Yuarm: A low cost Android platform for vision based manipulators control

Robot technologies are accelerating rapidly. Today many robots are inspired by the shape and function of the human body. One of the popular system is the vision based robot control. By mimicking human eyes, a cameras dictates the movements of the robots. Considered as an expensive solution, the emerging smart phone technologies make a vision based system an accessible and cheap solution for mimicking human eye behavior. This paper proposed a low cost platform for vision based manipulator control that consists of used smartphone, hobbyist servo motors, custom robot links, and android operating systems. Designed as a 4-axis manipulator equipped with a smartphone, the platform is able to track an object in front of the robot. A custom Android application based on OpenCV is built so that the center of the object can be detected. HSV color segmentation is used to differentiate target and environment. The control goal is make the target at the center of the of the camera reading as fast as possible. Using a PD control for individual joints with κ
_p
= 0.09 and κ
_d
= 0.01, the manipulator can be aligned with target within 7 to 12 [s], depending on the initial and target positions. The platform can be used for both teaching and research in robotics.