Could a stealth video game like Assassin’s Creed 5 use the new BYU object recognition algorithm?
In a related report from The Inquisitr, the Assassin’s Creed 5 release date is apparently being set for 2014. Not much is known about the sequel except that Ubisoft creative director Jean Guesdon teased the next location could be in Japan:
“Feudal Japan remains – like numerous other time periods and locations – a possible future setting for the franchise because thanks to the Animus. The entire human history is our playground.”
And this playground can also expand the gameplay as the technology driving the video games expands. While everyone is focused on how the Xbox One and PS4 can increase the graphics detail and resolution, the much faster CPU and the vast increase in system memory also offers up gameplay potential that has yet to be explored. The obvious addition is more physics effects, but AI could potentially take a leap forward, as well.
Years ago, I recall playing heavily modified versions of Quake and Descent that added neural network-based learning. The Non Player Characters (NPCs) could progressively become more and more difficult as they learned the player’s strategies. Unfortunately, while this tech is neat it’s really only useful if the NPCs have a shared knowledge pool and this defeats any sense of realism since enemies you’ve previously killed pass on their advice to future combatants. So unless you’re playing a game where the story focuses on an alien hive mind this tech doesn’t make much sense.
But have you ever hated ever hated how AI can “see” you and shoot you accurately through bushes, grass, etc. when you can’t even see them at all? BYU engineer Dah-Jye Lee released an object recognition algorithm that can accurately learn to update its own pattern recognition. Without getting into detail, this algorithm is capable of recognizing objects in images and video with 100 percent accuracy while others tend to do worse:
“In most cases, people are in charge of deciding what features to focus on and they then write the algorithm based off that. With our algorithm, we give it a set of images and let the computer decide which features are important.”
After reading up on it a bit, I’m just curious if this tech could be implemented in games like Assassin’s Creed 5. Never mind the Kinect, computer AI tends to detect objects, including the player, via a simple cone or ray check. In the video game world NPCs technically always “know” you are there in their closed world but programmers are tasked with dumbing them down to meet player expectations. While stealth games like Splinter Cell and Assassin’s Creed have extra parameters to determine visibility, most games still come off as pretty fake. It’s possible a newer stealth game could use this algorithm by rendering a simplified black and white image for the AI to analyze to determine if it’s “seeing” things.
I would assume games could use this in conjunction with cone and ray checks along with other reasonable limitations to humanize the AI. For example, in a stealth game only if suspicion has been triggered would this algorithm even be used at all when a NPC “focuses” on an object once at a time. This would be done for realism and to keep the CPU and GPU load from getting stupidly high since this algorithm requires rendering from the AI’s point of view. The Xbox One and PS4 are powerful… just not that powerful.
The algorithm would be most useful for when shadows, fences, and other objects obscure the player. For example, a ray check could determine the player is standing behind or in a bush and then use this algorithm to determine if there is a visible shape (head, limb, gun) the AI can see. This algorithm would also prevent scenarios where the player is “in” a shadow, and invisible to the AI, but still has a lit background behind them which should be outlining the player clearly. Another big gameplay change would be that clothing and camouflage would actually make a difference.
Now do I actually see this being implemented in Assassin’s Creed 5? While I’d like to see it happen I sincerely doubt it since game developers will be producing multi-platform games that will be bridging the generational gap for at least two years. The Xbox 360 and PS3 probably couldn’t handle it so we’re looking at Assassin’s Creed 6, or even Assassin’s Creed 7, at the earliest assuming Ubisoft decides to do something like this. At the earliest we might expect an enterprising game development studio to try out tech like this in a next gen only game. Of course, in the end there’s also the fact that the assassin’s are like a spotlight in the dark with those white cloaks…