Voice-Controlled Navigation for TurtleBot3 using ROS Noetic and Speech Recognition

This study presents a unique approach for operating the TurtleBot3 Waffle-Pi robot with voice instructions using the ROS Noetic framework. The system uses the SpeechRecognition library to translate voice inputs into movement commands, with the goal of improving user engagement and providing an intuitive control mechanism that can be applied to a variety of domains including assistive robotics. The system naturally interfaces with ROS Noetic, which uses Python for scripting and control logic. The method entails real-time recording audio with a microphone, filtering it to remove noise, then interpreting commands with Google's speech recognition API. For the TurtleBot3, commands like “forward,” “back,” “left,” “right,” and “stop” are assigned to matching movements. Noise filtering and command disambiguation problems are solved by modifying the recognizer's energy threshold and using a dynamic energy threshold. A complete test was carried out in controlled situations to determine command recognition accuracy, response time, and navigation precision. The results showed good reliability and effectiveness, with accurate command recognition and minimum reaction delay.