使用模糊邏輯之雙足機器人即時平衡控制系統設計

Abstract

長久以來，即時平衡一直是雙足機器人急需解決的重要課題。傳統上解決平衡問題之作法為利用動態方程式與逆向運動學來獲得機器人目前重心與相對姿態，但此作法複雜且運算量龐大，只適用於使用高運算能力處理器之機器人。為了解決此問題，本論文使用運算能力較為低階之微處理器(PIC-18F452)為控制核心，以仿鴕鳥構造之雙足機器人『PAPA-Man』為研究發展平台。成功提出一個以ZMP觀點搭配智慧型演算法-模糊邏輯的平衡方法，並且發展出一套機器人雙足穩定之分散式平衡控制系統架構，藉此能即時適應於不同地形與斜面。機器人並協同以Visual Basic設計pc端人機介面的操控軟體-『Balance Control Agent』，透過此視窗介面，可即時觀看目前機器人重心位置而下達行為操控之指令；進一步，本研究整合二足步行機器人之電子模組與電腦端之軟體通訊，實現即時平衡之適應能力。本設計同時希望以此達到低成本、高效率及高擴充性之優點。

Real-time balance in biped robots is an important research issue to be solved. Traditionally, inverse kinematics from dynamic equations is solved to obtain the centre of gravity and the gesture of the robot. But the computing effort induced from the complicated technique prohibits the realization of the real-time balance. In this thesis, an intelligent control algorithm using fuzzy logic and ZMP is developed to achieve real-time balance. Our research is implemented on a biped robot named “PAPA-Man” whose mechanical architecture mimics an ostrich. The goal of the system is to make the robots balance on different landforms and incline. Further, a “Balance Control Agent” is designed using Visual Basic as the control interface. The software computes the location of the central gravity synchronously and sends commands to the robot for balance control. This work integrates the electronics and firmware of the biped robots with the communication software on the personal computer to realize the ability of real-time balance. Other objectives such as low cost, high efficient and easy to expand are also considered.