• @osa_k
  • Write-up
• @Mi_Sawa
• @camypaper

Implementing Scheme-like language (Rabbit) to write Lambda-man AI. Final submission is rabbit/uct.rabbit (Despite of its name, it doesn't use UCT algorithm.)

The AI performs breadth-first search to find pills then scores them by distance. It also consider nearby ghosts which affect with decreasing score on surrounding cells. Then AI goes along to a highest-scored cell.

Implementing macro preprocessor (Mofu) to write Ghost AIs. Implementing the EDITOR (not plugin) for the Mofu and GHC. Final submission are mofu/tourist.mofu and mofu/traveler.mofu .

We implemented two types of AI, tourist and traveler. They generally calculate the relative position of the "target" and aims the "target". When they visited the same junction frequently, they changed the priority of the direction. The difference between these AI is "target" position. Traveler aims the present Lambda-man's position, but tourist aims the future Lambda-man's position. Future lambda-man's position was calculated from the difference between the present Lambda-man's position and past position of him. As a result, they will perform a pincer movement.

• bin/chino.rb
  • Rabbit -> GCC compiler
• bin/tippy.rb
  • Mofu -> GHC preprocessor
• bin/rize.rb
  • GCC simulator
• bin/syaro.rb
  • Game simulator (Incomplete; only simulates lambda-man)
• chino/*.rb
  • Library for bin/chino.rb
• maze/*.txt
  • Maze generated by maze_create.cs
• mofu/*.mofu
  • GHC preprocessor language
• rabbit/*.rabbit
  • Scheme-like language compiled into GCC
• rize/*.rb
  • Library for bin/rize.rb and bin/syaro.rb
• submission
  • Submitted files
• gcc_parse.rb
  • GCC specification parser
• gcc_spec.txt
  • GCC specification copied from reference page
• maze_create.cs
  • Maze map creator
• MainWindow.xaml
  • Mofu and GHC editor's design
• MainWindow.xaml.cs
  • Mofu and GHC editor's logic
• GHC.xshd
  • Editor's setting file for syntax highlight