An embarrassingly parallel hoppingwindow noise removing algorithm for lidar based robot mapping

This paper presents an embarrassingly parallel hopping window algorithm to remove noise from Lidar (Light detection and ranging) sensor for robot mapping applications. The algorithm works by analyzing the density of Lidar data inside a window which hops over the entire input sensor data. For faster execution of the algorithm, multiple window hopping is done intelligently without omitting the processing of actual data which is checked by a `reverse-hop´ mechanism. We discuss the parallel implementation of the proposed algorithm. Results show that the proposed algorithm efficiently removes noise from Lidar data, is very fast, embarrassingly parallel and its implementation is very straightforward.