UAV-AGV heterogeneous multi-agent robotic system has drawn the attention of researchers to explore its capabilities in different perspectives. The UAV-AGV can concatenate their individual capabilities to overcome the drawbacks of each. UAV will benefit in payload and AGV will have the navigation guidance due to the presence of UAV. Collaborative kinematics between both agents is basic requirement of the system. Vision-based method is one of the techniques to implement collaborative motion. A high-level sliding mode controller is developed and validated for the vision-based navigation of UAV for reaching the target/AGV. Gazebo simulations are performed for trajectory tracking in the image frame to reach the target by the UAV. UAV autonomously detects the target and plans the trajectory to reach it. Apparent size-based depth controller is developed for the UAV and simulated in the Gazebo. Altitude trajectory tracking is implemented for the UAV using sliding model controller. Sliding mode based high-level controllers are performing well for the navigation of UAV and trajectory tracking in the image frame opens a different approach for the reaching of AGV by UAV. A non-linear depth controller is developed and simulated in Gazebo which can be useful for the landing task of UAV over AGV.