Voice commanded mobile robot for search and follow operations in unknown environments

Abstract

We are increasingly becoming accustomed to living and working side by side with robots. Apart from supporting us by doing hazardous and mundane jobs, robots are becoming indispensable in a number of industries. Most mobile robots implement a speech interface in order to enhance their interaction with humans. On the other hand, autonomous mobile robots are becoming more robust and self-sufficient, hence requiring less intervention from human operators. The scope of this project is to integrate human-robot interaction with autonomy. This is achieved by enabling a speech interface, using the versatile and state-of-the-art Google speech recogniser and eSpeak speech synthesiser. The Khepera III mobile robot is used to follow the user‟s spoken instructions, and is guided solely by the sonar and infrared sensors mounted on the robot itself. Therefore, no maps or expensive sensors, such as laser rangefinders, are needed. The programmed robot is able to explore and reach an environment without crashing into any obstacles. Furthermore, with the use of the on-board CMUcam5 camera, the mobile robot is able to identify a user-specified target by its colour, approach it, and keep tracking it. This dissertation starts by presenting several works about speech interfaces, including Speech Recognition and Speech Synthesis. Furthermore, a study on different methods for autonomous robot search, exploration and navigation in the absence of a map is given. A selected number of these methods are then implemented and tested. Finally, the integrated system is designed and implemented on a real, experimental setup. This is followed by a thorough experimentation and testing procedure using both simulated and real environments. The obtained results are then analysed and discussed.