This week I've been playing Splinter Cell Blacklist and HITMAN again to be inspired in how to improve our AI. I ended up implementing what we're calling "Hierarchical Detection", which can be described as detection dependent on the disguise the player is using and the detecting NPC's rank. Imagine a scenario with a Guard - lower in the Military Hierarchy (defined by us) than any other rank - seeing the player, which is disguised as an Sargent or a General. In this case, the Guard NPC doesn't detect the player. Several improvements to this approach can be made, including lower ranking NPCs recognizing every NPC in the rank immediately above, and not all higher ranking NPCs recognizing all lower ranking ones.
After a few hours of enjoying Splinter Cell's movement system, I realized ours was terrible, and so, I used the rest of the day improving it, with one of the biggest additions being strafe: before this, our player character moved sideways in a way I can't really describe, only by "not enjoyable to any player that tested our game".
 |
| Fig. 1: AI Perception in Splinter Cell |
I also revamped the detection system, adding multiple detection "cones" to the NPCs, including a small area behind them, and "cones" that require certain conditions, such as a larger circle surrounding the player that detetcs running. We still don't have in-game sounds, so I didn't think it was a priority to have additional sound awareness, such as guards detecting doors opening, but it's something I want to do in the near future.
Sem comentários
Enviar um comentário