Fun find. I've started writing my own Utility AI reasoner (behavior trees for the actual AI, but my own utility decision system) and got the same conclusion: Curves. A simple way to create complex and intelligent behavior.
I've been searching for this topic over months! And I couldn't find a single code along tutorial about this topic! Thanks youtube for not showing these series when i search the exact keywords. Subbed and will watch these in a burst. Thanks in advance!
This is excellent, thank you so much to contributing to our industry. I am attempting to make truly NextGen experiences that I always dreamed about as a kid, and without these random videos, blogposts, comments etc. i dont know if I could do it. This also motivates me to eventually make videos like this of my own. Thank you.
This is great! Just what is needed. You are doing great! Keep going.Your style is just perfect! Thank you so much for doing this! You are providing a wonderful learning experience to the AI community.
I already came up with my own idea how to create a form of utility AI -- and was then disappointed to find my idea for a "novel AI system" wasn't actually a new idea and even already had a name. I'm still looking forward to setting up and trying out my Utility AI in a life simulator.
Thanks for all you have taught me through your RUclips channel. But, please I will like you to put me through on ways to go about some new things in the utility AI series.... I want a system where the AI must be assigned to a building before it can do work Some Illustrations: 1. AI assigned to a logging camp must only cut down wood when gathering resources. 2. An AI assigned to storage will do work by get resources stored in other building and place them at the main storage building. 3. An AI not assigned to any buildings will only randomly wonder around, sleep and eat but can't work. Please I would really appreciate it if you could help me with all this suggestions.
Hello! Cool video and I'm looking forward for the next parts! I just wish you had better audio quality on the black board part. Besides that it was very nice to watch and listen :)
Thanks for watching! That should be the only blackboard section in this tutorial series. So sorry haha. I'll work on improving the quality of that style of presentation next time.
I like the Khan Academy style blackboard! Just a note for future videos. Since you said you'll be delving into the code later on, please make sure thr font is enlarged or you zoom in. A lot of tutorials become harder to follow simply because they record the entire screen.
That is true. Makes alot easier for the viewers, even if for the creator is more pleasing to see it small. So far i'm enjoying his video, because TooLoo is not going fast in the teaching. cheers
Hello ST (From LostRelicGames discord), i enjoyed the concept of AI implementation. Subbed, liked and waitting for the next video. Do you have a roadmap or schedule for the part 2? Edit: With only this part 1 i am able to implement this concept. Thank you for sharing this! Cheers
Wow! I'm glad the conceptual explanation was clear enough for you to already implement it. I'll be releasing part 2 next week when I'm done editing. Stay tuned!
@@tooloo3658 The bell is set to notify me already! Hopes up to compare. So far i'm making ScriptableObject of the Actions and the Considerations. The brain have a reference of all the Controllers in order to have all the information the for the calculation of the score. Now i'm gonna wait for your video. I think i can make this concept using ECS. But i have to create a mockup first. Cheers!
@@tooloo3658 I was thinking about the concept of the "action", because if i want to make them generic to the result be re-used in other actions the "Action" itself must have another list of "execution" items. For example a spaceship game. If we think about it, there is some actions that must be combined because in one state it can just move and not attack, another it can move and attack at the same time. As well as standby and attack. There is 2 actions but diferent 3 states where the actions must be a generic action for a state if we think of the state as the "Action" you describe in the video. Unless i determine that some actions belong to certain a group. This way one group of actions cannot override the decision of another group. In one group (A) i might have: "Idle, Move, Flee and Follow", the other group (B) i might have: "None, charge, fire, shield, missile". Therefore i can combine execution of actions of diferent groups together. - Idle, None - Move, None - Move, Fire - Idle, Missile May i ask you whether can you advice me if this is a good approach or am i overthinking?
@@pliniomourao I haven't figured out how to implement parallel execution of actions yet. For your case of the spaceship though, a hack you can try is to separate the spaceship into 2 pieces: 1) the weapon and 2) the spaceship body. Each can be controlled by their own "AIBrain" - the spaceship AIBrain will be only responsible for movement decisions while the weapon AIBrain will be responsible for shooting decisions.
@@tooloo3658 ControlBrain and WeaponBrain. I like your thought. thank you so much i'm gonna try it out. Seems simplier than parallel execuction. Cheers
Fun find. I've started writing my own Utility AI reasoner (behavior trees for the actual AI, but my own utility decision system) and got the same conclusion: Curves. A simple way to create complex and intelligent behavior.
I've been searching for this topic over months! And I couldn't find a single code along tutorial about this topic! Thanks youtube for not showing these series when i search the exact keywords. Subbed and will watch these in a burst. Thanks in advance!
You're welcome! Hope the series can help with your work!
This is excellent, thank you so much to contributing to our industry. I am attempting to make truly NextGen experiences that I always dreamed about as a kid, and without these random videos, blogposts, comments etc. i dont know if I could do it. This also motivates me to eventually make videos like this of my own. Thank you.
This is great! Just what is needed. You are doing great! Keep going.Your style is just perfect! Thank you so much for doing this! You are providing a wonderful learning experience to the AI community.
I am going through the 6th video in the series. This is a great system. Please keep making videos!
i cant believe this video doesnt have more views
just found this now wow this is right what i need
The way you explained it is so easy to understand, Thanks !! just subscribed
I already came up with my own idea how to create a form of utility AI -- and was then disappointed to find my idea for a "novel AI system" wasn't actually a new idea and even already had a name. I'm still looking forward to setting up and trying out my Utility AI in a life simulator.
sad, had the same experience when found out about Wave Function Collapse
Thank you from pro game developer. for implementation details
Great video. It can be tough to find information about designing NPC AI, but this is very helpful.
Do you have any resources for design patterns of ai systems? It is really easy to get high coupling in ai system design which i don't really like.
Thank you! Interesting material! But I still can't watch to the end, because every time I want to start kenshi :)
Very well written, thanks for the video mate
Very good explanation for me =)
Kenshi!!!!!!
Incredible tutorial!
Thanks for all you have taught me through your RUclips channel. But, please I will like you to put me through on ways to go about some new things in the utility AI series....
I want a system where the AI must be assigned to a building before it can do work
Some Illustrations:
1. AI assigned to a logging camp must only cut down wood when gathering resources.
2. An AI assigned to storage will do work by get resources stored in other building and place them at the main storage building.
3. An AI not assigned to any buildings will only randomly wonder around, sleep and eat but can't work.
Please I would really appreciate it if you could help me with all this suggestions.
Is the game in the background yours? The world kinda remindes me of Kenshi
Hello! Cool video and I'm looking forward for the next parts! I just wish you had better audio quality on the black board part.
Besides that it was very nice to watch and listen :)
Thanks for watching! That should be the only blackboard section in this tutorial series. So sorry haha. I'll work on improving the quality of that style of presentation next time.
TooLoo is a weird name, guess I have to sub
Subbed
I've always wondered how people,developers and programmers learn these thing. Like if someone told them or they just search through files or something
sounds like discrete state machine
I like the Khan Academy style blackboard! Just a note for future videos. Since you said you'll be delving into the code later on, please make sure thr font is enlarged or you zoom in.
A lot of tutorials become harder to follow simply because they record the entire screen.
Thanks! I'll definitely keep that in mind for the next videos.
That is true. Makes alot easier for the viewers, even if for the creator is more pleasing to see it small.
So far i'm enjoying his video, because TooLoo is not going fast in the teaching.
cheers
Hello ST (From LostRelicGames discord), i enjoyed the concept of AI implementation. Subbed, liked and waitting for the next video.
Do you have a roadmap or schedule for the part 2?
Edit: With only this part 1 i am able to implement this concept. Thank you for sharing this!
Cheers
Wow! I'm glad the conceptual explanation was clear enough for you to already implement it. I'll be releasing part 2 next week when I'm done editing. Stay tuned!
@@tooloo3658 The bell is set to notify me already! Hopes up to compare.
So far i'm making ScriptableObject of the Actions and the Considerations. The brain have a reference of all the Controllers in order to have all the information the for the calculation of the score.
Now i'm gonna wait for your video. I think i can make this concept using ECS. But i have to create a mockup first.
Cheers!
@@tooloo3658 I was thinking about the concept of the "action", because if i want to make them generic to the result be re-used in other actions the "Action" itself must have another list of "execution" items.
For example a spaceship game. If we think about it, there is some actions that must be combined because in one state it can just move and not attack, another it can move and attack at the same time. As well as standby and attack. There is 2 actions but diferent 3 states where the actions must be a generic action for a state if we think of the state as the "Action" you describe in the video.
Unless i determine that some actions belong to certain a group. This way one group of actions cannot override the decision of another group. In one group (A) i might have: "Idle, Move, Flee and Follow", the other group (B) i might have: "None, charge, fire, shield, missile". Therefore i can combine execution of actions of diferent groups together.
- Idle, None
- Move, None
- Move, Fire
- Idle, Missile
May i ask you whether can you advice me if this is a good approach or am i overthinking?
@@pliniomourao I haven't figured out how to implement parallel execution of actions yet. For your case of the spaceship though, a hack you can try is to separate the spaceship into 2 pieces: 1) the weapon and 2) the spaceship body. Each can be controlled by their own "AIBrain" - the spaceship AIBrain will be only responsible for movement decisions while the weapon AIBrain will be responsible for shooting decisions.
@@tooloo3658 ControlBrain and WeaponBrain.
I like your thought. thank you so much i'm gonna try it out. Seems simplier than parallel execuction.
Cheers