• Data Analytics
  • Nov 24, 2020
  • By QOMPLX

QOMPLX Intelligence: Pathfinding In the Fog of War with StarCraft

QOMPLX Intelligence: Pathfinding In the Fog of War with StarCraft

This is the third in a multi-part QOMPLX Intelligence series that examines how QOMPLX uses the real-time strategy (RTS) game StarCraft as a training ground for research to support advanced reinforcement learning and effective decision making using machine learning and artificial intelligence. Download a PDF copies of  Part 3: Pathfinding in the Fog. We have also provided links to the previous installments of this series below.

In recent posts we've been digging deep into our team's ongoing research and development on machine learning and artificial intelligence. Like other researchers, QOMPLX's team is using Blizzard’s StarCraft real time strategy (RTS) game for experimentation and concept validation in areas like reinforcement learning, multi-agent reinforcement learning techniques, and select experiences in applied research supporting these improved decision-making goals.

In our latest installment in this series, we take a look at the problem of pathfinding in StarCraft. As we've noted, StarCraft is a strategy game that presents particular challenges to applying machine learning because it has only partial observability. In contrast to a game like Chess or GO, in other words, StarCraft does not enable players to see the entire play space - at least initially. Instead, large parts of the play space are obscured by what is referred to as the "Fog of War." AI that wants to succeed at StarCraft must learn to navigate this obscured terrain efficiently. That's where "pathfinding" and pathfinding algorithms come in.

Pathfinding is not a new topic, and various algorithms have been developed to tackle the problem. Yet there are still significant challenges and the Fog of War which is usually present in RTS games poses another layer of challenge as features such as terrain obstacles, (static) defensive buildings, and (moving) enemy units need to be learned effectively.

At QOMPLX, we have been working on ways to tackle these challenges using cutting-edge machine learning and artificial intelligence (AI) technologies. In our latest installment in this QOMPLX Intelligence series, we are discussing the pathfinding algorithms themselves and some of the different approaches these algorithms take to pathfinding problems. We also weigh their relative strengths and weaknesses.

Use the download button below or the following link to download and read StarCraft: Pathfinding in the Fog in PDF format.


Additional Reading

Part 1: Build Order Selection in StarCraft

Part 2: Opponent Strategy Identification Using StarCraft  

You might also be interested in

Turning the data you have into the data you want

Turning the data you have into the data you want

Having data is no great challenge. Most insurance companies have plenty, and while they could perhaps use more–most could–it needs to be usable otherwise it serves no value.

Read more
Creating value through insurance data infrastructure

Creating value through insurance data infrastructure

Interoperable solutions backed by a flexible data fabric are the key to digital transformation and multi-system interoperability. In the blog series “Creating Value Through Insurance Data Infrastructure”, we will explore the problems, decisions, and solutions to this challenge.

Read more
From Data to Intelligence: Systems Alchemy for the Insurance Sector

From Data to Intelligence: Systems Alchemy for the Insurance Sector

Major amounts of data live within insurance carriers but the challenge lies in getting it out in useful form. Learn how to extract the value from data without the need to replace your existing systems, spend thousands of hours coding or rekeying data, or commit millions to a new data architecture.

Read more
Request a Demo

Interested in learning more?

Subscribe today to stay informed and get regular updates from QOMPLX.