What's happening these days? What about the next update
[h2]What's with the updates[/h2]
The last time I spoke here, I promised to focus on developing a more fleshed out hearts and minds interaction for the game (more social projects, political leaders, more lasting consequences to how you interact with the villages). This hasn't changed, and will still remain the main task for the nearest future.
I have, however, reached a limitation in how the AI brain works. I initially made a somewhat simple system, where each actor senses the events around him, and tries to pick the most suitable action for his current mission. This works fine, but the problem is - the more complicated a mission is, the more unreadable the code that manages it becomes.
For example, if the enemy units are doing a foot patrol, their goal is to move from point to point, looking for anything interesting they might wanna check out. So far, so good. But, what if they should check out a village, instead of a random map point? What if they stumble upon and unguarded enemy car? What if they get enegaged in an unexpected place? What if their allies need help? What if they're in cars themselves? They then need to stop, exit them and continue the mission.
All these things are being tracked by the mission script, making it more and more hellish to manage and read. I cannot continue writing the AI this way, cause it completely discourages me from making any more complex interactions (whenever actors leave their cars, I need to remember why - is it the location they were going to, are they stopped in traffic, is it something else), and in general - if at any point the AI gets engaged by your troops - it has to abandon the original mission completely, because it has no way of remembering what it was trying to do.
The solution I'm currently working on, is switching to a more modern and robust way of writing this code - a node based system, similar to a behavior tree. The AI will now always remember at which stage of it's mission it is, and after I'm done writing the "nodes" for various interactions, I can start making really complicated plans for it.
The same system will be used for other parts of the game, starting with behavior in combat, to the in-game politics.
[h2]How will the AI be different when I'm done[/h2]
This is a sample view of an AI mission, where instead of a single monolith of unreadable code, I now have a structure I can track, and think of as "point to point" interactions for the AI. This lets me plan entire behaviors, like: 1) go to an area 2) exit cars 3) enter a building 4) if engaged, abort and return to this plan later 5) etc.
[h2]What does this mean for the game[/h2]
The AI will get smarter, and I will again start making new interactions for it. The first thing I'll do will be adding political leaders to the game. The leaders would be moving from village to village, changing the civilian sentiment in favor or against you, promoting their local allies, maybe organizing protests. As the player, you'll be getting new missions related to them, new demands to meet, and so on.
This sort of complicated interaction would be impossible to make with the old system.
[h2]What's coming next[/h2]
The next big update will add politics and more hearts and minds strategies for the player, like new social projects for you to develop, and more interesting interactions with the locals.
[h2]When is it coming[/h2]
With the summer season in full swing, I'll take two weeks off by the end of August (gonna be my first time getting any downtime since 2019). Before I leave, I plan on finishing the switch to the new system for all existing AI missions. This might not be immediately visible for the players (after all you rarely see what the AI is planning to do, and why), so I also plan on implementing smaller, unrelated changes as well. Expect new units, as well as some QOL improvements.
Planned release date - end of September.
The last time I spoke here, I promised to focus on developing a more fleshed out hearts and minds interaction for the game (more social projects, political leaders, more lasting consequences to how you interact with the villages). This hasn't changed, and will still remain the main task for the nearest future.
I have, however, reached a limitation in how the AI brain works. I initially made a somewhat simple system, where each actor senses the events around him, and tries to pick the most suitable action for his current mission. This works fine, but the problem is - the more complicated a mission is, the more unreadable the code that manages it becomes.
For example, if the enemy units are doing a foot patrol, their goal is to move from point to point, looking for anything interesting they might wanna check out. So far, so good. But, what if they should check out a village, instead of a random map point? What if they stumble upon and unguarded enemy car? What if they get enegaged in an unexpected place? What if their allies need help? What if they're in cars themselves? They then need to stop, exit them and continue the mission.
All these things are being tracked by the mission script, making it more and more hellish to manage and read. I cannot continue writing the AI this way, cause it completely discourages me from making any more complex interactions (whenever actors leave their cars, I need to remember why - is it the location they were going to, are they stopped in traffic, is it something else), and in general - if at any point the AI gets engaged by your troops - it has to abandon the original mission completely, because it has no way of remembering what it was trying to do.
The solution I'm currently working on, is switching to a more modern and robust way of writing this code - a node based system, similar to a behavior tree. The AI will now always remember at which stage of it's mission it is, and after I'm done writing the "nodes" for various interactions, I can start making really complicated plans for it.
The same system will be used for other parts of the game, starting with behavior in combat, to the in-game politics.
[h2]How will the AI be different when I'm done[/h2]
This is a sample view of an AI mission, where instead of a single monolith of unreadable code, I now have a structure I can track, and think of as "point to point" interactions for the AI. This lets me plan entire behaviors, like: 1) go to an area 2) exit cars 3) enter a building 4) if engaged, abort and return to this plan later 5) etc.
[h2]What does this mean for the game[/h2]
The AI will get smarter, and I will again start making new interactions for it. The first thing I'll do will be adding political leaders to the game. The leaders would be moving from village to village, changing the civilian sentiment in favor or against you, promoting their local allies, maybe organizing protests. As the player, you'll be getting new missions related to them, new demands to meet, and so on.
This sort of complicated interaction would be impossible to make with the old system.
[h2]What's coming next[/h2]
The next big update will add politics and more hearts and minds strategies for the player, like new social projects for you to develop, and more interesting interactions with the locals.
[h2]When is it coming[/h2]
With the summer season in full swing, I'll take two weeks off by the end of August (gonna be my first time getting any downtime since 2019). Before I leave, I plan on finishing the switch to the new system for all existing AI missions. This might not be immediately visible for the players (after all you rarely see what the AI is planning to do, and why), so I also plan on implementing smaller, unrelated changes as well. Expect new units, as well as some QOL improvements.
Planned release date - end of September.