The Evolution of robotics over the years has seen rapid progress, significantly transforming with the incorporation of Artificial Intelligence and Human-Interaction techniques. Traditional robot control methods rely on manual controls like controllers, human-machine interactions (HMIs), joysticks, intricate algorithms, remote devices, or predefined path planning, which may limit the accessibility and ease of use. To overcome and mitigate the drawbacks, technologies like Artificial Intelligence, and Robot Operating System (ROS), which aids in creating robotic applications with its collection of software libraries and tools, can be utilized to improve robot development for real-time applications. The proposed approach involves the ROS for robot development, incorporates google speech recognition, gazebo simulation, and RViZ visualization, facilitating both virtual testing and real-world implementation. Moreover, the robot is designed with Autodesk Fusion 360 and deployed on physical hardware to validate its practical usability. Experimental evaluation across 90 trials under varying noise conditions demonstrates a 90.2% voice command success rate, an average response time of ~1.0 second, and a navigation accuracy of 94.3% for straight paths and 87.1% for complex paths. By leveraging speech recognition and ROS-based control, Voxbot enhances accessibility, providing a hands-free and efficient approach to robotic navigation. An ablation analysis further quantifies the contributions of noise filtering, multithreading, and continuous listening to latency and accuracy.
Shanmuganathan et al. (Sat,) studied this question.