利用學習機制對機器人運動的速度軌跡做指令簡化與調變

Abstract

摘要 一個稱為方波的學習策略以運動的精確度來換取簡化的指令且擷取了運動軌跡的相似性，已經在先前被提出。然而，我們考慮到的除了運動的軌跡還有運動的速度需要被學會。我們延續了這個工作而提出了另一個方法稱為三角波法來實現速度的追蹤與指令的簡化。當原來複雜的指令在容許精確度的運動下，被簡化成簡單的形式且用簡單結構及較少記憶空間的的學習機制來儲存與操作。此外，這個學習機制也可以從調整指令的大小來改變其速度，所以簡化的指令提供了一組具有不同距離及速度的運動，而毋須重算系統的動力學。以網求揮拍動作為主的實驗報告證實了我們所提出的學習機制履行了精確度與簡化指令的交換與速度的追蹤。

A learning scheme, called rectangular-wave method, has been proposed previously. It trades motion accuracy for motion command simplification and captures the similarity of position trajectory, (eg., human signature). However, we consider not only the position but also the velocity trajectories should be learnt. As a continuing work, we propose another method to implement motion command simplification and velocity tracking. We refer it as the triangular-wave method. When the original complex motion commands capable of accurate motion tracking are simplified into simple forms, the simplified motion commands can then be stored and manipulated by using learning mechanisms with simple structures and scanty memory resources, and they can be fast and smoothly executed. In addition, this learning scheme can also perform motion command scaling in speed, so that simplified motion commands be provided for a number of similar motions of different movement distances and velocities without system dynamics re-calculation. Experiments based on tennis swing motions are reported that demonstrate the effectiveness of the proposed learning scheme in implementing these accuracy-simplification and velocity tracking.