Attitude-Motion Estimation of Tumbling Objects Using Visible Region Information
- Paper number
IAC-07-C1.8.05
- Author
Mr. Hideyuki Tanaka, University of Tokyo, Japan
- Coauthor
Dr. Takehisa Yairi, University of Tokyo, Japan
- Coauthor
Dr. Kazuo Machida, University of Tokyo, Japan
- Year
2007
- Abstract
The attitude-motion estimation of tumbling objects is a fundamental problem for space missions. We proposed a novel approach for solving this problem using only qualitative observation data. Our method does not require any precise image processing or feature tracking. Instead, we use very rough qualitative information, that is, “Which aspects of the target are visible?”\\ \quad In our system, the observation data at a given time becomes a binary vector that consists of the values of 1 (visible) or 0 (invisible) for each aspect of the target. We introduce radio frequency identification (RFID) technology to acquire such qualitative information. The target is assumed to have multiple RFID tags on its surface. Since each RFID tag has a certain communication area, only the RFID tag facing the observer returns its ID. The returned ID corresponds to the visible aspect. Accordingly, the observation model becomes strongly nonlinear and non-Gaussian.\\ \quad In order to solve the attitude-motion estimation problem, we apply a particle filter, which is one of the sequential Monte Carlo methods applicable to any state-space model and which generalizes the conventional Kalman filter.\\ We conducted simulation experiments under several conditions. The results showed that our method enables the accurate estimation of the attitude motion of a tumbling object and is robust to false-negative errors such as occlusions. We also checked the applicability of our estimation framework to static parameters such as the moment of inertia, as well as to state variables. The estimation accuracy depends on the number of RFID tags and the configuration of RFID readers. We found that efficient observation and effective use of information are made possible by their appropriate configuration. If the target has 14 RFID tags and the observer has three RFID readers, the estimation accuracy of six degree is possible. This is sufficient accuracy for rough estimation in the initial step of satellite capture.\\ \quad Our method is not limited to only use of RFID technology, i.e., it is possible to use image data or methods of acquiring the information of visible aspects. Therefore the method can be extended to applications using other hardware configurations.
- Abstract document
- Manuscript document
IAC-07-C1.8.05.pdf (🔒 authorized access only).
To get the manuscript, please contact IAF Secretariat.