manif

Why we included this project

Rotations and rigid transforms don't compose like plain vectors, which is usually where pose estimation code starts to hurt. manif packages the relevant Lie groups, SO(3), SE(3), and the extended pose variants, as small C++11 types built on Eigen that slot into existing codebases without a build-system overhaul, and it computes Jacobians analytically for the common operations. That last part saves real time when you are handwriting an extended Kalman filter or an optimization residual. Python wrappers mean you can prototype in Python and keep the same operations for the deployed C++ pipeline. For robotics teams doing SLAM, odometry, or inertial state estimation, it's a practical way to stop re-deriving the group math and get on with the estimation problem.