JAIST Reposi https://dspace.j Title ターン制戦略ゲームにおけるベンチマークマップの提 案 Author(s) 木村, 富宏 ; 池田, 心 Citation ゲームプログラミングワークショップ 2016 論文集, 2016: 36-43 Issue Date 2016-10-28 Type Conference Paper Text version author URL Rights http://hdl.handle.net/10119/14084 社団法人情報処理学会, 木村富宏, 池田心, ゲームプログラミングワークショップ2016 論文集, 2016, 36-43. ここに掲載した著作物の利用に関する注意 : 本著作物の著作権は ( 社 ) 情報処理学会に帰属します 本著作物は著作権者である情報処理学会の許可のもとに掲載するものです ご利用に当たっては 著作権法 ならびに 情報処理学会倫理綱領 に従うことをお願いいたします Notice for the use of this material: The copyright of this mate retained by the Information Processi Japan (IPSJ). This material is publi web site with the agreement of the a the IPSJ. Please be complied with Co of Japan and the Code of Ethics of t any users wish to reproduce, make de work, distribute or make available t any part or whole thereof. All Right Copyright Information Processing S Japan. Description 第 21 回ゲームプログラミングワークショップ Japan Advanced Institute of Science and
1,a) 1,b) TUBSTAP TUBSTAP Offering New Benchmark Maps for Turn Based Strategy Game Tomihiro Kimura 1,a) Kokolo Ikeda 1,b) Abstract: Tsume-shogi and Tsume-go, mating problem of Shogi or Go, are sub-problem of these games. They have been created by many authors, have been played by many players for training, and have been used for evaluating the performance of computer algorithms. Also, benchmark problems are often used in the area of optimization and mathematical programming, for evaluating the specific/total performance of algorithms. We consider such benchmark problems are needed also for TUBSTAP, turn based strategy games. In this paper, we propose many benchmark problems, according to required abilities for these games, and from easy to difficult problems. Finally we show the performance of the existing open-source programs. 1. 1 a) kt499887@gmail.com b) kokolo@jaist.ac.jp TUBSTAP [1] TUBSTAP 10 [2]
2. TUBSTAP [2] [3] [4] [5] web 3. 8 8 HP HP HP HP HP HP HP 10 TUBSTAP 2016 3 UEC-GAT [6] 3 Military, M-UCT, M3Lee TUBSTAP
1 run A1 2 run A2 3 run A3 4 chase A1 Military HP 1 Military HP 4. 4.1 [7], [8] (chase) (run) chase run chase A1 5 chase A2 6 chase A3 7 run B1 8 run B3 9 chase B1 10 path A1 4.2 1 AI HP run chase 1 2 HP HP
1 AI Military M-UCT M3Lee run A1 39 0 0 6 run A2 39 0 0 2 run A3 19 0 2 0 chase A1 19 0 10 9 chase A2 19 0 7 10 chase A3 19 0 2 0 run B1 10 10 1 0 run B3 30 0 0 0 chase B1 19 4 8 7 2 chase A3 9 10 11 12 13 14 127 52 437 15 92 0 33 11 113 5 29 0 0 289 0 1236 0 290 3 AI path A1 29 0 10 10 path A2 11 0 9 3 path A3 39 0 1 2 5 10 M3Lee 39 3 6 2 run A1 5 4 6 7 run A1 Military 8 run B1 2 9 run B3 19 3 3 3 3 3 4.3 chase A3 6 13 13 14 2 MINMAX 2 9 10 12 14 14 12 4.4 TUBSTAP path 3 AI 10
11 path A2 12 path A3 15 block A3 HP 1 16 block B1 HP 10 13 block A1 4 14 block A2 AI block A1 9 0 6 0 block A2 9 0 3 0 block A3 9 0 0 0 block B1 9 10 4 0 block B2 9 0 0 0 11 12 5. 4 AI 14 17 block B2 HP 4 6 2 13 15 2 AI 16 4 Military HP10 HP 1 HP 17 1 3 AI block AI 6. select A,B
18 select A1 HP 4,6,8,10 2,3,4,5 19 select C1 HP 3,10 HP 1,1,10 22 guard A1 23 guard B1 HP 6,4,4 24 guard B2 25 guard B3 20 select B1 21 select B2 5 AI select A1 1 0 10 2 select B1 1 0 10 2 select B2 1 0 2 0 select C1 9 9 7 10 5 AI 18 HP 4 24 1 M-UCT 20 M-UCT 21 6 M-UCT 19 AI HP10 7. 26 guard C1 27 guard D1 28 tag A1 29 tag B1 6 AI tag A1 3 9 guard A1 9 9 0 4 guard B1 19 1 5 1 guard B2 19 0 0 2 guard B3 19 0 0 0 guard C1 19 10 1 10 guard D1 19 0 0 3 tag A1 9 10 10 10-7 7 5 6-5 3 6 6-3 0 0 0 tag B1 9 0 0 0 guard tag 6 AI
30 puzzle A1 31 puzzle A2 32 puzzle A3 33 puzzle B1 22 2 23 25 AI 26 guard A1 AI guard A1 27 guard A1, guard C1 2 1 2 AI 28 2 1 2 3 AI 9 AI 7,5,3 29 5 9 5 AI 8. 7 AI 31 34 puzzle B2 35 puzzle B3 36 puzzle C1 37 puzzle C2 38 puzzle C3 7 AI puzzle A1 19 10 10 10 puzzle A2 19 1 10 10 puzzle A3 9 0 0 0 puzzle B1 29 0 8 10 puzzle B2 29 0 2 10 puzzle B3 29 0 0 9 puzzle C1 50 0 3 7 puzzle C2 50 0 0 1 puzzle C3 50 0 0 0 select B
30 AI puzzle A2 32 AI puzzle B 2 tag 33 34, 35 AI puzzle C 38 2 AI 37 36 puzzle B 8.1 AI 39 40 39 40 (7,5) 39 40 AI AI AI HP TUBSTAP [1] GPW2013 pp 146 153 [2] TUBSTAP http://www.jaist.ac.jp/is/ labs/ikeda-lab/tbs/ [3] TSPLIB http://elib.zib.de/pub/packages/ mp-testdata/tsp/tsplib/tsplib.html [4] Huang, S.-C., Muller, M.. Investigating the limits of Monte-Carlo tree search methods in computer Go. CG 2013,pp. 39 48 [5] : 2006 [6] GAT (Game AI Tournaments @UEC) http: //minerva.cs.uec.ac.jp/gat_uec/wiki.cgi?page=\% C2\%E81\%B2\%F3GAT2016 2016-09-28 [7] : 2006 [8] Bourg, D.M. and Seemann, G. : AI 2005 9.