Through this project, my team explored the effectiveness of DeepMapping, a tool created by the AI4CE lab at NYU for point cloud registration using Deep Learning. My primary focus was to contrast the effectiveness when used on synthetic environments versus real 3D datasets, sparse versus cluttered scenes, indoor versus outdoor, large versus small trajectories, high versus low frame overlap. We found DeepMapping to be highly robust to local minima, even with highly misaligned point clouds. We found it to be weaker when attempting to register highly dense scenes.