StrataSLAM

StrataSLAM is an open-source 3D simultaneous localization and mapping (SLAM) system for ROS 2, developed by the Intelligent Robotics Lab at Universidad Rey Juan Carlos (URJC) under the EasyNavigation organization. Its key idea is to keep one point cloud per floor (a submap) plus one global map, without floors collapsing into each other.

The system solves common SLAM problems when a robot changes floor (elevator, stairs, ramp, drone): losing height, closing false loops between identical-looking floors, and resetting the reference frame in a dark stairwell. Online, no code path accepts a loop closure between two floors; floor transitions are detected and classified, and floors are fused offline — using transition constraints, the barometer, structural anchors and a weak floor-height prior, in a gravity-aligned 4-degree-of-freedom group SE(2)×ℝ.

StrataSLAM supports an RGB-D camera, 3D LiDAR or both, with IMU, barometer and external odometry optional; everything is configured with a single YAML profile, and the 'strataslam doctor' command checks the setup before mapping. The project (version 0.2.0) is released under the Apache-2.0 license and depends on, among others, the GTSAM and small_gicp libraries; it builds for ROS 2 Jazzy.