Investigation of Robust SLAM and Multi-Robot Exploration
| Autor: | Viet-CuongPham, 范越強 |
|---|---|
| Rok vydání: | 2013 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 101 As a result of developing advanced technology and automation, robots are becoming more human-like. They are employed in many applications for improving our daily tasks such as exploration of unknown or distant territories, search and rescue, surveillance, reconnaissance, examination of hazardous areas, assistance in hospital, intelligent home, farm work, and entertainment. This dissertation addresses two fundamental problems in robotics: the simultaneous localization and mapping (SLAM) problem and the exploration problem. SLAM is an important and challenging task for the operation of autonomous mobile robots in which both the pose of the robot and features of the environment need to be estimated at the same time. In particular, it is desirable to achieve robustness and efficiency in the SLAM implementation. Robots in unknown environment are likely to be subject to modeling errors which cannot be easily characterized in terms of statistical properties. To mitigate the effect of uncertainties and disturbances, robust filters such as the filter can be employed. However, robust filters are complex to implement, demanding a significant amount of computational resources. Therefore, we propose a reduced order filter to solve the robust SLAM problem in which robot dynamics are subject to uncertainties and measurements are subject to bounded-but-unknown disturbances. To achieve both efficiency and robustness, techniques such as state partition and relative landmark representation are employed. Simulations reveal that results obtained from the proposed reduced order filter, which has a lower computational complexity, closely approximate those from the full order filter. Moreover, the reduced order filter is more robust than EKFs and FastSLAM. Although the problems of mobile robot exploration have been studied for a long time, there are still many practical issues to overcome. For example, in classical exploration approaches, robot poses are assumed to be known or estimated by SLAM. Ordinary SLAM approaches only handle the obtained sensor data and lack a mechanism to control the robot motion to perform the SLAM and exploration tasks more efficiently. It is well known that the path control strategy can have a substantial impact on the quality of the resulting map as well as the overall efficiency. As a result, integrated exploration methods have been proposed recently to simultaneously consider localization, mapping, and motion control. However, most integrated exploration methods only deal with the single robot case. Therefore, we propose an integrated exploration approach for multiple robots. In particular, robots are controlled to minimize the overall exploration time and to maintain a sufficient level of accuracy of the robot pose and map estimates. This approach is further enhanced to globally disperse robots in the environment, to balance the exploration performance and localization quality as well as t o deal with limited communication. Simulation results show that the proposed approaches outperform existing ones. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|