원문정보
초록
영어
In this paper, we present a vision based robust pose estimation system in different observed situations for a quadrotor in outdoor environments. This system could provide us with approximate ground truth of pose estimation for an outdoor quadrotor, while most of existing vision based systems perform indoors. We only use the own features of the quadrotor, while most existing systems modify the architecture of the quadrotor or put additional components such as colored markers on it. We propose the novel robust pose estimation algorithms for different observed situations. With good observed results, we get all of the four rotors and calculate the pose. But when fewer than four rotors are observed, all of existing external vision based systems for the quadrotor do not mention this and could not get right results. By combining inertial measurement unit (IMU) data, our robust pose estimation system has solved these problems and obtained accurate results of pose estimation. We demonstrate in real experiments that our pose estimation system for the quadrotor could perform accurately and robustly in real time.
목차
1. Introduction
1.1. Related Work
1.2. Overview of our Work
2. Hardware and Features Selection of the Quadrotor
2.1. Hardware
2.2. Why Only Use the Quadrotor’s Own Features
3. Preliminary Position
3.1. Relative Pose Relationship Between the Quadrotor and the Camera
3.2. Detection and Tracking of a Quadrotor
3.3. Preliminary Position of a Quadrotor
4. Robust Pose Estimation Algorithms
4.1. Different Observed Situations of a Quadrotor
4.2. Problem Formulation
4.3. EMRPP Algorithm for Four Rotors Observed
4.4. Three/Two Rotors Observed
5. Simulation Experiments
5.1. Four Rotors Observed
5.2. Discussion
5.3. Three/Two rotors Observed
5.4 Discussion
6. Real Experiments
6.1. Results of Real Experiments
6.2. Pose Error in Real Experiments
6.3. Discussion
7. Conclusion
Acknowledgements
References