Reliable navigation-path extraction system for an autonomous mobile vehicle

This paper describes the algorithms for path recognition and obstacle detection for an autonomous mobile vehicle. The Path Extraction Algorithm (PEA) recognizes drivable paths on the road by image processing. The Environment Extraction Algorithm (EEA) provides the spacial pose of the mobile vehicle and obstacle detection by the data processing of the 2D laser scanner. The Pattern Classification Algorithm (PCA), a machine learning process based on the supervised method, enables to classify road patterns by the use of trained Artificial Neural Networks. The Navigation-Path Extraction Algorithm (NPEA) is comprised of these three sub-systems. Our test results demonstrate that the Navigation-Path Extraction System (NPES) is reliable and robust to be implementable on a mobile vehicle to achieve self-driving.