Presentation is loading. Please wait.

Presentation is loading. Please wait.

Use “Search” for Pathfinding FactorySchool Library Hospital Park Newsagent University church Example from Alison Cawsey’s book start finish.

Similar presentations


Presentation on theme: "Use “Search” for Pathfinding FactorySchool Library Hospital Park Newsagent University church Example from Alison Cawsey’s book start finish."— Presentation transcript:

1 Use “Search” for Pathfinding FactorySchool Library Hospital Park Newsagent University church Example from Alison Cawsey’s book start finish

2 Breadth First Search library school hospital factory park newsagent universitychurch Put things on the back of the list “to visit later”

3 Depth First Search library school hospital factory park newsagent universitychurch Put things on the front of the list “to visit later”

4 Breadth vs. Depth Which is better? library school hospital factory park newsagent university church stadium grocers marketbridge fountain

5 Breadth vs. Depth Which is better? library school hospital factory park newsagent market church stadium grocers bridge university

6 Breadth vs. Depth Which is better?  Depends on problem  Breadth usually needs a lot more memory  Remember all the bits you need to expand next  Breadth could be good if  There are many long dead ends,  But one very short successful path  Depth could be good if  There are many successful paths  But all are quite long  Can also combine – set a depth limit

7 What about a big open space?  See demo…  Break it up into squares  Each node has 8 children (Be careful about looping)  That’s an awfully big tree!  Need some clever tricks…  How would a human do it?  Heuristics  Search we did before is called “blind” or “brute force” (not clever)  Heuristic is a clever rule of thumb

8 Hill-climbing with Heuristic FactorySchool Library Hospital Park Newsagent University church start finish Heuristic: how close to goal

9 Hill-climbing with Heuristic Library 9 School 7 Hospital 5 Factory 5 Park 6 Newsagent 0 University 3Church 4 finish Heuristic: how close to goal

10 Hill-climbing with Heuristic FactorySchool Library Hospital Park Newsagent University church start finish Heuristic: how close to goal

11 Hill-climbing with Heuristic Library 9 School 7 Hospital 5 Factory 5 Park 2 Newsagent 4 University 0Church 4 finish Heuristic: how close to goal

12 Hill-climbing with Heuristic Heuristic: how close to goal from Russell and Norvig’s book

13 Best first Search Library 9 School 7 Hospital 5 Factory 5 Park 2 Newsagent 4 University 0Church 4 finish Heuristic: how close to goal  Order the list of nodes “to visit later”  Do best first  But try others later  Very good, e.g. in open space  Doesn’t consider how far we’ve come though  A* - more in practical

14 Remember: General Problem Solving  Problem formulation  Initial situation  Goal situation  Actions that can be done  +cost of action  Constraints  Task:  Find the best sequence of permissible actions that can transform the initial situation into the goal situation. 617 34 582

15 Search is an abstract technique…  Jugs problem  Two jugs, 4 litre and 3 litre  Want to get 2 litres in 4 litre jug  Formulate problem  Can represent state as (0,0) or (4,0) or (4,2)…  Actions:  Fill 4 litre ( _, _ )  ( 4, _ )  Fill 3 litre ( _, _ )  ( _, 3 )  Empty 4 litre ( _, _ )  ( 0, _ )  Empty 3 litre ( _, _ )  ( _, 0 )  What else?

16 Search for games: Minimax

17

18 Alpha-Beta pruning example Alpha value I get at least this Beta value I get at most this (if I go here)

19 Alpha-Beta pruning example No question about 3 now

20 What about real (hard) games?  So far we searched all the way to the end of the game  Not feasible in chess, branching factor 35  So far we didn’t use heuristics  Do a limited lookahead  Distance to goal? Evaluate the board state  Requires intelligence: pieces, their positions, and stage in game  How much lookahead?  Modern computer?  Alpha beta can give about double  4 moves ≈ human novice  8 moves ≈ human master  12 moves ≈ Deep Blue, Kasparov  Deep Blue had extra tricks to look further on interesting paths  Go  Branching factor ≈ 300… forget it!  Use databases of patterns

21 What is search good for?  Pretty much everything!  Pathfinding, puzzles, general problem solver, games  Scheduling deliveries  Arranging the CS1013 timetable  Diagnostic systems  find a set of malfunctions that explain the symptoms  Speech recognition  find the right sequence of words  Finding templates/models to match a visual scene  Learning is search for a hypothesis  Planning systems  Find a sequence of actions that achieves a given goal  We will look at this next week

22 Defence A big user of AI. "... the deployment of a single logistics support aid called DART during the Desert Shield/Storm Campaign paid back all US government investment in AI/KBS research over a 30 year period." Tate A. Smart Planning. ARPI Proc. 1996.

23 Search is an abstract technique…  What are we really doing here?  What is the science of abstraction?  Mathematics  Look at problems abstractly  See that they’re the same  Use one technique for many problems

24 Note on Heuristics  Hard to come up with a good heuristic  Often use human intelligence  Is the chess computer smart?  What about TD-Backgammon  Try the Missionaries and Cannibals  http://www.learn4good.com/games/puzzle/boat.htm http://www.learn4good.com/games/puzzle/boat.htm  Interesting because human heuristics go awry  Computer is not confused  Remember the first law 617 34 582 123 45 678

25 Recap: What have you learned about “Search”  “Blind” or “brute force” techniques  Breadth first  Depth first  Heuristic techniques  Hill-climbing  Best first  A* - more in practical  General problem solving  Game playing  Minimax  Alpha-Beta pruning  Search can apply to many diverse problems  Makes some tasks simple for computers  Heuristics need some intelligence  Musings…  Computer: Some simple tricks can go a long way  Power of computer to store so much and go so fast  Just like life – simple blocks


Download ppt "Use “Search” for Pathfinding FactorySchool Library Hospital Park Newsagent University church Example from Alison Cawsey’s book start finish."

Similar presentations


Ads by Google