Revision history [back]

Among the open source implementations, RGB-D SLAM, DVO ( http://vision.in.tum.de/data/software/dvo ), MRS-MAP (http://code.google.com/p/mrsmap/) show the state-of-the-art performance on RGB-D cameras. They all rely extensively on the RGB information. CCNY_RGBD is also promising too.

I am not aware if your device (Fotonic) provides the RGB data or not. If not, then you can try calibrating a RGB camera with Fotonic sensor. And then run any the above mentioned methods by renaming the appropriate topics.

But one of the drawback is that they accumulate error over time and there is need to revisit the place to improve the accuracy.( loop closure ) .

I am not sure of what is your target environment and the trajectory you might follow. I would suggest you to first try these open source implementations on Kinect (as setting up the system would be easy and fast), look for the accuracy of the odometry estimates. Based on this you can decide if you would like to use these existing implementations for your task.

Good luck and report what your observations are and your results!