A Robust Lane Recognition Technique for Vision-Based Navigation with a Multiple Clue-Based Filtration Algorithm

A Robust Lane Recognition Technique for Vision-Based Navigation with a Multiple Clue-Based Filtration Algorithm

This paper proposes a novel multiple clue-based filtration algorithm (MCFA), which is developed to detect lane markings on roads using camera vision images for autonomous mobile robot navigation. The main goal of the algorithm is the robust estimation of the relative position and angle of the lane in the image by using multiple clues based on different characteristics of the lane. In particular, robustness against environmental changes is enhanced greatly since a dynamic model of the lane, be-sides static features of the lane such as color, intensity, etc., is incorporated for reliable estimation. The efficiency of the algorithm is verified through mobile robot experiments under various extreme illumi-nation conditions in outdoor environments. The increased robustness performance enables reliable closed-loop control of a mobile robot that operates in a variety of navigation-related missions.

This paper proposes a novel multiple clue-based filtration algorithm (MCFA), which is developed to detect lane markings on roads using camera vision images for autonomous mobile robot navigation. The main goal of the algorithm is the robust estimation of the relative position and angle of the lane in the image by using multiple clues based on different characteristics of the lane. In particular, robustness against environmental changes is enhanced greatly since a dynamic model of the lane, be-sides static features of the lane such as color, intensity, etc., is incorporated for reliable estimation. The efficiency of the algorithm is verified through mobile robot experiments under various extreme illumi-nation conditions in outdoor environments. The increased robustness performance enables reliable closed-loop control of a mobile robot that operates in a variety of navigation-related missions.