In this paper is presented the design of a tracking and trayectory generation robotic system in a testing environment Robocup (small size category) by the prediction of movements using Kalman filters and the optimal trajectory generation by implementing B-spline curves. In the different results obtained it was found that the Kalman filter facilitates the prediction of robotic agents with small processing times and curves B-splines by optimizing the control points allowed to generate smooth curves for obstacle avoidance.