This class focuses on modern techniques, based largely on probability theory, to address problems in sensing and estimation in mobile robotics. A running theme throughout the class is that robotics involves uncertainty at several different levels. The machinery of estimation theory and probability theory (Bayesian filtering), has been applied with great success to cope with uncertainty in sensing and actuation. The class will cover the relevant theory and applications to problems in robot localization, mapping, perception and manipulation. If time permits we will cover additional topics such as planning and extensions to multi-robot systems.

READ THIS CAREFULLY IF YOU PLAN TO TAKE THE CLASS

The treatment in class will be mathematical. Students are expected to know elementary probability theory, calculus, and linear algebra at the undergraduate level. There is no formal prerequisite for the class (don't believe what the catalogue says). If you want to register, and the office won't let you, send email to the instructor. Based on the mathematical treatment presented in class, students will be expected to complete a fairly sizeable class project. Most students will be expected to use a standard robot simulator (ROS) for the project. There are limited opportunities for advanced students to use a state-of-the-art robot (the PR2, shown on the right). In order to complete the project, students will need to have a reasonable background in programming. (Typical projects are expected to run into a few thousand lines of code). All the software tools used in the class run on Linux. Students are expected to be reasonably competent in programming under Linux (or some *nix flavor).

This page was created by Gaurav Sukhatme and is maintained by him. Last updated 8/20/11