부산대학교 도서관

UAV-AGV multi-agent system is attracting the mobile robotics research community due to its applications in several engineering disciplines. It's a heterogeneous combination of UAV and AGV. Collaborative kinematics between both agents is necessity of the system for better utilization of independent capabilities. Vision-based method is most focused in the literature to build the collaboration between the agents which is of great use in the GPS denied environments. One of the collaborative tasks of UAV is to approach the AGV and hover over it at certain altitude. A trajectory tracking technique in the image frame is developed for the task in the present work. UAV detects the static target/AGV in the image plane and plans 3- dimensional trajectory to reach and hover at a desired depth from it. Minimum acceleration trajectories are designed using position and apparent size of the target in the image plane. Sliding mode control theory is applied for designing the controllers for approach, depth motion of UAV and validated with the simulations. The proposed image plane-based trajectory tracking technique is further validated in the Gazebo simulation environment. Simulation results shows that the method is useful in GPS denied environments for the UAV to plan its collaborative navigation in the image frame for smoothly reaching a desired position from the target/AGV.