Classes
Spring 2011
16-899C ACRL: Adaptive Control and Reinforcement Learning
Website: Webpage for ACRL
Spring 2010
16-899C ACRL: Adaptive Control and Reinforcement Learning
4:30-5:50 TR , SyllabusAdaptiveControlAndReinforcement, NSH 3002
Website: http://robotwhisperer.org/16899cS10/
Fall 2009
16-831: Statistical Techniques in Robotics
Website: http://robotwhisperer.org/16831F09/
16-831_syllabus
Probabilistic and learning techniques are now an essential part of building robots (or embedded systems) designed to operate in the real world. These systems must deal with uncertainty and adapt to changes in the environment by learning from experience. Uncertainty arises from many sources: the inherent limitations in our ability to model the world, noise and perceptual limitations in sensor measurements, and the approximate nature of algorithmic solutions. Building intelligent machines also requires that they adapt to their environment. Few things are more frustrating than machines that repeat the same mistake over and over again. We’ll explore modern learning techniques that are effective at learning online: i.e. throughout the robots operation. We’ll explore how the twin ideas of uncertainty and adaptation are closely tied in both theory and implementation
Previous classes
16-899C ACRL:
Adaptive Control and Reinforcement Learning
Machine Learning Techniques for Decision Making, Planning and Control