Hi all,
I’m not sure if this is the appropriate forum, but I’m looking for some guidance on AI asset packages in the Unity store. I am making a turn-based “sports” game involving multiple customizable teams. Since the combatants on the team, the arena and even the goals might change match to match, I’m looking for an AI system that’s more dynamic than a simple stimulus-response system. The AI would need to evaluate the strengths and weaknesses of all of the combatants in the match, evaluate the terrain, select from available actions and respond to unexpected events.
I have enough skills/experience to be able to modify a system that’s close to doing what I want, I’m just trying to avoid having to write the whole thing.
Does anyone have some suggestions for systems that might work?
Thanks!
Well, it’s going to be a challenge to write good AI for that. I would look for something like this Goal Oriented Action Planning for a Smarter AI | Envato Tuts+
And here’s open source implementation for Unity A new open source GOAP (Goal Oriented Action Planner) for AI
Also look into this chess algorithm tutorial. A step-by-step guide to building a simple chess AI It is a very solid explanation of base principles for turn based AI.
Thank you! The GOAP article mirrors one of the approaches I was thinking of taking.
Are there existing AI assets on the Unity store that might implement this approach?
Sure there are, but I’m unaware of them and can’t recommend any.
In practice, a decision tree system will get you to your goal fastest. The Asset Store has DecisionFlex and Decision Tree Toolkit.
If you have more time to devote to training your AI, start with How To: Unity Machine Learning Agents.
I have nothing against GOAP (Quest Machine uses it), but a decision tree will let you implement complex behavior faster, and with more time MLagents will probably let you train better agents.