STEREOVISION DATA PROCESSING WITH 3D DENSITY MAPS FOR AGRICULTURAL VEHICLES

摘要

Along with the advent of compact stereo cameras, stereovision information has become an essential part of many perception units of robots and autonomous vehicles. The ease of capturing stereo data clouds is in contrast with the difficulty of processing them reliably in real-time operations. This research developed a methodology to deal with stereo point clouds by extracting useful information for an autonomous vehicle to travel between two points avoiding the obstacles interfering with its trajectory. The novel conceptof 3D density and its embodiment through density grids was introduced and verified. Experimental results demonstrated that the application of 3D density and density maps is a suitable technique to cope with complex stereo data, providing useful information to aid in the path-planning routine of a mobile platform