Vision Processing for Position Based Visual Servo Control of Spherical Objects

There is an need for an autonomous system that can organise spherical objects and reposition these objects with six degrees of control. A mechatronic system is intended to partially fulfil the needs of this system by providing the means to position an object's three degrees of rotational freedom. In this paper we present the development of a vision motion tracking algorithm. This algorithm is then to be used for creation of motion models and a visual servo system. The vision algorithm is based around intra-frame blob tracking to determine the motion of a manipulated object. The common axis and magnitude of the motion is determined by tracking randomly placed dots on the object's surface. An intersection of planes approach is utilised to generate a best fit axis of rotation. The performance of the dot tracking and axis rotation algorithms are investigated with simulations by varying variables test against other methodologies.