Andrew's Leap

Summer 1997

Planning and Control

Draft: June 14, 1997.

Back to Top Curriculum Page

1. Planning and Control
2. 
   (Not sure where to go with this, but I would like to give them AI x Modern Control. A robot is a Turing Machine wired to a ... Markov chain? Discrete state space discrete time system? State space formulations of perception, control, planning?)
   
   (Why is this last? As it provides conceptual organization for the whole project, it should be first. Also, we want students to work with the whole beast from the start, so they can worry about control/planning first.)
   
   (Maybe it should be dispersed through the curriculum. Intro to state space models at the front. Then more detailed treatment of planning, perception, and control following programming, sensing, and powertrain topics, respectively?)
   
   Handouts: Look around. Excerpt from Simon's book? Discrete models of action, graph searching. Write something about using slip to follow walls?
   
   Lab: Program robot to navigate stripes on the floor? Program it to follow walls? Implement mobile robot path planner?
   
   
   1. Simon's ant: simple mechanisms can interact with the environment in interesting ways.
   2. Changing task state: what actions are available
   3. Estimating task state: fusing sensory information with motor history
   4. Control: feedback loops can modify system behavior. A wrapper on the task system.
   5. Planning: what sequence of actions will get you there? Searching the task state graph.

Robot projects