AI Offers Video Game Design Possibilities Far Beyond Virtual Reality

Intro to Gaming Industry

It has already been more than 60 years since the first video game was invented, and thanks to tremendous improvements in hardware capacity and innovations in game design, today’s players have countless excellent options across countless game categories. The video game industry was worth US$139 billion in 2018, with a projected annual growth rate of 12 percent through 2025. As visual quality and gameplay becomes increasingly rich and sophisticated, leading video game companies are accelerating their investments in machine learning to take their games to the next level.

Advanced computer vision technology is supercharging virtual and augmented reality, one of the latest milestones in video game design. Other AI technologies are enabling powerful enhancements not only in the development processes, for example with animation generation and intelligence enhancement of non-player characters (NPC), but also to implement breakthrough features such as infinite maps and character customizations.

How Machine Learning Helps Game Developers Express Creativity

• Unity Engine is one of the most well-known cross-platform game engine for designing video games. In 2018, Unity3D introduced an open-source plugin “ML Agents Toolkit” that enables game developers to use machine learning algorithms such as deep reinforcement learning through Python APIs to train intelligent agents. These agents can be used to enhance NPC controls to implement more interactive and challenging gaming experiences. Unity also holds a community challenge with its ML Agents and has multiple experimental projects on exhibition.

• Ubisoft has also experimented with the possibility of using neural networks to automatically generate character animation. The method was first proposed by a researcher at Ubisoft Montreal and is designed to output accurate animations based on complex player controls and changing scenes. The reinforcement learning network takes player controls, scene geometry and previous character motion as inputs and produces animation based on current control orders and character position.

How Machine Learning Powered Features Are Reforming the Gaming Experience

• No Man’s Sky is a science-fiction survival and exploration game in which users visit fictitious planets with varying landscapes and creatures. To mimic the endlessness of the real cosmos, the game deploys machine learning algorithms to generate new unique maps while the player is exploring. This feature not only saves time for game programmers and artists, but also enables players to literally have an infinite universe to discover.

• Designed by the Fu Xi AI Lab of the Chinese video game firm Netease, Ni Shui Han is an online Chinese traditional-style desktop game that was released in 2018. The developers brought computer vision techniques such as face recognition and features detection into the game, implementing a feature that allows players to control virtual doppelganger characters by simply uploading a selfie. Players can also upload two human face pictures and the algorithm will generate an animated character that has features of both.

Future Trends and Obstacles

As we have entered a new decade and major console producers are scheduling next-generation product upgrades, a new era of video gaming awaits. Although the full extent of AI integration in the industry remains to be seen, there is no doubt that machine learning algorithms will play an increasingly important role in interactive entertainment like video games.

Opportunities however often come with obstacles. Most popular 3D games already require intensive compute and memory, and machine learning algorithms will only increase the demand. Running such games simultaneously with machine learning can be expected to increase players’ computer hardware expenses. Also, the available data on player behaviour is still scarce, which makes it difficult to develop highly individualized and engaging games.

Every video game imitates a complex imaginary world. Just as AI technologies are revolutionizing nearly all industries in the real world, there are endless possibilities to explore in the chemistry between AI and game design. In the next few years, we may see games that can detect the players’ actions, skill levels, or even mental and emotional states to create a more responsive gameplay experience. We may also communicate directly with NPCs through human or body languages rather than using keyboards or controllers.

---

Author: Ziyang Lin | Editor: Michael Sarazen