Generate samples from multivariate gaussian covariances
It would be useful for some tests/demos to be able to generate samples from multivariate gaussian distributions (emulating sensor data). I ran into this small project that we could simply add to wolf:
https://github.com/beniz/eigenmvn
which is GNU LESSER GENERAL PUBLIC LICENSE.
It provides a simple API that can do things like:
Eigen::EigenMultivariateNormal<double> noise(Vector3d::Zero(), pow(std, 2)*Matrix3d::Identity());
Vector3d sampled_noise = noise.samples(1);
We could also provide our own implementation in the include/math/covariance.h
file.
Edited by Mederic Fourmy