機器手控制器設定的運動學參數與實體結構間之差異會導致機器手程式定位之精確度不良。為避免硬體修正所需之昂貴成本與技術限制，本文建議由適當的校準與軟體程式以提昇機器手之性能，期望機器手在具高重現度之特性外同時兼備較佳之精確度。

　　本篇論文討論建立線上校準自動化之雛型，藉自行設計的量測系統與機器手的程式指令配合，自動收集校準過程中所須之相關數據。規劃之校準過程中僅須機器手在工作平台之移動，因而避免了三度空間量測之障礙與高校準成本。實驗之數據經由兩步驟先後估算代表機器手機何結構之旋轉參數與平移參數。因實驗資料存在不可避免之量測誤差，本文應用數值分析中之最小平方法配合較多之量測數據，以降低量測誤差之影響。

The differences between the kinematic parametres used in a robot controller and the actual kinematic structure of a given robot can cause a significant degradation in robot accuracy. There are basically two methods to improve robot accuracy. The first method is hardware modification which will increase the cost of robots significantly. The other approach which was adopted by this paper is to calibrate the robot and to modify the corresponed software for the robot controller.

　　This paper presents an algorithm which implements the concept of automatic on-line calibration to industrial robots. The proposed calibration scheme includes the usage of a selfdesigned measurement system which eliminates the necessity of three dimensional measurement. Based on the experimental data, the geometry of a robot may be identified by means of estimating twenty-four geometric parameters in two steps. Since the collected data always involve measurement error, least squara method will be used to approach a better estimation of the kinematic parameters.

none