視覺感測與慣性量測融合於同時定位與建圖

Fusion of Inertial Measurement and Visual Sensor for Simultaneous Localization and Mapping

吳皇毅(Huang-Yi Wu)

指導教授 ： 孫崇訓

淡江大學/工學院/機械與機電工程學系碩士班/碩士(2016年)

https://doi.org/10.6846/TKU.2016.00108

摘要

本論文探討慣性量測元件輔助單眼視覺同時定位與建圖的議題。本論文使用加速強健特徵(Speed-Up Robust Feature, SURF)演算法偵測並描述特徵點，再以反深度參數化方法描述地標點位置，並利用擴張型卡爾曼濾波器(extended Kalman filter, EKF)估測攝影機及地標點狀態。使用慣性量測元件的位移可以推測出尺度基準，並能以此初始化單眼視覺估測，並透過實驗證實慣性量測元件能使單眼視覺同時定位與建圖成功初始化。

關鍵字

單眼攝影機 ； 慣性量測元件 ； 視覺式同時定位與建圖 ； 單眼視覺估測初始化

並列摘要

This study investigates the issues of inertial measurement unit (IMU) assisted monocular simultaneous localization and mapping (SLAM). The speeded-up robust features (SURF) algorithm is used for interest point detection and description. The positions of environment landmarks are represented by inverse depth parameterization method. The positions of camera and landmarks can be estimated by using the extended Kalman filter (EKF). The map scale for monocular SLAM initialization can be estimated by the displacement of IMU. The experiment results demonstrate that the IMU successfully initialize monocular SLAM.

並列關鍵字

Monocular vision ； Inertial measurement unit ； Simultaneous Localization and mapping(SLAM) ； Monocular SLAM initialization.

參考文獻

[4] 陳庭瑋，"飛行機器人單視覺式定位與建圖之影像深度初始化與模糊資料關聯"，淡江大學機械與機電工程學系碩士班碩士論文，2014

[11] 邱明璋，"基於極線限制條件之單眼視覺式移動物體偵測與追蹤"，淡江大學機械與機電工程學系碩士班碩士論文，2011.

[2] J. Civera, A.J. Davison and J.M.M. Montiel, “Inverse Depth Parametrization for Monocular SLAM,” IEEE Trans. Robot., vol.24, no.5, pp.932-945, 2008.

[5] H. Bay, A. Ess, T. Tuytelaars, and L. Van Gool,“Speeded-up robust features (SURF),”Computer Vision and Image Understanding (CVIU), vol.110, pp.346-359, 2008.

[6] P. Viola and M. Jones, “Rapid object detection using a boosted cascade of simple features,” In Proc. IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp.511-518, 2001.