應用於非連續河石地形四足機器人的創新設計與實作

Abstract

本研究針對非連續河石地形，設計輕量化、跨步靈巧的四足機器人，透過本研究規劃的平衡步驟，與足部摩擦的調整，完成可用單晶片嵌入式系統控制的跨越河石地形的足部規劃與實作平台，可提供自行訓練或擴增智能型地形偵查定位與跨步理論（例：ANFIS）的附加程式功能，逐步加強與增進四足機器人（NC-F4-09）跨越河石地形的功效。 隨著近代機器人的發展，開始走入生活之中，故接下來機器人設計所將面對，是越來越複雜的地形條件。本研究旨在發展具備高度地形適應能力之四足機器人，在兼顧機動性的前提下，適應廣泛的地面複雜度。 研究中先訂定快速動作、高地形適應性與輕量化為目標，分析先前開發之四足機器人（Quadruped Robot）NCTU-CIDM-F4-2005，運用TRIZ理論推導，提出新的四足機器人之設計架構，並對控制系統之架構與動作流程進行規劃。依據提出之設計架構，以CAD軟體Solidworks進行虛擬模型干涉檢查與零件強度分析，並尋找較佳的機器人工作空間配置。另對於機器人行走動作規劃，透過地形資料座標化，應用反向運動學配合ZMP理論，建立機器人穿越非連續複雜地形的移動規則。 接著將所完成之設計結果，建立以單晶片Basic Atom為處理核心，致動器為伺服馬達之四足機器人原型，運用先前所規劃之跨越非連續地形移動規則所發展之路徑產生程式，對此四足機器人之非連續地形行走能力進行測試，經由穿越隨機產生之非連續非平坦地形，證明此研究所推導之設計架構，能夠滿足本四足機器人平台設計之目標。

With the development of modern robots, they are involved in people's daily life. There is no doubt that more and more complex terrain conditions will be the most important challenge to be dealt with. The main target of this study is trying to develop a quadruped robot with wide range of terrain adaptability, and also high mobility. In this study, we have set fast movement, wide range of terrain adaptability and light weight as our design goals. Through analyze our previous quadruped robot---NCTU-CIDM-F4-2005, we got new hints for design with TRIZ theory. The design hint will use in quadruped robot structure, control system structure and motion planning process. To establish an effective design of robot structure, a virtual robot has been use in structure adjustment and motion simulation. That is to avoid interference problems and optimize working space. And for path generation, by setup a coordinate on the terrain, here we use inverse kinematics and ZMP theory to build a system, which will be used to guide quadruped robot over through discontinuous terrain. Follow-up with the above design results, a test prototype of quadruped robot was established with servo motor and Basic Atom processer. With the path generating system for discontinuous terrain, the terrain adaptability of this robot would be experimented by random discontinuous terrain test. The result of these test have prove the robot design could achieve our final goal.