Unleashing Intelligent AI in F.E.A.R. with Goal Oriented Action Planning

Table of Contents

1. Introduction

• Overview of AI in video Game development
• Previous episodes and techniques discussed

2. Goal Oriented Action Planning: Theory

• Introduction to automated planning
• Modeling problems and world state
• Actions, preconditions, and effects
• Planning process and A* search

3. Goal Oriented Action Planning: Implementation

• Adaptation of STRIPS planning for games
• Dr. Jeff Orkin and the development of GOAP
• Finite State Machine and animation states
• Smart objects and their role in GOAP

4. GOAP in F.E.A.R

• Application of GOAP in F.E.A.R
• Goals and actions in the game
• Validation and re-planning
• Performance overhead and the case of rats

5. Impact of GOAP in Video Games

• Popular games adopting GOAP
• Comparison with behavior trees and HTN planning
• Other resources on GOAP

6. Conclusion and Thank You

🤖 Goal Oriented Action Planning: Bringing Intelligence to Game Characters 🎮

In the world of video game development, Artificial Intelligence (AI) plays a crucial role in creating realistic and engaging experiences. In previous episodes of AI 101, we have explored various AI methodologies used in games, such as navigation meshes, finite state machines, and behavior trees. In this episode, we delve into another popular technique called Goal Oriented Action Planning (GOAP) and its implementation in one of the most iconic games of its time – First Encounter Assault Recon (F.E.A.R).

1. Introduction

Video games have come a long way in terms of AI. From simple rule-based systems to complex decision-making algorithms, developers constantly strive to create AI characters that challenge players and provide immersive gameplay. In the opening episodes of AI 101, we have discussed established AI methodologies used in games, setting the stage for our exploration of GOAP.

2. Goal Oriented Action Planning: Theory

To comprehend GOAP and its application in F.E.A.R and other games, we need to understand its theoretical foundations. GOAP is derived from the concept of automated planning, wherein a system attempts to determine a sequence of actions to achieve a specific goal. The problem is modeled using a language or encoding that represents the current state of the world and the facts or predicates associated with it.

Actions in GOAP are divided into three parts: the items involved, preconditions that must be met, and effects that result from completing the action. For example, opening a closed door requires the door to be closed and the non-player character (NPC) to be in the same room. Planning involves creating a plan that incorporates these actions and changes the world state accordingly, allowing the NPC to navigate from one room to another.

3. Goal Oriented Action Planning: Implementation

In the development of GOAP, Dr. Jeff Orkin, the AI lead for Monolith Productions, played a significant role. Building on the STRIPS planning system, GOAP utilized a Finite State Machine (FSM) as its core driver. Unlike traditional FSMs with numerous states, GOAP's FSM focuses on three key states: moving into a position, playing an animation, and interacting with smart objects.

The key insight behind GOAP's animation-driven FSM lies in the observation that many game characters primarily perform animations that convey intelligence and responsiveness. Each state in GOAP represents a specific animation, such as walking, running, or interacting with the environment. By executing the appropriate animations in the right sequence, GOAP creates the illusion of intelligent behavior.

4. GOAP in F.E.A.R

F.E.A.R, released in 2005, brought GOAP into the spotlight within the gaming community. By assigning goals to AI characters, F.E.A.R leveraged the planning system to generate action plans tailored to each character's situation. The game's AI system utilized over 70 goals, including patrol and combat objectives. These goals were prioritized based on the character's knowledge of the player's proximity.

The planning process in F.E.A.R involved searching for suitable actions within the state space to achieve the assigned goals. Once a plan was devised, it underwent validation to ensure its feasibility and consistency. However, the dynamic nature of the game environment necessitated continuous re-planning to accommodate changes caused by player actions or interactions with the environment.

5. Impact of GOAP in Video Games

The success of GOAP in F.E.A.R prompted its adoption in numerous video games, both popular and cult classics. Games like Condemned: Criminal Origins, S.T.A.L.K.E.R: Shadow of Chernobyl, and Deus Ex: Human Revolution incorporated GOAP or similar planning methodologies. However, in recent times, the popularity of GOAP has declined, with developers favoring Hierarchical Task Network (HTN) planning. Nevertheless, GOAP continues to be a valuable technique in the repertoire of AI systems in video games.

6. Conclusion and Thank You

In conclusion, Goal Oriented Action Planning revolutionized the portrayal of AI characters in video games. F.E.A.R showcased the capabilities and potential of GOAP, and its influence extended to subsequent games. We have explored the theoretical underpinnings of planning, the implementation of GOAP in F.E.A.R, and its impact on the gaming industry. Thank you for joining us on this AI 101 journey, and we appreciate your support as we continue to explore the exciting world of AI and Games.

Highlights:

• GOAP (Goal Oriented Action Planning) is a technique used in video game development to create intelligent AI characters.
• GOAP utilizes planning algorithms to determine a sequence of actions that achieve specific goals set by designers.
• F.E.A.R introduced GOAP to the gaming community, demonstrating its effectiveness in creating realistic and engaging AI behavior.
• GOAP incorporates a Finite State Machine (FSM) with three states – moving, animating, and interacting with smart objects – to simulate intelligent behavior.
• Other popular games, including Condemned: Criminal Origins, S.T.A.L.K.E.R: Shadow of Chernobyl, and Deus Ex: Human Revolution, have adopted GOAP or similar planning methodologies.

FAQs

Q: What is the difference between GOAP and behavior trees? A: While both GOAP and behavior trees are AI techniques used in games, they have different underlying mechanisms. GOAP focuses on planning sequences of actions to achieve goals, while behavior trees organize AI behavior using hierarchical structures of tasks and conditions.

Q: Can GOAP accommodate dynamic changes in the game environment? A: Yes, GOAP can adapt to dynamic changes by continuously re-planning and validating its action sequences. However, this process can incur a performance overhead, as observed in the case of rats in F.E.A.R.

Q: What other planning methodologies are widely used in the gaming industry? A: In addition to GOAP, Hierarchical Task Network (HTN) planning is gaining popularity in modern games. HTN planning allows for the hierarchical organization of tasks, enabling more flexible AI behavior and complex decision-making.

Q: Where can I find more resources on GOAP and its implementation? A: The video description provides additional resources on GOAP and its application in games. Additionally, exploring the AI source code of F.E.A.R can provide valuable insights into the implementation of GOAP in a real-world game scenario.

Resources:

• Link to F.E.A.R AI source code
• Additional resources on GOAP