Zachary Dodds, Steven Santana, Brandt Erickson, Kamil Wnuk, Jessica Fisher, and Matt Livianu
Monte Carlo Localization (MCL) is a robust, probabilistic algorithm for estimating a robot’s pose within a known map of the environment. Now a crucial component of many state-of-the-art robotic systems, MCL’s simplicity, flexibility, and power make it a technique particularly suitable for an undergraduate AI or robotics classroom. Several popular low-cost mobile platforms, such as the Lego RCX and the Handyboard, lack the visualization bandwidth and/or the processing power required to effectively experiment with MCL. This paper describes a series of assignments that guide students toward an implementation of MCL on a relatively new, moderately priced platform, Evolution Robotics’ ER1. We also report on experiences and lessons learned over the past year of using the ER1 for teaching undergraduate robotics.