Comparing analytical simulation with experiments using the flexible manipulator, it has been demonstrated that the new position control method gives significant improvement in control performance with high precision in order of a figure. In addition, further optimization can be possible by adding other non-linear predictors such as radial basis function and fuzzy modeling. This paper describes the details of a sub-scaled flexible manipulator and a neural network position control system as well as results of analytical simulation and benchmark experiments. (author). (ERA citation 20:011860)