Home

Our final project for Computational Robotics consisted of developing SLAM and planning algorithms in tandem. The initial goal was to merge these two implementations and provide a continuously updating map to a Neato robot attempting to reach defined waypoints in an unknown environment. Upon visitation of each waypoint, the robot would return to its starting point in the most efficient way.

Our general approach was to use built-in packages and to incrementally fill in the functionality ourselves. In the end, the SLAM portion of the project focused on stitching maps together, while the planning focused on optimizing a Markov Decision Process (MDP) based on this paper. Because the paths of exploration diverged from the original goal, we did not integrate them and focused on each part separately.

Please see the blog posts #1 (mapping, planning) and 2 (mapping, planning) for a detailed description of the development process and blog posts #3 (mapping, planning) for our results.