This paper proposes a new object manipulation method by means of a multi-fingered hand using a particle filter to estimate a position and attitude of an object. In our previous work , the object virtual frame has been presented instead of an actual frame to manipulate an object without any external sensing information. A desired position and attitude of the object, however, has not been necessarily realized using the method because of no external sensing information. In order to cope with that, a new virtual frame is proposed based on a 3D laser range sensor data including considerable noise and time-delay in this paper. Firstly a particle filter is designed to estimate the position and attitude of the object from the sensing information. Then the new virtual frame is designed based on estimations of the particle filter.
|Translated title of the contribution||1P2-O08 Object Manipulation using a Particle Filter to Estimate a Position and Attitude of an Grasped Object|
|Publication status||Published - May 17 2015|