Abstract— A thrilling pass time in many lives, video games have been a source of entertainment since it was first released in the 1950s at a science fair. Video games are a place where creativity and art meet science and mathematics, which is clearly depicted through the usage of Artificial Intelligence. Its role is as important as narrative, art, graphics, audio, and any other element. One can not overlook the role of AI in making Non-Playable Characters (NPCs) more human-like, and how they make a game more immersive. This paper reviews the franchise of The Last of Us by Naughty Dog and discusses the usage of AI for decision-making by Non-Playable Characters in the game as well as in other fields.
Keywords— Artificial Intelligence, Video Games, Non-Playable Characters, Path Finding, FSM.
The inception of AI in video games- Artificial Intelligence plays an important role in video games
ever since its inception in ‘The Nimatron’, an ancient mathematical game. The game was
commercially released at the Westinghouse pavilion at the New York world fair in 1939. ‘Nim’ is a
game played by two players where each player takes a turn to remove some coins from a single heap.
There are two versions of ‘Nim’, one where the winner is the one to take the last coin and another
where the player must avoid being the former. “A sturdy chrome-edged cabinet topped with rows of
light bulbs representing the game tokens”, was built and named ‘The Nimatron’. The 8-foot tall
entertainment device stood as an attraction to anyone who wanted to defeat the machine. Almost as if
it had a mind of its own, ‘Nim’ made its decisions promptly and confidently, of which the players
were quite taken aback. Although ‘Nim’ might not have a lot to contribute to the Gaming Industry
directly, its development did bring the idea of humans going against mechanical components into the
limelight.
NPCs or Non-Playable Characters- NPCs have been a part of video games since the era of ‘Pong’.
They are computer-controlled bots or characters that interact with the player and give the illusion of
playing against or with other human players. These characters allow the game to provide an
open-ended experience to its players in addition to making the game more interactive. It removes the
restriction of having human players fill in additional roles in a story.
The Last of Us franchise is set in post-apocalyptic America, where about 60% of the population was
either infected by the deadly fungal infection Cordyceps or killed surviving from the infected
following an explosion that spread the fungus around the country around mid-September in 2013.
AI can be implemented in various ways in the game through NPCs. Starting with character dialogues
and the buddy trailing system to the behavioural integrity of the human enemies as well as the
infected. This paper will discuss these aspects in detail.
This section explains the repetitive AI models in depth for the sections followed later. The Last of Us
is a stealth-based game, and to ensure a smooth transition between the different states of NPC
awareness, the game uses Finite State Models or FSMs.
An FSM is a mathematical model used to describe the behaviour of a system by distributing the
system into a number of ‘states’ and the transitions between them. Each state in the FSM represents a
specific behaviour that the buddy AI or the enemy AI follows, such as following the playing,
searching for them or engaging in combat. The transitions are triggered by certain events or conditions
in the game. However, they can become repetitive if too simple and struggle with unforeseen
situations. To address this, developers can use advanced FSM techniques like hierarchies for complex
behaviors or even explore alternatives like Behavior Trees for more nuanced decision-making.
In The Last of Us, an FSM could control the buddy AI's behavior. It might have states for following
the player, being on alert, and assisting in combat, with transitions triggered by the player's actions
and the environment.
Download the
PDF↩︎ for rest of the paper.