Operative trajectory calculation for control algorithms in mobile-robotics using Superpixel segmentation

摘要

Superpixel segmentation has been developed in many ways, like an optimized recognition system's research, making faster and sturdy ways to calculate superpixels and segmenting information from the real world, by now the result has lead to implement SLIC Superpixels method, which is a low hardware requirement code, inside an embedded system, making it as fast as possible, to accomplish the need of placing it inside a robot. So why is this need so important? Because the way the robot has to execute an entrusted task is really important, and time is a primary factor of it, to improve its functions and benchmarks reactions. The project improves the SLIC Method of making superpixels with average from the original color segmented matrix; and then calculate with KMEANS method the centers and clusters from objects, but before that an attached method based in a section of set labels an average and a bilateral filter already applied this. Finally with a fuzzy positioner it can be found the optimal route of its movement.