UC5: AI for Games
Over the past decade, the game industry has undergone drastic changes. There has been an increase in online games, digital distribution platforms, mobile and social games. Digital games are now services that rely on continuous developing and adding new content over the lifetime of the product in a winner-takes-all market. Before launch, optimizing a game for the market is crucial for success.
AI can provide a radically new approach for testing and optimizing games by letting developers test thousands of different configurations and cases instead of a few hundred or less. It can replace or augment existing practices by providing product evaluations faster than current methods. Additionally, AI may even provide methods for generating content automatically. The game industry is seeing an increased need for automation and improved efficiency. Finally, AI has the capacity to provide new gaming interactions. For the past decade, new control devices have been appearing with the aim to replace the traditional gamepad and provide more natural interaction.
To address these issues, this use case aims to advance game testing, design and interaction through AI algorithmic innovations. It focuses on three topics: automated testing, improved music analysis and synthesis, and natural game interaction.
“Automated testing for Games” implicates the use of AI agents that can automatically detect bugs and glitches in video games by autonomously playing them and compiling test reports as human testers would do. Video game testing is a time-consuming process that traditionally has been entrusted to human testers. However, the future of the game market sees an increasing adoption of the game-as-a-service model, where new content has to be provided to players on a monthly basis. Therefore, quality assurance (QA) becomes a vital part of the game development cycle, as the new content produced has to be continuously tested.
“Improved music analysis and synthesis for Games” aims to help developers choose or conceive background music or sounds. In video games, the audio track has the role to strengthen the atmosphere of the present scene, and it has an impact on the emotion felt by the user. Developers can either choose music from a catalog or delegate this task to music composers or sound designers; but automatic AI tools may help, especially in the case of dynamical changes and procedural content generation.
“Natural game interaction through edge analysis of camera stream”: With cloud gaming, end users can play games without dedicated hardware and directly interact with games running on the cloud and streamed to their home. A Smart TV equipped with a standard RGB camera coupled with AI at the edge will replace the gamepad for natural game interactions. The camera stream will be analyzed with AI models that dynamically adapt to the available resources and offloaded to the edge. The output of the analysis (e.g. human body detection, gestures) will be interpreted into game commands within stringent latency requirements and ensure the privacy of end-users.
Discover more on this topic on the whitepaper on AI for Video Game Testing and Music Processing.
Watch the Demo-Video
