This paper describes in converting the assembled Bioloid premium humanoid robot from a remote controlled robot into a fully autonomous soccer playing robot with vision system. The various parts of the Bioloid humanoid robot along with camera system are assembled for real time application to soccer competition. The tasks which are accomplished in this paper include locating the ball present in the arena, walking towards the ball, identifying the goal and kicking the ball into the goal. This paper illustrates the various real-time vision techniques implemented and the inverse kinematics for the robot to execute the decisions taken by Vision Systems. The task of localizing the robot with respect to the ball and goal is accomplished by a custom developed technique. The results obtained by the vision algorithms are also included in this paper. The humanoid robot is integrated with a CMOS VGA Web Camera, which is connected to a Raspberry Pi, which runs on Linux based platform in which the Open source computer vision library (OpenCV) is installed and vision algorithms are written in C++ programming language. Humanoid robot CM530 control unit is then interfaced with the Raspberry Pi for generating TTL communication with servo actuators at each joint for obtaining humanoid motion such as kick, run, walk etc., which is required for soccer competition. The developed vision based humanoid robot is very stable, low cost, and less power consumption. Developed Vision based humanoid robot can be used for various applications like international soccer competition, security or safety, defence, assistance in medical, harvesting in agriculture and environmental hazards and other humanitarian services.
A. B. Krishnan, Aswath, S., and Udupa, G., “Real Time Vision Based Soccer Playing Humanoid Robotic Platform”, in Proceedings of the 2014 International Conference on Interdisciplinary Advances in Applied Computing, New York, NY, USA, 2014.