6D Object Pose Estimation using RGBD Data and Fast-ICP

Problem Statement

Given RGBD Images and a model of the object, segment, reconstruct and estimate 6DOF pose of the object.

Usage Guide:

1. Change the flags in Wrapper.m as needed.
2. By default, the code runs uses Liquid Container dataset (change folder name if needed).
3. By default, the code uses sub-sampling, truncation and point to plane method, does not display all iteration and save video.

Report:

You can find the report here.

Sample Outputs:

References:

1. Arun, K. Somani, Thomas S. Huang, and Steven D. Blostein, Leastsquares fitting of two 3-D point sets, IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 5, pp. 698–700, 1987.
2. Low, Kok-Lim, Linear least-squares optimization for point-to-plane icp surface registration, University of North Carolina Chapel Hill, Vol. 4, 2004.
3. Rusinkiewicz, Szymon, and Marc Levoy, Efficient variants of the ICP algorithm, Proceedings of 3-D Digital Imaging and Modeling, 2001.