I had a big break from it - mainly because I was focusing on the group project (too much, perhaps) - but, it has come to the time when I should have started really digging in a month ago, so I'm still, as always, behind.
I started looking over my project again about 2 weeks ago - looking at areas where I could improve, and the implementation I have now is not very good.
The choices that leaders and people seem to make almost seem random because of the ridiculous number of factors I've implemented (usually only affecting one-off choices too...) - and so it doesn't really seem like there's much AI at all - there's no cohesion.
Similarly, many interactions within the game don't make sense, for example, when a natural disaster strikes, it cuts off Tech Trees at a random point in the city's investment - wiping out not only say, road infrastructure that makes sense, but... also potentially the civilisation's knowledge of making musical instruments.
Similarly the code is starting to get bloated with poor planning and one-off variables and functions that don't do much. It's getting hard to expand upon.
So I guess what I'm trying to say is that my current project is stupid because I rushed in, just trying to get the programming done - rather than properly designing the system and the data/choice interactions... which frankly, if you're coding AI - seems really obvious. My bad.
So, I'm going to be redesigning my AI system from the group up - paying particular attention to all the relationships and the effects of choices like I should have been. Similarly, I will design my rulesets now as well, ensuring that every variable in each AI entity has a significant purpose. If I want to code in intricacies, I can do that when I'm done.
Similarly I'm thinking of coding a program to create Behavioural Trees. For example, a simple drag and drop interface and such. While this might be a risky detour, it could potentially benefit me a lot in the long run.
So, see how that goes!
No comments:
Post a Comment