Simultaneous localization and mapping (SLAM) is a technique used by robots and autonomous vehicles to build up a map within an unknown environment (without a priori knowledge), or to update a map within a known environment (with a priori knowledge from a given map), while at the same time keeping track of their current location. The robot can move autonomously if SLAM solved perfectly. Several solutions of SLAM problem have become apparent over the past 10 years. SLAM problem is formalized, as a theoretical problem of various forms have been resolved. In addition, variety methods have been proposed which can be operated various environment, that is, indoors, outdoors, in water and in the air. In theory, it can be seen SLAM is a problem that has been solved to some extent, there are still many problems to be solved yet. In this study, to carry out SLAM with sensor data log which is collected by real robots, real-world environments, we implements the Extended Kalman Filter(EKF)-SLAM which is one of the typical SLAM algorithms and Thin Junction Tree Filter(TJTF)-SLAM which is one of the latest algorithms. In addition, by performing the EKF and TJTF SLAM in Large-Scale, I was able to perform a comparison and verification of performance and practicality predictability in speed, the amount of computation and memory.

Paper

• LINK (Korean)