Roboshark : a gantry pool player robot

Robots are coming to conquer new domains of everyday life. They are used in fields such as education, entertainment and housekeeping more than ever. Roboshark, a gantry robot capable of playing pool games, can be used for both training amateur pool players and entertaining amateur or professional ones. Roboshark belongs to gantry robots family. It is located directly above the pool table – a cm cm 45 90 × model in our project and has four degrees of freedom. The first three are prismatic, correspondent to movement along the X, Y and Z axis in a 3 dimensional Cartesian space, and the last is revolutional in order to rotate the kicker arm. For gathering the environmental information including the location and the color of the balls, a camera is mounted over the table. Having the information sent by the camera, a frame grabber captures each frame as a still image in memory. Due to the importance of ball colors, a color blob detection based algorithm is used. It contains some noise removing and merging filters, which are applied on the found objects. Roboshark uses crisp mathematics to compute the strength and the angle of the shot (named "shot vector" in this paper) and a fuzzy approach to select the best shot. Only direct shots are taken into account at this step. We will show that with 16 fuzzy rules, in most cases the ball selected by the fuzzy target ball selection module is the same as the one which an expert will choose in the same situation