PROJECT TITLE :

An SfM Algorithm With Good Convergence That Addresses Outliers for Realizing Mono-SLAM

ABSTRACT:

Monocular simultaneous localization and mapping (mono-SLAM) may be a key element of autonomous robot visual navigation. Recently, the structure from motion (SfM) approach has become an attractive suggests that of implementing mono-SLAM as a result of of its high accuracy; although, in this application, the SfM approach must be operable in real time and strong to outliers. However, as a result of of robust nonlinearity, typical SfM strategies, like bundle adjustment, should take into account multiple initial values to obtain the globally optimal result, that is time consuming. During this paper, a completely unique iterative SfM algorithm based on the object-house objective perform is proposed. To improve the method's robustness to outliers and to incorporate the knowledge obtained from different types of sensors, an approach to closely integrate the proposed SfM algorithm with extended Kalman filter-primarily based SLAM is proposed. Experimental results using both synthesized and real information are according to our theory, and verify that the most advantage of the proposed SfM algorithm is its sensible convergence. The algorithm is thus notably acceptable for realizing mono-SLAM, because when rotations can be obtained approximately using gyroscope info, the algorithm is globally convergent from any initial price.