Hi everyone,
We wanted to give a bit of insight on the things we’ve been working on so this is going to be a lengthy post.
Over the past few weeks we’ve been working on our AI system. Upon working on it we’ve discovered that our old AI system was performance intensive and extremely limiting, which made it difficult for us to add new functionalities to the system. After having a few test runs its clear that we would need to overhaul our AI system so that new functionalities could be added.
So far, we’re happy to report that we’ve successfully overhauled our core AI system. With this newly overhauled core system, we now have a solid ground to stand on when building our AI.
At this moment, the AI has already been able to behave much like how it behaves normally in the game but with better performance and stability. Here are screenshots comparing the performance of our old AI system and new AI system.
A bit of side note before we get to the explanation of things, we’ve previously decided to record a video to compare the two systems, however, our screen recorder has given a significant impact to performance. Because of this, we’ve used screenshots instead so that we could show the actual numbers with less interference.
Both the old and new system were tested in our laptop with each been given an empty scene with 30 AIs simultaneously activated at Ultra quality settings. From the screenshots it is clear that the new system has greatly increased the performance of the game, making it jump from less than 10 FPS in our our old system to a whopping 150 FPS in our new system.
Although there are still room for improvement, the results given by our new system have made us confident enough to take this system forward and build functionalities for our AI. One of the new functionalities we’ve managed to integrate is the perception functionality. The perception functionality enables AIs to define if others are dangerous or not. Unlike basic behaviors where AIs have a blanket reaction towards their opponents, the perception functionality allows AIs to be more sophisticated and consider which opponent they should face, stay passive, or run away.
The video shows how the perception functionality plays out in the game. It can be seen from the video that the Triceratops AI reacts differently when faced against a Nanotyrannus and a Tyrannosaurus. This is because it perceives the Nanotyrannus to not be dangerous, hence it keeps on wandering, while the Tyrannosaurus to be dangerous, which makes it run away.
The video shown is only a sample of the perception functionality, as when coupled with basic behaviors and our population system, this functionality would become more complex.
Anyways, that’s currently it for this devlog, there’s still plenty of things we’d need to do to further flesh out this system and we will inform you when we’ve made further progress. More updates coming soon.