Teleoperation and Control of a Humanoid Robot NAO through Body Gestures

摘要

The present project focuses on controlling the humanoid robot, NAO, in real time through the gestural interpretation of the Kinect sensor and implementing linear regression techniques to adjust the movements. The system was developed under the Python programming language and the Naoqi operating system. For the teleoperation part, seven pre-established operation commands were integrated by means of corporal gestures. The system was validated by a user performing random movements that control the NAO and verifying how similar those movements were with respect to those executed by the robot. For the teleoperation part, the user navigated the NAO through a structured environment by viewing streaming video obtained directly from the camera integrated into the robot. As a result, it was observed that the robot imitated movements of the user's full body. The proposed method offers the advantage of requiring a smaller number of arithmetic operations and a smaller amount of processing compared to related works in the literature.