Review Test1

1
Review Test1
2
Robotics Future Technology
3
Future of Intelligent Systems

• Ray Kurzweil futurist
• A Long Bet
• Robot Soccer
• BBC News - What happened to the Robotic Age?
• Robotic Dance?

4
CHAPTER 5

• REVIEW
• of some important concepts to remember

5
Incremental Problem Formulation (The N-Queens
Problem)
6
(No Transcript)
7
Constraint Satisfaction

• A constraint satisfaction problem (CSP) is a
  problem composed of variables and constraints.
  The objective is to assign values to variable in
  such a way that all of the constraints are
  satisfied.
• A CSP can be more formally viewed as a triple
  ltD,X,Cgt where
• D represents a set of n domains
• X represents a set of n variables, each variable
  xi takes it value from Di
• C represents a set of constraints.

8
Two types of CSPs

• Finite domain CSPs
• Queens, Domains, Boolean
• Infinite Domain CSPs
• If Job1 takes 5 days, then Job3 will start
• Job1 5 lt Job3

9
thrashing

• Wasteful backtracking is known as

10
Variable Value Ordering

• Variable order selection seldom results in the
  most efficient search
• One efficient strategy
• Minimum Remaining Value (MRV) which chooses the
  value with the fewest legal values
• A.K.A Most Constrained Value
• A.K.A Fail-First Heuristic because it is the
  most likely to cause failure first

11
Propagating through constraints

• Forward Checking is a better use of constraints

• When every variable X is assigned, the forward
  checking process looks at each unassigned
  variable Y AND Deletes from Ys domain any value
  that is inconsistent with the value chosen for X.

• Think of Forward Checking as a way to
  incrementally compute the info needed for MRV to
  do its job

12
Arc Consistency

• ?nother backtrack-based algorithm that is even
  more efficient is called the maintaining arc
  consistency (MAC) algorithm.
• Arc Consistency provides fast method of
  constraint propagation stronger than forward
  checking.
• Arc Consistency can be provided as preprocessing
  step (just as forward checking can be used to
  preprocess).

13
MAC Algorithm p146

• Rather than just revising each domain
  corresponding with the value of each instantiated
  variable MAC makes the network arc-consistent
  with respect to the instantiated variables (all
  arcs must be consistent).
• MAC is applied repeatedly until no more
  inconsistencies remain.
• Node Consistency
• 1- consistency means that each individual
  variable by itself is consistent (node
  consistency)
• 2- consistency is arc consistency
• 3- consistency means that any pair of adjacent
  variables can always be extended to a third
  neighboring variable (i.e. path consistency or
  AC-3)
• AC-3 uses a queue to keep track of nodes that
  need to be checked for inconsistency

14
Intelligent backtracking looking backward

• Chronological backtracking a policy for when a
  branch of the search fails by backing up to the
  preceding variable and try a different value for
  it
• Chronological uses the more recent decision point
  to revisit

15
Things that cause a failure

• The conflict set
• Backjumping backtracks to the most recent
  variable in the conflict set