These distributions are the base representation for the robot and the map elements in many localization and SLAM algorithms.
They include unimodal Gaussians, sum of Gaussians, sets of particles, and grid representations, methods to convert between them and to draw an arbitrary number of samples from any kind of distribution. Check out the classes:
- poses::CPointPDF for 3D points.
- poses::CPosePDF for 2D + heading poses.
- poses::CPose3DPDF for 6D (3D + yaw,pitch,roll) poses.
- poses::CPose3DQuatPDF for 7D (3D + quaternion) poses.
Each of the three above abstract classes has implementations for different PDF representation (particles, a gaussian, a sum of gaussians,…).
This technical report contains some of the derivations of the implemented Jacobians: