Thank You, Suryansh Kumar.

Basically, EKF-SLAM and FastSLAM are two filtering versions of SLAM, i.e., they estimate the state from a previous state and new observations. There is another bunch of algorithms that estimates the state based on all previous measurements, thus solving what is sometimes called the "full" SLAM problem. A popular member of that category would be GraphSLAM.

Thank You, Michael Hardegger for pointing out a full-SLAM algorithm solver.

Alternative approach is random finite set SLAM, e.g. the PHD SLAM (by Mullane, Adams, Vo). PHD SLAM is similar to FastSLAM by using Rao-Blackwellization. Recent results by Leung show superior performance compared to FastSLAM. Implementation is available at https://github.com/kykleung/RFS-SLAM

Hello, when we talk about SLAM algorithms there is lot of point of view: We can talk about the filter used to fuse data as you said EKF and FAST SLAM which are two examples of Probabilistic SLAM, we can also talk about the UKF and Hinfinity SLAM. The second point of view of the SLAM algorithm is the sensor used for observation, then we can talk about Laser SLAM, Visual SLAM (with mono or stereo camera), and now with the apparition of RGBD cameras (tof camera and Kinect, ...) we can talk about the RGBD-SLAM. The open doors in SLAM research area currently are in Long Life SLAM (Is A SLAM for long term, day and night and very large area),

Thank you

I imagine any change in structure of the SLAM would create new kind of SLAM and anybody could create his/her own SLAM and it is based on your innovation. Still, the following SLAM are more common:

• EKF SLAM -  which you mentioned.
• Fast SLAM - which you mentioned

Moreover:

• UKF SLAM, CEKF SLAM, COP SLAM, DP SLAM... - https://www.openslam.org
• Mono SLAM- http://perso.ensta-paristech.fr/~filliat/Courses/2011_projets_C10-2/BRUNEAU_DUBRAY_MURGUET/monoSLAM_bruneau_dubray_murguet_en.html
• Linear SLAM- http://www.cc.gatech.edu/~dellaert/html/linear_slam.html

Still there are plenty of SLAM procedures that are not mentioned here. I insist you to visit  https://www.openslam.org. I hope it works for you

https://www.openslam.org

http://perso.ensta-paristech.fr/~filliat/Courses/2011_projets_C10-2/BRUNEAU_DUBRAY_MURGUET/monoSLAM_bruneau_dubray_murguet_en.html

http://www.cc.gatech.edu/~dellaert/html/linear_slam.html

So far, I think the most extensive SLAM algorithm is the PHD SLAM with MOT capability. It seems to be two different things. However, they are two in one. A good SLAM application must work in cluttered, unstructured, and dynamic environment, and you cannot avoid to encounter the moving objects. How to distinguish the static landmarks with the interfering moving objects is the austere task each SLAM designer has to face. How to utilize this MOT (moving object tracking) feature to help SLAM is yet a good wide opened research topic. You may Google this topic --- SLAM + MOT

Good luck