SLAM TOOLBOX FOR MATLAB

LATEST NEWS

Upgrade 2015/08/05: Added Graph-SLAM using key-frames and non-linear optimization.

Older upgrades and news

Upgrade 2012/04/22: Added support for Omni-directional cameras for ahmPnt and eucPnt points.
From 2011/09/03 to 2011/09/08: a bug in the package released between these 5 days caused the toolbox to completely fail. Download the current version below.
2010/09/04: BUG FIX: Corrected bug in code for IDP. If you just want the bug fix, click here. Toolbox versions after 2011/09/08 are already fixed.

See the TROUBLESHOOTING section below for some fixes related to old Matlab versions.

INTRODUCTION

We present two toolboxes sharing a number of features, but using two radically different approaches to SLAM: EKF, and nonlinear optimization via graphical models. They are provided for free under the GPL license (please read the file COPYING and make sure you agree in the terms and conditions before using them).

EKF-SLAM: The first toolbox performs 6DOF SLAM using the classical EKF implementation. It is conceived as an "active-search" SLAM. With points:

and with lines:

Graph-SLAM: The second toolbox substitutes the EKF by a non-linear optimizer based on factor graphs and matrix factorization. Currently, QR, Cholesky, and Schur factorizations are implemented. It works with points:

USING THE TOOLBOX

Users employing the toolbox for scientific research should cite in their scientific communications one of the papers of the authors (especially SOLA-ETAL-IJCV-11, SOLA-ETAL-TRO-08) appearing in the References section of the documentation, and also acknowledging the use of this toolbox.

DOWNLOAD AND INSTALLATION

Read the README.txt file in the GitHub repository for installation instructions.
Download the 6DOF SLAM toolbox for Matlab, using one of the GitHub facilities to do so:

git clone, if you have git in your machine
zip download, if you do not have git.

Read the pdf doc to have an idea of the toolbox, focused on EKF-SLAM implementation.
Read this other doc for a complete tutorial on GraphSLAM. See this file for the objects organization in graphSLAM.

ADDITIONAL INFORMATION

Please email me with feedback, I will appreciate.
Features - EKF:

Full 6DOF.
Points and lines, with many different parametrizations. Consult the IJCV'11 paper on landmark parametrization.
3D graphics - highly customizable.
Works with any number of robots and sensors.
Monocular and stereo systems are treated alike.
Supports extrinsic self-calibration of multi-camera rigs as in our TRO'08 paper.

The toolbox supports undelayed initialization of the following landmark types:

ahmPnt : Anchored homogeneous points. See video. This parametrization performs equivalently to Inverse-depth points in filter consistency.

idpPnt : Inverse-Depth points, with conversion to Euclidean points (eucPnt) after confirming linearity. This follows Civera's TRO-08 paper on Inverse-depth parametrization. See video. Also called Anchored Modified-polar points (AMPP) in SOLA-IJCV'11.

hmgPnt : Homogeneous points. See video.

fhmPnt : Framed Homogeneous points. We follow a work by Simone Ceriani et. al., 2011.

plkLin : Plucker lines as explained in our IROS-09 paper. See video.

aplLin : Anchored Plucker lines. See video.

idpLin : Inverse-depth points lines. See video. Also called Anchored Modified-polar points lines (AMPPL) in SOLA-IJCV'11.

hmgLin : Homogeneous-points lines. See video.

ahmLin : Anchored homogeneous-points lines. See video.

See this video that compares idpPnt against ahmPnt and hmgPnt points, showing the superior consistency of idpPnt and ahmPnt because of over-estimation of the covariance in hmgPnt. (note: in the video, hmgPnt is labeled HP, ahmPnt is AHP, and idpPnt is AMPP.)

See this video with all line parametrizations running in parallel.

Features - GraphSLAM

Full 6DOF.
3D graphics - highly customizable.
Verified to work with 1 robot, 1 image+depth sensor, Euclidean points. Upgrades for greater generalization may come up in the future.
Cholesky, QR and Schur complement matrix factorizations for the solver.

BUG FIXES

BUG (fixed in toolbox releases after 2011/09/08): The Jacobians of the idpPnt initialization function were incorrect. This derived in a poor performance of idpPnt points. Now, idpPnt and ahmPnt show equivalent performance.
FIX: follow these simple steps:

Download this file. Unzip it.
Copy files idp2ahp.m and ahp2idp.m to toolbox folder %SLAMTB/Points/
Copy file retroProjIdpPntFromPinHoleOnRob.m to toolbox folder %SLAMTB/Observations/
Delete the obsolete bugged file %SLAMTB/Points/idpS2idpW.m from the toolbox.

TROUBLESHOOTING

Expression or statement is incorrect--possibly unbalanced (, {, or [.
In version 2009b and after, MATLAB introduced the tilde return operator '~' to ignore unwanted return parameters and avoid creating variables in memory that are not going to be used.
If you happen to be running this toolbox on older versions of Matlab, you will encounter the following kind of error:

" Expression or statement is incorrect--possibly unbalanced (, {, or [."

Your best option is to do the following:

Run slamtb as many times as you have errors of this kind.
Each time, navigate to the error line, and substitute the tilde '~' by a dummy variable name such as tmp, unused, dummy, or whatever you prefer.