SOKOBAN : Single-Agent Search term project Artificial Intelligence 2000, Spring Shin Saim Choi Dong-jin - PowerPoint PPT Presentation

Loading...

PPT – SOKOBAN : Single-Agent Search term project Artificial Intelligence 2000, Spring Shin Saim Choi Dong-jin PowerPoint presentation | free to download - id: af884-ZmZmY



Loading


The Adobe Flash plugin is needed to view this content

Get the plugin now

View by Category
About This Presentation
Title:

SOKOBAN : Single-Agent Search term project Artificial Intelligence 2000, Spring Shin Saim Choi Dong-jin

Description:

hard to estimate # of stone push needed to goal ... Cheap stack. cf. List. Limit of solution length. Earlier iterations are small : ... – PowerPoint PPT presentation

Number of Views:87
Avg rating:3.0/5.0
Slides: 18
Provided by: mir83
Learn more at: http://ai.kaist.ac.kr
Category:

less

Write a Comment
User Comments (0)
Transcript and Presenter's Notes

Title: SOKOBAN : Single-Agent Search term project Artificial Intelligence 2000, Spring Shin Saim Choi Dong-jin


1
SOKOBAN Single-Agent Searchterm project
Artificial Intelligence2000, SpringShin
SaimChoi Dong-jin
2
Overview
  • Challenging area in sokoban
  • Detailed implementation
  • Experiments
  • Result and conclusion

3
Challenging area in sokoban
  • Deadlock
  • unsolvable problem configuration
  • irreversible moves
  • Search-Space Size
  • average 1018
  • Lower Bound
  • hard to estimate of stone push needed to goal
  • require moving stones through and away from the
    goal

4
Detailed implementation
  • Deadlock
  • IDA Transposition Table
  • Search space size
  • Deadlock Table Macro
  • Lower bound
  • Move ordering Minmatching

5
IDA
  • Search tree space Linear size
  • cf. DFS Exponential size
  • Cheap stack
  • cf. List
  • Limit of solution length
  • Earlier iterations are small
  • Exponentially growing size

6
Transposition Table
  • Avoiding cycles
  • Avoiding duplicating work
  • Storing recently visited entry

7
Minmatching
  • Lower bound heuristic
  • Minimum of pushing stone
  • Each stone decide particular goal.
  • Minimize sum of distances
  • Deadlock detection

8
Minmatching(Cont.)
  • Example1

9
Minmatching(Cont.)
  • Example2

10
Move Ordering
  • Lower bound heuristic
  • Finding right sequence of moves

11
Deadlock Table
  • Avoiding trivial deadlock
  • Need off-line computation
  • Storing deadlock information
  • Problem
  • Computing time and space

12
Deadlock Table(Cont.)
13
Macro Moves
  • Reducing search space
  • Combining several actions to super action
  • Too many macros
  • Tunnel macros
  • One-Way Tunnel Macros
  • Two- Way Tunnel Macros

14
Tunnel Macros
  • Example1
  • Example2

15
Effect of Macros
16
Experiment
17
Conclusion
  • SOKOBAN Solution
  • Old techniqueNew technique
  • SOKOBAN need various heuristics and search
    technique.
  • SOKOBAN need very large search space
About PowerShow.com