Ball Trajectory Estimation and Robot Control to Reach the Ball Using Single Camera

Abstract

In robotics research, catching a projectile object with a robotic system is one of the challenging problems. The outcome of these researches can be used in a wide range of applications such as video surveillance systems, analysis of sports videos, monitoring programs for human activities, and human-machine interactions. In this paper, we propose a new vision-based algorithm to estimate the trajectory of a projectile, which estimates the time and the position of the projectile's collision with the robot's working space in real-time. We use sub-pixel calculations and present an improved algorithm for estimating the center of the ball. We evaluate the performance of different trajectory estimation algorithms and also provide a real-time hardware implementation of our method on a designed robot. Moreover, the combination of single-camera and gyroscope information is studied in this paper. The results show that the proposed algorithm is capable of correctly estimating the ball's trajectory and has a very good performance against the noise.