Obstacle detection over rails using hough transform

Abstract

Autonomous systems can assist humans in the important task of safe driving. Such systems can warn people about possible risks, take actions to avoid accidents or guide the vehicle without human supervision. Whether in cars or trains or ships the artificial vision algorithms offer an alternative for the design and implementation of autonomous driving systems. In railway scenarios cameras in front of the train can assist drivers with the identification of obstacles or strange objects on the rails. Multiple factors add huge complexity to this task. The changing conditions create a scene where background is hard to detect, lighting varies and process speed must be fast. This article describes a first approximation to the problem where using the Hough transform, the rails and area of interest are detected. On this area a systematic search is done for finding and delimiting possible obstacles. Our system accomplished a real time performance when employed in the analysis of recorded videos from the driver perspective. Using digital added obstacles our algorithm detects mostly all of them and warns if the objects over the rail can create a danger to the safety travel of the train.