跳到主要內容

臺灣博碩士論文加值系統

(44.211.31.134) 您好!臺灣時間:2024/07/22 20:37
字體大小: 字級放大   字級縮小   預設字形  
回查詢結果 :::

詳目顯示

: 
twitterline
研究生:黃品達
研究生(外文):Huang, Pin-Ta
論文名稱:論遊戲角色技能與其自動生成方法── 利用基因演算法演化行為樹結構
論文名稱(外文):Automatic Generation Method of Game Character Skill - Evolving Behavior Tree Structure using Genetic Algorithm
指導教授:孫春在孫春在引用關係
指導教授(外文):Sun, Chuen-Tsai
口試委員:陳一平袁賢銘孫春在
口試委員(外文):Chen, I-PingYuan, Shyan-MingSun, Chuen-Tsai
口試日期:2020-06-23
學位類別:碩士
校院名稱:國立交通大學
系所名稱:資訊科學與工程研究所
學門:工程學門
學類:電資工程學類
論文種類:學術論文
論文出版年:2021
畢業學年度:109
語文別:中文
論文頁數:84
中文關鍵詞:遊戲技能基因演算法行為樹自動生成
外文關鍵詞:Skills in GameGenetic AlgorithmBehavior TreeAutomatic Content Generation
相關次數:
  • 被引用被引用:0
  • 點閱點閱:241
  • 評分評分:
  • 下載下載:8
  • 收藏至我的研究室書目清單書目收藏:0
遊戲日新月異不斷推陳出新,究其原因還是因為人類總是傾向求新求變,大部分遊戲都經不起歲月的洗禮而被淘汰,其中也不乏有很多遊戲都給玩家留下深刻美好的回憶,但也幾乎都只能剩下回憶,能真正歷久不衰的遊戲少之又少,當一款遊戲不再能給玩家帶來新的刺激就容易被放棄,而現今也有一些遊戲透過生成技術自動產生遊戲內容,不斷帶來新的體驗,但其生成方法通常過於單調,跟不上玩家期待的變化。
對此狀況本研究利用基因演算法結合行為樹,試圖生成邏輯性較強的遊戲內容,其中以遊戲角色的技能為主要生成對象,實驗中會指定一些應該達成的目標,但這些目標大多不會直接影響模型生成的做法和結構,留給模型很高的自由度,模型會將多個簡單的效果加上迴圈、判斷式等結構組合成技能,再讓生成的各種技能實際進入遊戲環境中測試,經過不斷的嘗試摸索出什麼樣的結構和參數能獲得高分,藉此完成符合預期甚至是超出預期的技能。
最後的實驗結果顯示,本研究使用的方法確實能因應不同的要求生成不同的技能效果,並且有機會訓練出超乎使用者預期的技能,顯示出本研究使用的模型除了代工以外也具有創造的能力,而生成出的技能會是易讀且可以隨意修改的,不管是對於開發者或是玩家都能便利的使用。
The game is constantly progressing, and the reason is that humans are always inclined to seek innovation and change. Most of the games are not able to remain year by year. There are still many games that leave players with deep and beautiful memories, but almost only memories left. There are very few games that can last forever. When a game can no longer bring new excitement, it would be easy to be abandoned. Nowadays there are some games that automatically generate game content through generation technology constantly bring new experiences, but its generation method is usually too monotonous to keep up with the changes expected by players.
In this situation, this study uses Genetic Algorithm combined with Behavior Trees trying to generate more logical game content. Among them, the skills of game characters are the main objects to be generated. In the experiment, some goals that should be achieved will be assigned in the experiment, but most of these goals will not directly affect the procedure and structure of model generation, leaving the model with a high degree of freedom. The model will combine multiple simple effects with loops, conditional statement, and other structures into skills, and then let the generated skills practically enter the game environment. After lots of trial and error to find out what kind of structure and parameters can get high scores, it can finally complete the skills that meet or even exceed user expectations.
The experiment results show that the method used in this study can indeed generate different skill effects according to different requirements, and have the opportunity to get skills that exceed the user's expectations. The generated skills will be easy to read and can be modified at will. For both developers and players can easily use.
摘要 i
Abstract ii
表目錄 vii
圖目錄 viii
一、 緒論 1
1.1 研究動機 1
1.2 研究背景 2
1.2.1使用自動生成的原因──增加遊戲豐富度 2
1.2.2使用自動生成的原因──減少人力成本 3
1.2.3現今遊戲中的自動生成 4
1.2.4抽象行為 6
1.2.5技能 (Skill) 在遊戲中的定位 7
1.2.6角色扮演遊戲 (Role-Playing Game) 8
1.2.7行為樹 (Behavior Tree) 8
1.3 研究目標 10
二、 遊戲技能探討 11
2.1 技能系統介紹 11
2.1.1典型的技能系統 11
2.1.2技能本質探討 13
2.2技能的分析與分類 15
2.2.1常用分類──職業 (Class) 15
2.2.2常用分類──效果類型 17
2.2.3常用分類──使用方式 18
2.2.4常用分類──使用頻率與條件 21
2.2.5其他分類 21
2.2.6 技能連動 21
2.3技能的取得 22
2.3.1職業技能的取得 22
2.3.2深入探討 23
2.4技能系統實例──流亡黯道 (Path of Exile) 26
2.4.1職業 (Class) 和天賦 (Passive Skill) 26
2.4.2技能 27
2.4.3裝備 28
三、 研究方法 29
3.1 研究架構 29
3.1.1建立基本系統 30
3.1.1.1編碼 (Encoding) 30
3.1.1.2演化 33
3.1.2製作基礎元件 38
3.1.3投入遊戲環境訓練 42
3.2 實驗工具介紹 43
3.2.1遊戲引擎Unity 43
3.2.2行為樹工具Behavior Designer 44
四、 研究結果 45
4.1 Tracing 45
4.1.1第一版實驗 45
4.1.2第二版實驗 50
4.1.3第二版實驗 (參數搜尋優化) 52
4.2 HitBack 53
4.3 Deflection 56
4.4 Explosion 60
4.4.1第一版實驗 60
4.4.2第二版實驗 63
4.4.3第三版實驗 66
4.5 Explosion Arrow 68
4.5.1第一版實驗 68
4.5.2第二版實驗 71
4.5.3第三版實驗 74
五、 結論與建議 77
5.1 結論 77
5.2 研究限制與未來展望 78
5.3 未來展望 78
參考文獻 80
附錄一、實驗中使用的Task 82
附錄二、實驗中使用的Goal 84
[1] Pac-Man. [PC Game].Namco, 1980.
[2] Unity. [PC Software].Unity Technologies, 2005. https://unity.com/
[3] 梁世佑, "當抽卡只剩空虛:轉蛋機制的心理與樂趣理論." [Online]. Available: https://www.u-acg.com/archives/15227
[4] Souls series. [Video Game].FromSoftware, 2009~2016.
[5] League of Legends. [PC Game].Riot Games, 2009.
[6] Puzzle & Dragons. [Mobile Game].GungHo Online Entertainment, 2012.
[7] Candy Crush Saga. [Mobile Game].King, 2012.
[8] The Legend of Zelda series. [Multi-Platform Games].Nintendo, 1986~2019.
[9] Minecraft. [Multi-Platform Games].Mojang Studios, 2011.
[10] Overcooked. [Multi-Platform Games].Ghost Town Games, 2016.
[11] Diablo III. [PC Game].Blizzard Entertainment, 2012.
[12] Rogue. [PC Game].A.I. Design, 1980.
[13] "Roguelike." [Online]. Available: https://zh.wikipedia.org/wiki/Roguelike
[14] Dead Cells. [Multi-Platform Games].Motion Twin, 2018.
[15] Hearthstone. [Multi-Platform Games].Blizzard Entertainment, 2014.
[16] Path of Exile. [PC Game].Grinding Gear Games, 2013.
[17] Pokémon. [Multi-Platform Games].Nintendo, 1996~2019.
[18] D. Isla, "GDC 2005 Proceeding: Handling Complexity in the Halo 2 AI." [Online]. Available: https://www.gamasutra.com/view/feature/130663/gdc_2005_proceeding_handling_.php
[19] C.-U. Lim, R. Baumgarten, and S. Colton, "Evolving behaviour trees for the commercial game DEFCON," in European Conference on the Applications of Evolutionary Computation, 2010: Springer.
[20] Diablo II. [PC Game].Blizzard Entertainment, 2000.
[21] Jacek Wesołowski, "Beyond Pacing: Games Aren't Hollywood." [Online]. Available: https://www.gamasutra.com/view/feature/4032/beyond_pacing_games_arent_.php?print=1
[22] Extra Credits, "Pacing - How Games Keep Things Exciting - ". [Online]. Available: https://www.youtube.com/watch?v=5LScL4CWe5E
[23] Extra Credits. "Shiver with Anticipation - How Horror Games Create Tension Cycles -." https://www.youtube.com/watch?v=OyiAR2BXtKU (accessed
[24] 尋找靈感, "如何累積恐懼?以《直到黎明》為例談恐怖遊戲設計概念." [Online]. Available: http://www.u-acg.com/archives/5786
[25] MapleStory. [PC Game].Wizet, 2003.
[26] multi-person discussion, "【問題】關於練到200等之後," 2019. [Online]. Available: https://forum.gamer.com.tw/C.php?bsn=7650&snA=1003605.
[27] Nioh. [Video Game].Team Ninja, Koei Tecmo, 2017.
[28] Heroes of the Storm. [PC Game].Blizzard Entertainment, 2015.
[29] DOTA 2. [PC Game].Valve, 2013.
[30] Guild Wars 2. [PC Game].ArenaNet, 2012.
[31] Destiny 2. [Multi-Platform Games].Bungie, 2017.
[32] Borderlands 3. [Multi-Platform Games].Gearbox Software, 2K Games, 2019.
[33] Remnant: From the Ashes. [Multi-Platform Games].Gunfire Games, 2019.
[34] Mega Man Battle Network series. [Video Game].Capcom, 2001~2009.
[35] Sekiro: Shadows Die Twice. [Multi-Platform Games].FromSoftware, 2019.
[36] Watch Dogs. [Multi-Platform Games].Ubisoft, 2014.
[37] biaodianfu, "树形结构数据存储方案." [Online]. Available: https://www.biaodianfu.com/adjacency-list.html
[38] Y. Kaya, M. Uyar, and R. Tekin, "A Novel Crossover Operator for Genetic Algorithms: Ring Crossover," CoRR, vol. abs/1105.0355, 01/01 2011.
[39] M. Nicolau, D. Perez-Liebana, M. O’Neill, and A. Brabazon, "Evolutionary Behavior Tree Approaches for Navigating Platform Games," 2016.
[40] M. Estgren and E. S. V. Jansson, "Behaviour Tree Evolution by Genetic Programming – Learning Novel Bot Behaviours in a 2D Top-Down Arena Shooter –," 2017.
[41] Behavior Designer. [Unity Plug-in].Opsive, https://opsive.com/assets/behavior-designer/
[42] "Overwatch," ed: Blizzard Entertainment, 2016.
連結至畢業學校之論文網頁點我開啟連結
註: 此連結為研究生畢業學校所提供,不一定有電子全文可供下載,若連結有誤,請點選上方之〝勘誤回報〞功能,我們會盡快修正,謝謝!
QRCODE
 
 
 
 
 
                                                                                                                                                                                                                                                                                                                                                                                                               
第一頁 上一頁 下一頁 最後一頁 top