Robust mono-vision SLAM of mobile robots

A robust mono-vision SLAM system is designed and implemented for mobile robots working under indoor environment. Specifically, a two-direction matching method is proposed for the SIFT algorithm to obtain stable and distinct point features. The Cartesian coordinates of the feature points are extracted by fusing image and encoder information. To improve the reliability of the SLAM system, multiple feature points are then combined as a visual landmark to implement the subsequent measurement task. Based on that, the EKF technology is then adopted to implement a robust mono-vision SLAM system. As illustrated by experimental results, the SLAM strategy proposed in the paper presents such advantages as low cost, high reliability and high accuracy.