Timeline for How do I create an AI controller for Pacman?
Current License: CC BY-SA 4.0
17 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jun 23, 2022 at 6:22 | vote | accept | R3sist | ||
| Jun 22, 2022 at 14:13 | history | edited | hal9000 | CC BY-SA 4.0 |
added 11 characters in body
|
| Jun 22, 2022 at 14:02 | history | edited | hal9000 | CC BY-SA 4.0 |
added 910 characters in body
|
| Jun 22, 2022 at 13:09 | comment | added | Neil Slater | @R3sist I think some of the confusion is that although the term "game AI" is often used for enemies in video games, and they may even use some old school AI such as pathfinding via A* search, many experts here probably won't even register that as being AI (despite the name) | |
| Jun 22, 2022 at 11:58 | comment | added | R3sist | Hello there. I’m sorry the question is confusing, I’ll try to make it more clear. Like @hal9000 said, simple AI is deeply integrated with the game at hand to reduce the dimensionality. I want to create an AI , which can look at the pixel values or “something” else, some other dataype, just like a human does. I was asking resources around that. How can I create a controller - which runs on a separate thread - learn by taking in pixel values, and controls that game with that. Not by being a part of the “game” itself - like other simple AIs (like the enemies) | |
| Jun 21, 2022 at 19:54 | comment | added | hal9000 | "The OP has already looked into implementations of RL for pacman, so I assume he's not so unfamiliar with it, otherwise, they would not have understood anything in those implementations". I point him clearly to deep rl (what is different), as classical rl has problems with high dimensional spaces (input spaces) like pixels or other complex datatypes as they tend to be tabular methods. (because of that they are generally like how OP said, like any other AI in games (meaning simple AI) that are deeply integrated with the game at hand to reduce the dimensionality ) | |
| Jun 21, 2022 at 19:51 | comment | added | hal9000 | "The OP may be unfamiliar with RL, but that doesn't mean that the question is clear". Just that someone can't find the correct terms doesn't mean they are not clear. | |
| Jun 21, 2022 at 19:34 | comment | added | hal9000 | You're right about the "Since you want to learn directly from the pixels" I changed that to a more fitting sentence. | |
| Jun 21, 2022 at 19:28 | history | edited | hal9000 | CC BY-SA 4.0 |
added 33 characters in body
|
| Jun 21, 2022 at 19:27 | comment | added | nbro | Ok, I would edit your post then not to start it with "Since you want to learn directly from the pixels". To me, it seems that the OP is suggesting to use multi-threading for learning and interaction (that seems to be the crux of the post), but I am not sure because the post is not fully clear. | |
| Jun 21, 2022 at 19:26 | comment | added | hal9000 | It's clear that the controller thread is a separate thread/system interacting with the game in a complex way. (through observing pixels values or different (but the data type is complex not deeply integrated into the game)) | |
| Jun 21, 2022 at 19:24 | comment | added | hal9000 | I don't give an answer that takes only pixel values. I point him to deep rl with that he has his controller that can interact with openAI gym to play his pacman directly from pixel values or ram states or whatever he wants. | |
| Jun 21, 2022 at 19:20 | comment | added | nbro | The OP writes "taking in pixel values or something else.", so why are you so sure that the OP wants something that takes (only) pixel values? The OP may be unfamiliar with RL, but that doesn't mean that the question is clear. The title says "how to create an AI controller for pacman", which is different from "create something like a separated “controller” thread". The OP has already looked into implementations of RL for pacman, so I assume he's not so unfamiliar with it, otherwise, they would not have understood anything in those implementations. In my view, the post is just unclear. | |
| Jun 21, 2022 at 19:14 | comment | added | hal9000 | It's quite clear what he wants, he is just very unfamiliar with rl. He is dissatisfied with the idea of simple AI that follows some simple rules and wants some separate agent that looks at pixels (like the player) and also can only do the same actions as the player (while poorly formulated it's quite clear). So seeing that I point him into the right direction being deep rl textbooks (since he wants the AI to learn pacman from the pixels). | |
| Jun 21, 2022 at 18:57 | comment | added | nbro | I "advice" you to ask for clarifications before providing an answer because the question and post are currently unclear. | |
| Jun 21, 2022 at 13:45 | comment | added | hal9000 | (a,s,s',r) stands for action, state, next state, reward. | |
| Jun 21, 2022 at 13:42 | history | answered | hal9000 | CC BY-SA 4.0 |