Online multiplayer shooter Rainbow Six Siege is working on an AI that can act like a real player and has shared details on how the system will work.
Ubisoft’s Rainbow Six Siege shared its year eight roadmap and hinted toward a new Defender AI playlist. This mode will help players practise against operators and test new strategies. Siege was released in 2015 and has gained a dedicated and experienced player base. However, it can be difficult for new players to get up to speed with its steep learning curve.
Since these new players would spend much of their time already eliminated from the match, learning or enjoying the game proved troublesome, despite the current AI-generated bots system that players can practise against.
AI operators
However, the Defenders playlist aims to help players to learn the intricacies of Rainbow Six Siege and its maps that offer various tactical approaches. The Defenders AI playlist will consist of AI-generated bots that can behave as real players. The current in-game bots are procedurally generated enemies and act more as targets to shoot rather than opponents with tactical thinking.
A major element that sets Rainbow Six Siege apart from other first-person shooter games is its approach to strategic gameplay and its operators, who possess different skills and gear. The current bots don’t utilise the maps the way a human player would and take on the form of a generic gun person rather than one of the game’s skilled operators.
The team behind Rainbow Six Siege is developing dual AI frameworks, which use both machine learning and traditional AI to produce new bots that behave closer to actual players. This means that players can go up against bots that no longer stand around waiting to shoot the first thing that moves but instead strategise and utilise the gear at their disposal.
Bringing the power of AI to games
Technical architect Alex Busby stated, “Our goal is for bots to be completely indistinguishable from players, and we want them to emulate the strategies and movements of a real player. The initial goal of the bots is to prepare newcomers for PvP and help them get accustomed to the environment that is playing Rainbow Six Siege.”
To create the new bots system, the Rainbow Six Siege team, along with researchers and AI innovators at Ubisoft La Forge, built parallel frameworks, one using a traditional AI framework and the other using machine learning. The long-term plan for the new bot feature is to use machine learning AI exclusively. However, traditional AI was needed to get the feature released quickly and without causing disruption to the game.
The team used the existing AI architecture in Siege but turbocharged it with match data generated from the game. This taught the AI with data and was then able to behave more like a real player without reducing code complexity. It’s important to point out that the data used to build the AI system is based on information that players have already authorised access to through its terms of use, and no personal data is included.
New systems
Machine learning frameworks take on large data sets, such as match replays and analyse them. The more the system processes this data, the better it can replicate human-like decisions.
Busby notes that, “One of the ways we’re doing this is, for all cases where a player has thrown a grenade, we look at all the data around the ‘world state’ when it was thrown. Then, we produce a module within machine learning to predict, in a given situation, if it’s a good time to throw a grenade.”
Interestingly, rather than the AI bot understanding that it has a grenade available and simply throwing it, the system is considering ‘if’ it should throw it. The bot thinks more like a human player, using strategy and holding onto the grenade for a better opportunity later. Not only does this create a more immersive gaming experience, but according to Busby, it also positively impacts the Siege team’s ongoing production, reducing developer workloads.
The team opted to use match replays to train the system, as with this method, all the team needed to do was collect a series of replays from the Rainbow Six Siege servers and feed them to the frameworks. This would allow the AI to analyse the data and pull desired behaviours.
Changing the face of gaming
When these new features launch, bots will be able to play as five defenders with future plans of adding new operators and maps to their repertoire. The way this will work is new operators will learn from two weeks of internal Ubisoft studios data gathered from playtesting. This means bots will be able to play as new operators right away. In addition, the framework will collect match replay data from live players and continue improving through its reinforced learning loop.
The new AI bots will help new players get to grips with the game’s maps and unique operator features. They will also grant more experienced players the ability to hone their skills as the AI system will have different difficulty levels. In the long run, the team aims to have offline training for the bots to determine if attackers or defenders have a higher probability of winning. This will grant the team a way to analyse data on how bot behaviour will impact rounds and adjust the framework to ensure the right balance is met.
Busby comments, “We want to set a stage for elite players to hone their skills before a big match. My personal, deep, deep ambition is, at the end of an esport event like the Six Invitational, to have the winners play against our elite bots and see who wins.”
AI allows game developers to utilise new tools to streamline the way they work and offer players a better gaming experience. A recent survey highlighted details on game NPCs and the majority noted they wished to see NPC characters gain better situational awareness. With the power of AI that possibility is on the horizon.
Paige Cook is a writer with a multi-media background. She has experience covering video games and technology and also has freelance experience in video editing, graphic design, and photography. Paige is a massive fan of the movie industry and loves a good TV show, if she is not watching something interesting then she's probably playing video games or buried in a good book. Her latest addiction is virtual photography and currently spends far too much time taking pretty pictures in games rather than actually finishing them.
