OpenRatSLAM, is an open-source C++ (ROS / OpenCV) and MATLAB version of RatSLAM. RatSLAM is a robot navigation and SLAM system based on the neural processes underlying navigation in the rodent brain and hence is based on computational models of the hippocampus. The approach uses a combination of appearance based visual scene matching, competitive attractor networks, and a semi-metric topological map representation. The approach has been proven in a real time 40 hour robot delivery task, mapping an entire Australian suburb and on Oxford’s New College dataset. Notably, RatSLAM works well on images obtained from inexpensive cameras. The RatSLAM system contrasts many of the other SLAM approaches that involve expensive precision laser sensors and occupancy grids.
For this project Scott and myself translated the original implementation, and, in doing so, reduced the number of functions and lines of code by an order of magnitude. OpenRatSLAM is also more portable that the original due to the ROS wrapper.
The following video shows the OpenRatSLAM system in action. OpenRatSLAM is available from https://github.com/davidmball/ratslam under the GPL.
D. Ball, S. Heath, J. Wiles, G. Wyeth, P. Corke, M. Milford (2013) OpenRatSLAM: an open source brain-based SLAM system. Autonomous Robots 34 (3), 149-176.