本論文主要目的是研究及控制四節機器人，以DSP晶片來實現控制四節機器人行為，藉由其對伺服馬達的控制來完成平面及斜坡站立，及在機構配重改變時嘗試在未知重心位置時，完成機器人所規劃的動作。 本機器人是由三個直流伺服馬達及四節連桿機構所組成，其長短左右對稱，外側連桿較短，內側連桿較長。並利用傾斜計量測機器人傾斜角度，在馬達軸心加裝編碼器以得知馬達旋轉角度及角速度。整個系統以DSP為主控核心，利用類神經演算法、模糊理論來設計控制器，整合伺服馬達控制、電路訊號擷取與訊號處理的技術。最後，模擬實驗結果印證機器人在未知載重狀態下的自我學習能力。 The objective of this thesis is to develop and control a four-link robot. A DSP chip is used to control the motion of this robot. The four-link robot will stand up vertically and stably on the horizontal and inclined surface with unknown loading on one of the link. The robot will execute the planned motion by controlling the servo motor based on the unknown COG position. The robot is composed of three servo motors, four unequal-length links and a tilt sensor. There are three joints and four links in the robot structure. The length of this robot is symmetric with respect to the second joint. The outside links are short, and inside links are long. The tilt sensor is used to measure the inclination angle of the robot. The encoders are applied to measure rotational angles and velocities of motors. The DSP is the major controller of the system, which includes artificial neural network(ANN) algorithm, fuzzy control algorithm, motor control, A/D converter and signal process etc. The simulated results indicate that the four-links robot is able to stand up by itself on the horizontal and inclined surface with unknown loading on one of the link.