This paper proposes a method to reduce the pose error and to solve the dead reckoning issue which occurs when the mobile robot with continuous-tracks travels in the unstructured environments. When the continuous-track type mobile platform travels on terrain such as sand, gravel, stairs and etc., slippage occurs and thus the driving state of the mobile robot cannot be recognized normally. To compensate for this pose error, the proposed method utilizes optical flow estimation detected by camera. This method is tested through experiment. Finally, This method reduces the pose error detected on inertia measurement unit within some limit, while the pose error of without compensation increases without limit during robot move.

ABSTRACT
1. 서 론
2. 비전 기반 모바일 로봇 상태 추정 시스템
    2.1 무한궤도 모바일 로봇
    2.2 옵티컬 플로우 기반 모바일 로봇 상태 추정 알고리즘
3. 실험 방법 및 결과
    3.1 요철이 있는 평지에서의 실험 환경
    3.2 실험 결과
4. 결 론
References

• 김민규(School of Mechanical Engineering, Kyungpook National University) | Min-Gyu Kim
• 심유섭(School of Mechanical Engineering, Kyungpook National University) | Yu-Seop Sim
• 이상룡(School of Mechanical Engineering, Kyungpook National University) | Sang-Ryong Lee Corresponding Author
• 이학(School of Mechanical Engineering, Kyungpook National University) | Hak Yi Corresponding Author