Course Descriptions
MECH_ENG 495: Selected Topics: Active Learning in Robotics

Quarter Offered

Spring : MW 3-4:20pm ; Todd Murphey


Active learning takes advantage of control authority to improve data as it is being acquired.  Robotics applications are particularly sensitive to how data is taken.  For instance, machine vision is obviously dependent on where a camera is pointed, and learning algorithms cannot independently overcome a camera pointing in an incorrect direction.  This class will cover recent results in control, planning, and machine learning, focusing on creating a closed-loop relationship between information needs and physical actions of a robot.  Some of the course will be in lecture format, and a significant portion of the course will be reading current papers in the area.  Students should have strong programming skills and experience in robotics.