Exploring the Potential of Intel Realsense T265 Camera as Visual Odometry and Tracking System for Autonomous Navigation of Four-Wheel Omnidirectional Mobile Robot

摘要

This research presents a study on exploring the potential of the Intel Realsense T265 Tracking camera implemented in Four Wheel Omni Mobile Robot (FWOMR) Navigation, where the method used by the camera device system is based on Visual odometry and Tracking. The advantage of using visual odometry over conventional (wheel) odometry in mobile robot navigation is that if there is slippage between the mechanical wheel and the floor, it will not affect the measurement of the robot's position displacement. Visual Odometry reading of Intel Realsense T265 camera provides feedback data in the form of position changes in the 2D cartesian plane coordinate system from time to time in the FWOMR control system. The greatest RMSE real measurement, measured in the z-axis at 1.61 cm, was acquired by testing the Camera T265 sensor's readout without having it installed on the robot. Using the T265 camera, Visual Odometry was successfully applied for basic kinematics movement control and tracking on FWOMR with <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\textbf{Kp} \boldsymbol{=2.0,}\ \mathbf{Ki}\boldsymbol{=0.1}$</tex> and <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{Kd}\boldsymbol{=0.1}$</tex>. The measurement results showed that the greatest RMSE was 3.47 cm on the x-axis. Additionally, FWOMR effectively completes tracking and special tracing on square and triangle trajectories, with the greatest RMSE measurement value on the x-axis being 3.24 cm.