Smarter AI for the 2D RPG Project Accelerator

Hey devs 👋

I’m excited to share a big step forward in the AI system of the 2D RPG Project Accelerator: the first working version of GOAP (Goal-Oriented Action Planning) is now live and integrated into the core framework!

If you've ever wanted to build AI that thinks for itself, makes decisions dynamically, and feels truly alive — GOAP is where the magic happens.

---

✨ What is GOAP?

GOAP stands for Goal-Oriented Action Planning, and it’s a flexible AI approach that lets agents figure out their own way to reach a goal. Rather than scripting every behavior, you give the AI a set of goals and actions with preconditions and effects — and it handles the planning on its own.

This means instead of hardcoding logic like:

if (IsHungry)
    GoEat();

...you define actions like "FindFood" and "EatFood", and a goal like IsFed: true, and the AI connects the dots. 🙌

---

🕹 How It Works in the Project

Here's a breakdown of how GOAP is implemented inside the 2D RPG Project Accelerator:

✅ World State

A list of key-value facts about the game world or the agent's internal state.

worldState = {
    "HasAxe": false,
    "IsHungry": true,
    "HasWood": false,
    "EnemyNearby": true
};

⚙️ Actions

Each action has:

• Preconditions: What must be true to perform this action

• Effects: What will be true after this action

• Cost: Used to determine the most efficient plan

public class ChopTreeAction : GoapAction
{
    public ChopTreeAction()
    {
        AddPrecondition("HasAxe", true);
        AddEffect("HasWood", true);
        cost = 2f;
    }

    public override bool Perform(GameObject agent)
    {
        Debug.Log("Chopping tree...");
        return true;
    }
}

🔐 Goals

Agents evaluate which goal to pursue, like:

{ "HasWood": true }

📏 Planning

The planner finds the best sequence of actions to get from the current world state to the goal. It's similar to pathfinding, using a search algorithm to navigate action dependencies.

Example:

If the goal is HasWood = true, and the agent doesn’t have an axe, the AI will automatically plan:

• CraftAxe

• ChopTree

And execute them in that order. Fully dynamic.

---

📊 Goap system in place with every agent has it’s own debug data

Here's where you can see the GOAP debug view

---

🚀 Why GOAP Matters

This system is a huge win for building more immersive gameplay:

• ✅ AI feels alive and reactive

• ✅ Modular – new actions can be dropped in and everything just works

• ✅ Easier to manage than giant state machines or behavior trees

With GOAP, your NPCs can plan and act based on what’s happening in the world — not what you told them to do step-by-step.

---

🕹 GOAP AI Combat In Action

---

🌟 What’s Next?

Now that the base GOAP system is live, I’m looking at some exciting expansions:

• ⚔️ Better combat AI (plan escape, flank, call for backup)

• 🌳 Resource gathering and trading

• 🧳 Multi-agent collaboration

• 🤔 Emotional states (fear, anger, etc.) that affect decision-making

---

📃 Try It Now

GOAP is now part of version 1.2.0 of the 2D RPG Project Accelerator. Update your Unity project and start building smarter agents today!

https://assetstore.unity.com/packages/2d/2d-rpg-project-accelerator-296300

Got feedback, questions, or ideas? Drop them below. Always happy to chat AI.

Stay tuned for more updates — and keep building! 🚀