Robust Cost

In robotics, we often come across estimating maximum likelihood or maximum-a-posterior problems quite frequently. And the choice of white noise that we introduce to estimate makes a lot of difference. This blog is about analyzing different cost functions and benchmarking them on some standard problems.

This post is based on the paper titled At All Costs: A Comparison of Robust Cost Functions for Camera Correspondence Outliers [1]. This blog is accompanied by code written in C++14 at: kvmanohar22/robust_cost.