Writing, Rewriting and Re-Rewriting Armello's AI System
summary
Rebuilds Armello’s AI by running the real Unity/C# gameplay code in a lighter “sim” path so decisions score actual game-state changes rather than hand‑coded heuristics, replacing a tangled inheritance tree with componentized managers, change‑based simulation and aggressive caching to avoid GC hits. Debug UI shows full path and card-play scoring, combats run ~50 Monte Carlo sims for outcomes, and planning runs in background while others act. Key outcomes: component over inheritance, change-only state diffs, integrated movement + card play.
AI-written summary · about
chapters (14)
- 0:00Intro
- 9:11Simulation
- 9:51Objections
- 11:09Why
- 14:40Message System
- 22:16Inheritance
- 29:14Debugging
- 34:56Designers
- 36:46Speed
- 39:45Good things
- 44:42Why is this good
- 52:15Conclusions
- 55:58Standing Questions
- 56:50Outro
related talks
01h 18mTotal War: how AI can deliver entertainment.- Duygu Cakmak & Guy Davidson [ACCU 2018]
39mEmbarrassing Mistakes: a short story about AI
20mStudio Speaks: The Reality of AI Adoption in Games
46mMind your NPCs and Q’s: what’s next for Game AI?
08mInterview with
01h 01mThe Forgotten AI of Video Games
38mPractical AI in Unity
48mMachine Takeover - Improving Your Game With Machine Learning & AI