舉重機器手臂之路徑規劃與動態模擬

Abstract

對於機器人相關領域中，機器手臂是一種研究最早且最基本的機構之一。在傳統工業中，我們經常使用串聯式機器手臂來幫助我們工作。按照作業的要求，往往需要控制機器手臂從某個空間位置運動到另一個空間位置。然而串聯式機器手臂最大的缺點，就是離底盤越近的驅動軸，所承受的力與力矩就越大，所以離底盤越近的桿件與驅動器便越笨重。因此讓機器手臂能自動規劃出一條省力的路徑，不但可以避免離底盤越近的桿件與驅動器過於笨重的情形，並且可以減輕驅動器的負擔。在本文中，我們以人工智慧相關領域中的一種經典演算法－Dijkstra演算法來搜尋最省力的最佳路徑。我們提出一種修正Dijkstra演算法的新方法，使其可以兼顧計算效率與最佳解的可靠度。另外我們根據B-spline Curve的理論將最佳路徑參數化，使由格點所組成的最佳路徑轉換為一條連續且平滑的曲線，藉此我們也可以計算動力學中的瞬間角速度、角加速度與力矩。最後我們使用OpenGL函式庫撰寫C++程式，來作舉重機器手臂動態模擬的工作。目前我們已經完成初步動態模擬的成果，使舉重機器手臂可以循著以Dijkstra演算法所搜尋出來的最省力路徑，將重物從低處舉到高處。

Robotic arm is one kind of mechanism, and is most basically and early in robotic research. In the traditional industry, we usually use the serial type robotic arm to help us to work. According to the work request, we often need to control the robotic arm from some space position movement to another position. However, the most serious shortcoming for serial type robotic arm is that the more drive shaft is near chassis, the more force and torque which it supports are large. Thus, links and drivers are more unwieldy. Therefore, automatically planning a path which reduces effort for a robotic arm, can avoid the situation that links and drivers are too unwieldy, and also can reduce the load of driver. In this paper, we will use one of classical algorithm in artificial intelligent, Dijkstra algorithm, to search a optimal path which take minimum effort. We suggested a method which revise Dijkstra algorithm to look after both sides the efficiency of search and the reliability of solution. Moreover we formulize the optimal path according to the theory of B-spline Curve, so the path which is composed by the grid point can transform into a continuous and smooth curve. By formulizing path we can also calculate angular velocity, angle acceleration, and torque in dynamics. Finally we use OpenGL library to design a C++ program which do the dynamic simulation of weight-lifting robotic arm. At present we already get the preliminary results in dynamic simulation. The weight-lifting robotic arm can follow the path of minimum effort that search by Dijkstra algorithm to lift a payload from low place to high.