الفهرس 
INTRODUCTIONOnly 14 pages are availabe for public view
 |  | from | 95 |  |  |
| | | from | 95 | | |
Abstract
In This thesis, development of a human assistant is discussed. This assistant could help humans based on a given utterance together with the help of machine vision using a camera installed on the robot. The utterance, Modern Standard Arabic (MSA) or Egyptian dialect, could be a question about something in the assistant’s workspace or a request e.g., a task of holding anything or a task to pick and place. The utterance is processed through a mix of three previously used techniques such as natural language processing through semantic parsing, sentence similarity, and pattern matching rather than using each one alone. The techniques are tweaked to evolve an algorithm that can deal with an utterance even if two languages or more exist. The algorithm can handle grammar free utterances. The algorithm managed to solve the problems found by the previously mentioned techniques. The user isn’t forced to speak in a specific order. Mainly, the focus of this thesis is the Arabic language, however it explains how we can merge different languages together.
In chapter 1, Important definitions, and jargons in the field of natural language processing are explained. Quick briefs are stated about the NLP divisions and the researched topics in this field.
In chapter 2, Literature review is provided about the previously used algorithms and the commonly used techniques discussing their pros and cons. Mainly, there are two approaches in NLP field. The first is rule based technique and the second uses machine learning.
In chapter 3, Details are given about the analysis of the utterance and how the conversation act is deduced. Problems and difficulties in the processing are reviewed.
In chapter 4, The machine vision software is developed. The algorithm used to detect shapes is clarified. Linking between the processed image and the processed utterance, to build an embodied agent, is explained.
In chapter 5, A real robot equipped with a camera is used to apply and test the novel algorithm in the industry. Extra functions are added such as object picking and placing. Therefore, Extra difficulties are added to the utterance.