Introduction

Fovis is a visual odometry library that estimates the 3D motion of a camera using a source of depth information for each pixel. Its original implementation is described in the paper:

Visual Odometry and Mapping for Autonomous Flight Using an RGB-D Camera.
Albert S. Huang, Abraham Bachrach, Peter Henry, Michael Krainin, Daniel Maturana, Dieter Fox, and Nicholas Roy.
Int. Symposium on Robotics Research (ISRR), Flagstaff, Arizona, USA, Aug. 2011 (PDF).

Quick links

Build requirements

Fovis is intended to be relatively portable. There is requirements for building and using the software are:

Fovis was developed on Ubuntu, but may work with other platforms.

Build instructions

For system-wide installation:

$ mkdir build
$ cd build
$ cmake ..
$ make
$ sudo make install
$ sudo ldconfig

This usually installs Fovis to /usr/local or something like that.

For use in the source directory:

$ mkdir build
$ make

Intel SSE

By default, Fovis is configured to use Intel SSE2 and SSE3 instructions. You can disable this option (e.g., if you're targeting a platform without those instructions) by using the CMake option USE_SSE:

$ cd build
$ cmake .. -DUSE_SSE=OFF
$ make

Building documentation

Fovis is documented using Doxygen. To build the documentation:

$ cd doc
$ doxygen

Following that, open up doc/html/index.html in your web browser.

Usage requirements

For portability reasons, the actual library itself is sensor agnostic and provides no data acquisition capabilities. To use fovis, your program must acquire data on its own and pass it through to the fovis API. Some examples are provided with the source code.

Effective use of fovis for visual odometry requires the following:

A source of 8-bit grayscale camera images.
A camera calibration for the images that provides an accurate mapping between image pixel coordinates (u, v) and 3D rays (X, Y, Z) in the camera's Cartesian coordinate frame.
A depth source for each image. A depth source must be able to provide a metric depth estimate for as many pixels in the camera image as possible.

Fovis provides built-in support for the following types of depth sources:

An RGB-D camera such as the Microsoft Kinect.
Calibrated stereo cameras.

You can also create your own depth sources using the Fovis API and adapt it to other sensor types.

Getting started

The best way to get started is to look through the examples provided with the source code in the examples/ directory.

Next, look through the Fovis C++ API. The primary class of interest is fovis::VisualOdometry.

License

fovis is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.

fovis is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.

A copy of the GNU General Public License is provided with the fovis source code.