Are you feeling good and in place? Or maybe you’re hot and bothered? Angry and upset? Or maybe sadness and depression? While all kinds of games exist for many different moods, it would be a good idea for a video game to adjust its difficulty depending on how the player is feeling. Because feeling constantly angry at the game might not be fun if that’s good for you.
Scientists in South Korea, at the Gwangju Institute of Science and Technology, have developed an even more surprising method for such a thing. The researchers developed a stress model that would adjust according to the players’ emotions and adjust it accordingly to ensure increased player satisfaction. Because who doesn’t want great satisfaction?
Game developers have always been aware of the balance needed when it comes to game difficulty and player progression, trying to find a sweet spot that isn’t too hard or too easy to make sure the game feels good. While settings are often changed, this often requires the player to adjust the settings. Korean scientists are saying something powerful.
Their model involves training dynamic difficulty adjustment (DDA) agents, using machine learning that has collected data from human players, which will adjust the difficulty of the game in order to increase one of the four different factors related to player satisfaction: difficulty, efficiency, flow, and valence.
The scientists used a fighting game for their model and trained their DDA agents, as human players played a fighting game against AI opponents, generating data for the agents, and people also had to answer questions about their knowledge. Using an algorithm called Monte-Carlo tree search, each DDA agent uses real game data and simulated data to play and adjust the opponent‘s AI combat behavior in a way that maximizes specific emotions or “affective state”.
Assistant professor Kyung-Joong Kim, who led the study, said the advantage of their method is that the player does not need to be monitored by external sensors to detect their emotions. “Once trained, our model can predict player worlds using just-in-game,” he said.
The study was small, using only 20 volunteers, but the team said the DDA agents created AIs that improved the players’ experience. However, fighting games provide a specific answer, so it begs the question of how it can be applied to other types of games, but the professor had an answer for this.
“Commercial sports companies already have a lot of player data. They can use this data to track players and solve various issues related to sports comparisons using our method,” Professor Kim said.
Their paper describes the example, “Diversifying dynamic difficulty to solve the agent by combining the player state model in Monte-Carlo tree search”, will be published in the journal Expert Systems With Applications on November 1. But for those interested, it is already available online and can be found here.
Next: Tekken Bloodline Needs More Courage
