This research focuses on the principle of human-machine interaction (HMI) to develop a real-time teleoperation system for an omnidirectional three-wheeled robot using voice commands. HMI allows humans to interact with machines, and teleoperation enables remote control through IoT technology and machine learning, enhancing the robot’s efficiency and accessibility. The research designs a mobile robot with Jetson Nano developer kit and an IMU 9DOF sensor for localization. The control interface is implemented through a mobile application using the Long Short-Term Memory (LSTM) algorithm for voice command recognition. Extensive testing with 160 trials, 40 repetitions for each voice command, evaluates the system’s accuracy and response time. The real-time teleoperation system demonstrates effective control of the omnidirectional robot with voice commands. Most commands are executed accurately, with slight deviations in specific directions. The internet-based approach ensures global control accessibility, catering to individuals with disabilities, and enhances human-robot interactions. The developed teleoperation system offers a promising solution for intuitive and efficient remote control of omnidirectional three-wheeled robots. The integration of voice commands and IoT-based technology enhances accessibility, convenience, and versatility in robot control. This research contributes to the advancement of HMI technologies and holds potential for various applications in industries and daily life, furthering progress in robotics and automation.