POLAR (Preference Optimization and Learning Algorithm for Robotics) is a MATLAB toolbox that provides a unified framework for human-in-the-loop preference-based learning.