Work in Progress: Enhancing Human-Robot Interaction through a Speech and Command Recognition System for a Service Robot Using ROS Melodic

Luis Emiliano Rodríguez Raygoza, Juan Carlos Tudon Martínez, Jorge de Jesús Lozoya-Santos, Luis Carlos Félix Herrán

TL;DR: The paper presents the development and evaluation of a speech recognition system for an autonomous service robot, focusing on human-robot interaction. The system aims to improve command accuracy and reliability in real-world scenarios.

Abstract: This paper presents the development and evaluation of a Speech and Command Recognition system integrated into PiBot, an autonomous service robot developed at Tecnológico de Monterrey. The system executes on Robot Operating System (ROS) Melodic framework running on a Jetson TX2 embedded computer to enable natural language interaction through Automated Speech Recognition (ASR). The study focuses on the challenges and opportunities of implementing speech recognition in real-world environments, particularly within constrained hardware platforms. The system achieved a 25\% Word Error Rate (WER) and a 73\% Command Accuracy, with performance varying across different testing environments. The system achieved a 25\% Word Error Rate (WER) and a 73\% Command Accuracy, with performance varying across different testing environments. Difficulties were noted in recognizing uncommon or non-Spanish words. A comparison with state-of-the-art models indicates room for improvement. Future work will focus on fine-tuning the model using datasets with ground truth transcriptions to enhance reliability in complex, noise-prone settings.

Submission Number: 152