wtf is bievr-lio?

BIEVR-LIO is a research system that helps mobile robots figure out where they are and how they are moving using a spinning laser sensor called a LiDAR combined with an inertial sensor that measures motion, like the one in a phone. This general task is called odometry, and it is one of the core building blocks robots need to move around safely. The project's contribution is aimed at a specific hard case: environments with very little distinctive shape to them, such as long straight tunnels or featureless corridors, where many existing systems lose track of the robot's position because there is not enough geometric detail to match one scan against the next. BIEVR-LIO addresses this by storing the environment as a map of small height images arranged in a grid of cubes called voxels, which captures subtle bumps and surface variation that flatter representations miss. It also focuses its comparison work on the parts of the environment that actually carry useful shape information, which both improves accuracy in these hard cases and reduces computing cost compared to comparing everything everywhere. The core system does not depend on any robotics framework and can be used on its own, with separate add on packages provided for both major versions of ROS, the common robotics software framework, labeled ROS1 and ROS2. Setup instructions and Docker images are provided for both versions so you can try the system without manually installing every dependency. The project also depends on two established math and optimization libraries, Eigen and Ceres, and can optionally support additional LiDAR hardware from a manufacturer called Livox. This is research code accompanying a published paper, aimed at robotics engineers and academics working on autonomous navigation, not a general purpose consumer tool. It is released under the BSD 3-Clause license.