A novel Kinect-based system for 3D moving object interception with a 5-DOF robotic arm

Abstract

This study proposes a method for performing velocity estimation by motion blur in a single image frame and intercepting a moving object with a robotic arm. It is shown that velocity estimation in a single image frame improves the system's performance up to five times. The majority of previous studies in this area require at least two image frames to measure the target's velocity. The speed estimation is practiced by a Kinect camera and the implemented control system converts the position and velocity information of the moving object to the suitable joint angles and torques to intercept within a certain time interval. To test the system, two different sets of experiments are designed in which the target swings on a string or is thrown within the working volume of the robotic arm. According to the recorded velocity data and the success of the experiments, the limitations of the method and setups will be discussed.