| 摘要: | 本論文目的為模糊控制的設計與實現應用於機器人系統。本文將提出兩個獨立的模糊控制架構，分別運用於兩輪倒單擺自走車(two-wheel inverted pendulum, TWIP)和機器手臂系統(robot arm system)。首先，針對兩輪倒單擺自走車所提出的模糊控制架構包含四個模糊控制器，分別為模糊站立平衡控制器(fuzzy balanced standing control, FBSC)、模糊定速行進控制器(fuzzy constant velocity control, FCVC)、模糊移動定位控制器(fuzzy traveling and position control, FTPC)及模糊車身轉向控制器(fuzzy yaw steering control, FYSC)。根據兩輪車的T-S模糊模型(T-S fuzzy model)，建立對應的平行分散補償器(parallel distributed compensator, PDC)以達到兩輪車的站立平衡控制。再基於兩輪車的運動特性，可利用Mamdani型態的模糊規則庫(Mamdani-type if-then fuzzy rule base)描述並建立兩輪車的定速行進控制、移動定位控制、車身轉向控制。完整的模糊控制架構將以嵌入式設計於系統可程式化晶片(system-on-a-programmable-chip, SoPC)上的軟核心處理器(soft-core processor)實現兩輪車的控制。電腦模擬的結果將說明控制器設計概念，而實際實驗結果呈現此模糊控制架構對於兩輪車的控制效能。再者，機器手臂系統之目的在於實現物體抓取控制(object grasping control)。利用標準的逆運動學(inverse-kinematics, IK)的觀念操控機器手臂的運動，再運用雙攝影機(two-CCD)視覺回授量測機器手臂位置誤差，使用模糊規則庫去描述並建立模糊位置誤差補償器(fuzzy position error compensator, FPEC)，以調整機器手臂的定位點進而縮減位置誤差，使得機器手臂可精準到達目標位置。最後，有效抓取區域(effective grasping region)的觀念則配合模糊位置誤差補償器，使得機器手掌得以成功抓取目標物體。實驗結果將驗證控制架構對於機器手臂系統的可行性。整體而言，對於建構實體的雙輪自走車和機器手臂系統，所需要使用到的硬體與軟體技術都將在本論文做說明。This dissertation introduces the design and implementation of fuzzy controls on robotic applications including a two-wheel inverted pendulum (TWIP) system and a robot arm system. Two fuzzy control schemes are proposed for the TWIP and the robot arm, respectively. First, the control scheme for the TWIP includes four kinds of fuzzy controls which are fuzzy balanced standing control (FBSC), fuzzy constant velocity control (FCVC), fuzzy traveling and position control (FTPC), and fuzzy yaw steering control (FYSC). Based on the Takagi-Sugeno (T-S) fuzzy model of the TWIP, a parallel distributed compensator (PDC) is constructed as the FBSC. Based on the motion characteristic of the TWIP, the FCVC, FTPC, and FYSC are designed in terms of Mamdani-type if-then fuzzy rule bases (FRBs). Then the fuzzy control scheme is embedded into a system-on-a-programmable-chip (SoPC) developmental soft-core processor to implement the controls of TWIP. Computer simulations are given to illustrate the control design ideas and practical experiments are conducted to demonstrate the effectiveness of the fuzzy control scheme for the TWIP. In addition, the concerned control for the robot arm is to realize the object grasping behavior. The standard inverse-kinematics (IK) technique is utilized to manipulate the robot arm. Based on the two-CCD visual sensory feedback, an FRB is proposed as fuzzy position error compensator (FPEC) to adjust the robot gripper position and to reduce the position error, such that the gripper can accurately reach a target position. The concept of effective grasping region is further presented to collaborate with FPEC such that the robot arm can grasp a target object precisely. Experimental results are exemplified to verify the feasibility of the control scheme for the robot arm. In summary, the requisite hardware and software techniques are introduced to establish a real TWIP and a robot arm system. |
