Project P6:
3D SFM with an IMU

Project Goal

This project is extremely challenging: The field of "structure from motion" seeks to reconstruct 3D models from sequences of camera images. In general, this is a difficult problem, as we shall see in class. In this class we are interested in a simpler version: SFM supported with an initial measurement unit. The idea is to attach to the camera an inertial measurement unit which measures accelerations and rotational velocities.

Project Scope

A system for developing 3D models of an indoor environment.

Tasks

• implement an optical flow algorithm for measuring motion in image coordinates
• develop a Kalman filter for state estimation
• develop an interface to the IMU and integrate the IMU
• technique for tracking features in camera images
• estimate the egomotion of the camera
• estimate the location of features in the environment

Project Status

Mark Woodward (mwoodward at cs dot stanford dot edu),
Justin Tansuwan (jjtswan at stanford dot edu),
Kayur Patel (kdpatel at stanford dot edu), and
Kurt Miller (tadayuki at cs dot stanford dot edu).

Point of Contact

Sebastian Thrun

Midterm Report

not yet submited

Final Report

not yet submitted