運用移動式與固定式攝影機於大範圍環境之機器人定位及建圖系統

Localization and Mapping for Mobile Robot Navigation Employing Onboard and Surveillance Cameras in Large-Scale Environments

SamuelSimon(Samuel Simon)

指導教授 ： 張文中

國立臺北科技大學/電資學院/電機工程系研究所/碩士(2012年)

摘要

本論文提出一種運用於大範圍未知監視環境之高精度機器人導航與控制系統,搭配地面及飛行機器人以視覺伺服方式達成同步定位與建圖。在此系統中, 天花板上將安裝多個攝影機用於監控環境, 利用這些攝影機, 整合EKF-SLAM 可進行額外的校正,此方法將會使機器人定位和環境建圖更加精準,然後,將建圖之結果從三維轉換至二維以用於軌跡規劃,並使用Bezier曲線產生機器人的規劃軌跡,讓機器人順利行進。為了擴大監控攝影機的覆蓋範圍,我們採用裝有攝影機的飛行機器人,此飛行機器人可在空中穩定地飛行, 拍攝監控攝影機未能監控的區域, 增加監控攝影機的覆蓋範圍。此系統已於一實驗室環境實驗驗證具可行性及有效性, 並預期此架構與方法將可進一步擴展機器人於未知環境中之應用性。

關鍵字

3D VSLAM ； 移動式機器人導航 ； 飛行機器人 ； 監控攝影機

並列摘要

This thesis presents a high-precision simultaneous localization and mapping of mobile robot in unknown large-scale surveillance environments. Utilizing ceiling-mounted PTZ surveillance cameras, additional correction step is integrated into the EKF-SLAM system to ensure high-precision localization and mapping. The 3D sparse map is transformed into 3D and 2D navigation maps. Based on the navigation maps, robot trajectory can be generated using Bezier spline curves to allow smooth movements. Effective navigation and control approach for the mobile robot is proposed to allow mobile robot to follow the generated trajectory. In order to extend the coverage area of surveillance cameras, aerial robots equipped with cameras is employed to cover un-surveillance area. Control law for the aerial robots is also applied allowing them to stabilize themselves and move around in un-surveillance area. Experiments were performed in laboratory environments to validate the feasibility and effectiveness of the proposed system. This system can be further developed to potential robotics applications in unknown environments.

並列關鍵字

3D VSLAM ； mobile robot navigation ； quadcopter ； surveillance camera

參考文獻

[1] M. Montemerlo and S. Thrun. Simultaneous localization and mapping with unknown data association using fastslam. In Robotics and Automation. Proceedings. IEEE International Conference on, volume 2, pages 19851981, September 2003.

[2] J. Civera, A. J. Davison, and J. M. M. Montiel. Inverse depth parametrization for monocular slam. IEEE Transactions on Robotic, 24(5):932945, Oct. 2008.

[3] Chieh-Chih Wang and C. Thorpe. Simultaneous localization and mapping with detection and tracking of moving objects. In Robotics and Automation. Proceedings. IEEE International Conference on, volume 3, pages 29182924, September 2002.

[4] T. Suzuki, Y. Amano, and T. Hashizume. Development of a sift based monocular ekf-slam algorithm for a small unmanned aerial vehicle. In SICE Annual Conference (SICE), 2011 Proceedings of, pages 1656 1659, sept. 2011.

[5] Christopher Mei, Eric Sommerlade, Gabe Sibley, Paul M. Newman, and Ian D. Reid. Hidden view synthesis using real-time visual slam for simplifying video surveillance analysis. In ICRA11, pages 42404245, 2011.