Predicting The Trajectory Of A Flying Body Based On Weighted Nearest Neighbors

The task of predicting the kinematical trajectory of a flying body is considered within the new approach for transportation of objects by throwing and catching. The basic method for prediction, k-nearest neighbors, is adapted for improving the speed of processing and more efficient use of the database. Two weighted neigbours are used to predict future values of object coordinates. Two modifications for the algorithm are proposed: division of sampling into a number of clusters and transforming 3D coordinates of the objects into a 2D coordinate system. The approach is evaluated using the simulation of ballistic motion.