WiFi遠端監控機器人的智慧型避障

Abstract

本論文採用Pololu公司所研發之3pi機器人與Baby Orangutan B328 控制器，實現可由Wi-Fi遠端操控之避障與防踏空智慧型監控機器人。本系統整合了無線通訊、紅外線感測、影像擷取、馬達控制、人機介面等模組，並基於模糊控制理論(Fuzzy Logic Control, FLC)，設計出一套完整且有效的控制演算法。整個機器人系統可以區分為Baby Orangutan B328通訊影像處理與3pi機器人避障控制的兩大控制核心。 第一部分為Baby Orangutan B328微處理器，其有體積小但效率高的特性，因此適合用於小型機器人的控制；在本系統中Baby控制器負責訊號間的傳輸與處理：透過與μcam攝影鏡頭相互溝通，可以獲得即時的影像訊號；而透過WiShield V2.0無線通訊模組可將訊號傳遞給使用者端(iPad)，亦或是接收使用者端(iPad)傳來的控制指令；除了接收影像和控制指令外，紅外線感測器之測量訊號亦會經由此核心處理，將所有接收到的控制訊號處理後傳給第二個核心：3pi機器人。3pi機器人接收到控制器傳來的控制訊號後，即會與其本身的線感測器的測量值作整合，針對障礙物與懸崖的位置，將感測值透過適當的模糊處理獲得相對應之避障訊號，進而完成行進間避障的目的。經過各種不同模式的避障實驗後，本論文所提出之避障演算法可以有效的閃避各種障礙物，而此演算法可以實際應用於監控機器人在巡邏任務中進行危險偵測處理。

In this thesis, the 3pi robot and the Baby Orangutan B328 controller, which are produced by the company Pololu, are adopted to implement a remote Wi-Fi surveillance robot with obstacle avoidance. This system incorporates the Wi-Fi module, infrared sensors, the camera module, the human-computer interface, and the other modules. Based on the Fuzzy Logic Control (FLC), a complete and effective control algorithm is proposed to realize a robot for obstacle avoidance. The proposed robot system consists of two main parts: one is signal processing system (Baby Orangutan B328 controller,) and the other is obstacle avoidance control system (3pi robot.) Baby is a compact but complete controller that is especially suitable for small robot control. In this system, Baby communicates with a camera module(μcam) and acquire the real-time images from it. Then, wireless communication is set up by a WiShield V2.0 between Baby and the user ends (iPad). Baby can transfer the received images to the iPad and receive the control instructions from the iPad. Moreover, the measured values of the infrared distance sensors are also transmitted to Baby. When Baby receives all the signals, it transfers the information to the obstacle avoidance control system, i.e. 3pi robot. Receiving the instructions, 3pi robot combines them with its line sensor measurements, and utilizes Fuzzy Logic Control theory to handle the obstacle avoidance. Demonstrated by different kinds of experiments with objects or holes, the results show that the proposed robot system can effectively dodge the obstacles and cliffs. Therefore, the proposed control algorithm can be applied to real surveillance robot systems for obstacle avoidance.