本論文主要目的是以DSP控制晶片作為控制核心，探討四節機器人的爬行運動控制及實現。此機器人之機構為長短左右對稱，由內側較長、外側較短的四節連桿及三個直流伺服馬達組成，控制部份主要採用DSP控制軟體操作介面，硬體方面包含四節機構、直流伺服馬達、馬達驅動器、編碼器、數位訊號處理器(DSP)的整合。在爬行動作規劃時，機器人之重心目標位置與各個馬達目標角度及速度是判斷其動態平衡的重要因素。研究中，先利用類神經網路演算法訓練來達成爬行動態平衡規劃，同時以模糊理論來對機構爬行動作的重心位置作修正，使機構重心保持在平衡穩定範圍內，完成四節機器人之爬行動作，達到向前爬行移動的目的。 實驗過程中首先於個人電腦設計四節機器人之爬行動作規劃，再配合類神經網路訓練及模糊理論來完成平衡動作與爬行時重心前後偏移的模擬。最後，完成模擬結果及四節機器人爬行動作的具體實現。 The objective of this thesis is to develop a four-link robot to crawl on a horizontal flat surface smoothly by the DSP chip. The robot is composed of four unequal-length links and three DC servo motors. The length of this robot is symmetric with respect to the second joint. The inside links are long, and outside links are short. The hardware of this system includes four-link mechanism, DC servo motors, motor drivers, encoders, digital signal process (DSP) board. The center of gravity (COG) of the robot is an important factor in the crawling processes. The robot is trained to crawl by the angles and the angular velocities of motors by using the ANN algorithm. The fuzzy theory is applied to design controller for the robot, which make the center of gravity (COG) of this robot maintain in the stable region. In the experiment process, the sequence of crawling behaviors of the four-link robot is designed and assigned by a developed software program in a PC. Then, the center of gravity (COG) of the robot is controlled to crawl stably by the artificial neural network (ANN) and the fuzzy theory. Finally, the developed robot is able to crawl smoothly on a horizontal flat surface.
