Speech commands recognition applied on a mobile robot (SCRAMR)

Abstract

Since the industrial revolution, people have been looking forward to technologies that will facilitate their lives, especially handicapped people who are willing to live normal lives. In this project, we will focus on providing a real-time, offline speech recognition system that will help handicapped communicate with a mobile robot easily by voice. In essence, we propose an embedded system of a mobile robot controlled using speech recognition. Our system utilizes deep learning algorithms for real-time speech commands detection in order to achieve a more natural approach of robotic systems control, which will help minimize the distance between the digital world and the physical world. In this project, we designed, implemented, and trained an intelligent small-scale speech-controlled car. The system was able to achieve great dynamic speech commands detection as it was tested with different people’s voices. We’ve tested our system in different environments with variant noise rates and different obstacles arrangements and, as expected, it worked greatly. It returned very good results in minimal response time. Keywords: Speech recognition, Deep Learning, Mobile robot, ESP-NOW.