The Power Of Machine Learning And VR Combined

VR and Machine Learning are usually seen as two distant worlds: while one is a purely architectural concept, the other is an established and progressing development and management tool that is currently probably the biggest focus within the tech industry. What is the possible correlation between these two then?

How Do They Intersect?

When applied to Virtual Reality, the basic machine learning process that may apply to the setup is related to calibration: for example, let's say that a new video game (developed exclusively for VR peripherals like Oculus Rift) needs ultra-precise calibration for whatever reason. In this case, having an algorithm that automatically controls the response of such inputs (height, sensibility etc.) would surely benefit the user experience and the overall quality of the game itself.

How is it possible then? In this particular scenario, the AI will automatically associate a certain height with determined movement paths that are following an algorithm. An advanced, more sophisticated AI, on the other hand, will develop a personalised movement path that will follow the user true sensibility and response, to guarantee a solid user experience.

What's SLAM?

Simultaneous Localization and Mapping is a VR application that tells the code what is the headset position and acts accordingly. Google Tango and Hololens (to reference a few) are the technologies that are currently using this application to track the headset position in a 3D map. Machine learning comes when the app has created many different version of the surrounding space, compiling and merging them into a solid, single projection that tells the exact position of the headset.

On Mobile

Mobile app development is currently at a stage in which AI, VR and Machine Learning are colliding into a solid instance: let's pick up the fact that Niantic has a new game in development that will be "The Harry Potter version of Pokemon GO". In this game, the AI is constantly mapping the surrounding area, using either SLAM and other scanning tools via camera, sensors, radars etc.

The current limit is the fact that the AI is limited to a standard in which it won't be able to damage the device's processor, ram etc, so it will likely be downgraded to a previous version of the most up to date IOS and Android.

The Game AI Itself

Machine learning has a big part when it comes to keeping the AI up in games that are VR focused: the AI, in fact, has to understand the player's movements, of course, but also the surrounding environment and how the player will react to it. The algorithm has to be responsive since the AI will have 4D duties. For now, the AI in VR games is pretty much basic, but we will see how it will develop in the future.