GOAL-ORIENTED ACTION PLANNER MODULE DEVELOPMENT FOR UNREAL ENGINE

• Main Author: Irfan Permadi, Muhammad
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/46846
• Institution: Institut Teknologi Bandung

Video game is a game that we can play in an electronic system. As technology advances, the expectations and demands from market on video game also increase. One of the ways to improve a video game quality is by improving its Artificial Intelligence (AI) quality in order to create a more entertaining game. There are some ways to develop an AI system in video game, one of the most common approach is by using Finite State Machine (FSM). One of other approaches is by using an architecture called Goal-oriented Action Planning (GOAP). GOAP architecture could work better than FSM because of its dynamic behaviour and easier to scale up. The problem that’s raised in this final project is the availability of GOAP architecture as an alternative that can perform better than FSM, but it’s still rarely used. To solve that problem, in this final project a module that will apply GOAP architecture will be developed to be used by video game developer. This final project focuses on developing GOAP architecture in a game engine called Unreal Engine. This final project research starts with study of literature about how GOAP architecture works and how its relation with video game and AI. After understanding how GOAP architecture works, analysis will be done to determine components that have to be developed in order to create GOAP module. The next step is designing the system based on analysis result then implement it into the Unreal Engine game engine. After GOAP module has been developed, then a set of tests are done to evaluate its performance. Tests are done by using black-box approach independently to test GOAP module’s functionality, making video game simulation to test GOAP module’s usability, and doing sample survey from video game developer to test GOAP module’s performance based on user candidate’s perspective. Based on test results in this final project, it was concluded that GOAP module that has been developed can plan actions based or perform action planning based on action’s data and AI agent’s data by user. Based on independent test result and questionnaire from three video game developers it was known that not only GOAP module could perform well, but also could be useful in video game AI development.