Range Estimation and Visual Servoing of a Dynamic Target using a Monocular Camera

This paper delves into the problem of tracking an unknown maneuvering target using only monocular visual feedback. It is usually difficult to perform target tracking using only monocular vision due to the absence of depth information. This restricts the use of commonly used Image and Position-based Visual Servoing methods. Hence, range estimation from monocular images has paramount importance if a conventional position based path planning algorithm is to be utilized. In this paper, we circumvent this problem using the angle subtended by the target on the follower in both 3D and image coordinate systems. Also, an Extended Kalman Filter is used to filter and get accurate range estimates of the moving target. The proposed method is extensively validated on a simulated setup created in Gazebo. Also, a Proportional Derivative (PD) controller is used to maintain a fixed standoff distance from the moving target while following it.