Developing and modelling of satellite docking algorithm

Abstract

In this study, a stereovision sensor system hardware and an algorithm are developed to be utilized in autonomous satellite rendezvous applications. A two dimensional representative environment is also being developed consisting of omni wheel robots to test breadboard model of the sensor system in a similar proximity operation scenario. The sensor hardware is designed according to predefined requirements keeping the limitations of the satellites. The sensor is able to estimate position, orientation, linear and angular velocity of a target object whose shape and sizes are known a priori. The detection system relies on specific reference markers and extracted BRISK feature points. Both of stereo and monocular vision approaches are used to detect object and estimate its distance, followed by reverse rigid body transformation to estimate the target?s 3D location and orientation. Time difference between two subsequent frames is used for estimating linear and angular velocities. Additionally, a path-planning algorithm is developed to approach target object in an efficient way.