Thought I'd share my AI trello for the summer

Hey all. Been having some health issues that have put me off my pace with tutorial videos and the like. Now we’re into the final few weeks of the University schedule before summer break, so things are crazy busy.

But I figured I’d post my trello plan which I’m working on for my summer AI work with Unreal Engine. Most all of this work will either be open sourced or released via the marketplace (with full source code either way).

So to cut a long story short, most of my plans for this summer are involving a lot of AI development work with Unreal Engine. I should get a fair chunk of the trello stuff done (a lot of it I’ve done before, so its mainly just redoing it for UE usage). You can see its biased towards tactical FPS needs. But actually some of the categories of work I’ve laid out are in truth the same (a digital actor is using essentially the same AI techniques as a squad member does, but conceptually I wanted to think of them separately for now).

I know, boring without videos and sparkly lights and stuff to show. But I’m a programmer dammit! Next time I promise to show a box near another box to show off the AI in progress :slight_smile:

Still got plans to get the next series of tutorials finalized, but won’t have time until the end of this week (its actually exam week this week).

Impressive list! If you find time between the coding to record or post any work in progress/design thoughts regarding what you’re working on, I’ll definitely be following.

Will definitely post what I’ve been working on when I get something worth talking about :wink: btw I’ll check out your util architecture in the next couple weeks too.

I have a functional best first or weighted random utility selector that I’ve been meaning to make a PR for. If you wouldn’t mind I could share it with you to get your feedback. I’m very interested in seeing where you go with most of the items on the board.

Sure, I’d be happy to have a look. I’m still in exam week and marking, so won’t be able to put much time into it until the end of this week at the earliest. But if you’re not in need of urgent feedback, I’d be happy to.

Interested in seeing work related to EQS.
I hope to use it to create contextual actions for the player and remove the need for complex moves or traces for others.
e.g. No need to check weapon collision if there is nothing near the player that could block the weapon.

I’m loving your work, dude. :slight_smile:

Hmm, for weapon collision I’d probably just stick with the raycast to be honest Kris. The physics engine will be doing some broadphase early-out tests anyway, so I doubt EQS would handle anything faster in that sense. EQS is really ideal for anything REALLY expensive, so that the cost can be amortized over multiple frames in an async query. So stuff like cover selection and target selection and the like, where each choice might need LOS tests and whatnot.

Why thank you sir!