In this paper, we propose a modified ORB-SLAM (Oriented FAST and Rotated BRIEF Simultaneous Localization And Mapping) for precise indoor navigation of a mobile robot. The exact posture and position estimation by the ORB-SLAM is not possible all the times for the indoor navigation of a mobile robot when there are not enough features in the environment. To overcome this shortcoming, additional IMU (Inertial Measurement Unit) and encoder sensors were installed and utilized to calibrate the ORB-SLAM. By fusing the global information acquired by the SLAM and the dynamic local location information of the IMU and the encoder sensors, the mobile robot can be obtained the precise navigation information in the indoor environment with few feature points. The superiority of the modified ORB-SLAM was verified to compared with the conventional algorithm by the real experiments of a mobile robot navigation in a corridor environment.

Abstract
1. 서 론
2. 모바일로봇 위치 및 자세추정
    2.1 IMU 기반 자세 추정
    2.2 엔코더 기반 위치 추정
3. Indoor ORB-SLAM
    3.1 ORB-SLAM
    3.2 Modified ORB-SLAM
4. 실험환경 및 실험결과
    4.1 실험환경
    4.2 실험결과
5. 결 론
References

• 옥용진(Pusan National University) | Yongjin Ock
• 강호선(Pusan National University) | Hosun Kang
• 이장명(Dept of Electronics Engineering, Pusan National University) | Jangmyung Lee Corresponding author