- 9
- 1 894 786
Linesight
เข้าร่วมเมื่อ 4 พ.ค. 2023
We train an Artificial Intelligence with Reinforcement Learning to play the game Trackmania Nations Forever, and post videos showcasing the progressive improvement of our A.I.
This channel is a collaboration between pb4 (github.com/pb4git) and Agade (github.com/Agade09).
You can contact us at the address pb4videos (at) gmail.com, via our github, or on Discord (server: discord.gg/tD4rarRYpj and channel: discord.com/channels/847108820479770686/1150816026028675133).
This channel is a collaboration between pb4 (github.com/pb4git) and Agade (github.com/Agade09).
You can contact us at the address pb4videos (at) gmail.com, via our github, or on Discord (server: discord.gg/tD4rarRYpj and channel: discord.com/channels/847108820479770686/1150816026028675133).
I Trained an AI for 2 Years on Trackmania. It's Breaking Records.
I trained an AI that plays Trackmania with reinforcement learning. It's getting good enough to challenge Official World Records. This is the story of how my AI reached such an incredible level.
Link to raw run playlist: th-cam.com/play/PLZ-JKrjYnD1GBlp-WS3Mr8bKmqsqSSO-U.html
Link to replay pack: drive.google.com/file/d/19p10V3g4RfFUoUUdVWNN2RXwJmcXfHC2/view?usp=drive_link
Link to input files: drive.google.com/file/d/12Jfp6nDNhAb7vQIeUIao_VU0W8652s_r/view?usp=drive_link
Project Github: github.com/Linesight-RL/linesight
Game: Trackmania Nations Forever
Buymeacoffee: buymeacoffee.com/linesight
00:00 Intro
00:43 map5 (v1 and v2)
03:58 A01
05:10 map5 (v3)
07:27 Official Campaign intro
08:42 A02
11:11 D06
14:22 Official Campaign
15:59 vs TAS
17:14 Minitrial
22:04 Method
23:14 Zero-shot
24:03 Next steps
25:38 Finale - Hockolicious
27:04 Credits & Times
#AI #reinforcement #learning #trackmania #world #record #official #campaign #tmnf
Music Tracks:
Luge - Geographer
Instant Crush - Corbyn Kites
Subway Dreams - Dan Henig
Hall of the Mountain King - Kevin McLeod
I got a stick - Kevin McLeod
Envision
Arp Bounce - Geographer
Up the steps - Geographer
Shitsubo Shita - Josh Lis
Just Breathing - NEFFEX
Tinker Time - Nathan Moore
Contact us:
First join the TMInterface Discord discord.gg/PvWYGkGKqd, then head over to the #programming-talk channel discord.com/channels/847108820479770686/885481860539498508
This project would not have been possible without the awesome TMInterface tool, provided by @Donadigo.
Link to raw run playlist: th-cam.com/play/PLZ-JKrjYnD1GBlp-WS3Mr8bKmqsqSSO-U.html
Link to replay pack: drive.google.com/file/d/19p10V3g4RfFUoUUdVWNN2RXwJmcXfHC2/view?usp=drive_link
Link to input files: drive.google.com/file/d/12Jfp6nDNhAb7vQIeUIao_VU0W8652s_r/view?usp=drive_link
Project Github: github.com/Linesight-RL/linesight
Game: Trackmania Nations Forever
Buymeacoffee: buymeacoffee.com/linesight
00:00 Intro
00:43 map5 (v1 and v2)
03:58 A01
05:10 map5 (v3)
07:27 Official Campaign intro
08:42 A02
11:11 D06
14:22 Official Campaign
15:59 vs TAS
17:14 Minitrial
22:04 Method
23:14 Zero-shot
24:03 Next steps
25:38 Finale - Hockolicious
27:04 Credits & Times
#AI #reinforcement #learning #trackmania #world #record #official #campaign #tmnf
Music Tracks:
Luge - Geographer
Instant Crush - Corbyn Kites
Subway Dreams - Dan Henig
Hall of the Mountain King - Kevin McLeod
I got a stick - Kevin McLeod
Envision
Arp Bounce - Geographer
Up the steps - Geographer
Shitsubo Shita - Josh Lis
Just Breathing - NEFFEX
Tinker Time - Nathan Moore
Contact us:
First join the TMInterface Discord discord.gg/PvWYGkGKqd, then head over to the #programming-talk channel discord.com/channels/847108820479770686/885481860539498508
This project would not have been possible without the awesome TMInterface tool, provided by @Donadigo.
มุมมอง: 1 495 153
วีดีโอ
Trackmania AI Learns To Drift and Beat Pros ? | Hockolicious
มุมมอง 351Kปีที่แล้ว
We trained an AI to play Trackmania on Hockolicious, one of the game's most prestigious tracks. The AI set a strong time, beating the 2012 World Record previously set by CarlJr. Link to map: tmnf.exchange/trackshow/414041 Link to replay: drive.google.com/file/d/1b0p2QSXAea1v7lVCGuWkNr-5xQ6PdqHo/view?usp=drive_link Game: Trackmania Nations Forever Project Github: github.com/Linesight-RL/linesigh...
Superhuman Trackmania AI Demo | map5
มุมมอง 29Kปีที่แล้ว
Linesight project. We used reinforcement learning to train an AI that plays Trackmania. On its training track, the AI is faster than the current human world record. We put this video out as a challenge: can a human still overcome our AI? Game: Trackmania Nations Forever (TMNF) Map: tmnf.exchange/trackshow/10460245 Replay (.gbx file): drive.google.com/file/d/1jfOyhRQCvOC5XSEhhajE2OvmJcCHKDjR Wir...
AI Plays Trackmania - Bloopers
มุมมอง 3.7Kปีที่แล้ว
In this video, and AI is trained with reinforcement learning to accumulate speed and finish a map as fast as possible. The AI learned a behavior where it turns around right before the finish line. This is not a one-off mistake, the AI repeatedly did similar things in back-to-back runs. Can you guess why ?
AI Plays Trackmania - Map5 2:04:91
มุมมอง 9Kปีที่แล้ว
The AI is trained via reinforcement learning. Game: Trackmania Nations Forever (TMNF) Map: tmnf.exchange/trackshow/10460245 Replay (.gbx file): drive.google.com/file/d/1hp1Mz0ooR2YBNpqNjvvxjFrBOGXs8DYD/view?usp=sharing
AI Plays Trackmania - Training Progression Side by Side
มุมมอง 3.4Kปีที่แล้ว
In this video, an AI is trained via reinforcement learning. In order from the top left corner, top right corner, bottom left corner and bottom right corner the AI has received progressively less training time. The video compares the lines taken by the various AIs in different parts of the map. Game: Trackmania Nations Forever (TMNF) Map: tmnf.exchange/trackshow/10460245
AI Plays Trackmania - Map5 2:07:00
มุมมอง 1.2Kปีที่แล้ว
The AI is trained via reinforcement learning. Game: Trackmania Nations Forever (TMNF) Map: tmnf.exchange/trackshow/10460245
AI plays Trackmania - Map5 2:09:12
มุมมอง 1.3Kปีที่แล้ว
The AI is trained via reinforcement learning. Game: Trackmania Nations Forever (TMNF) Map: tmnf.exchange/trackshow/10460245
(Teaser 01) AI learns to play Trackmania with reinforcement learning
มุมมอง 2.4Kปีที่แล้ว
This video is a recording of an AI currently training to finish a custom map as fast as possible with reinforcement learning. This run was played on Trackmania Nations Forever, with TMInterface to link the game and our AI. The run was played on a custom map. The .Gbx map file is available for download here : tmnf.exchange/trackshow/10460245 The AI took 2mn12s25 to complete the map during that t...
This proves that a fruit fly is physically capable of beating humans at trackmania
TH-cam ads are getting too much. I’m really looking forward to the day where this app is gone.
I wonder what do you mean "Mid Range Computer"? What generation of cpu and gpu do you train the ai on?
5700X / RTX 4070ti
u r swag surfin my friend
I think to make training faster I would do a pre-training to make a first model adapted a bit to every condition, but not specialized on a map. then take this model as a base for fine-tuning on the map. If the pre-trained model become so good it can generalize on all map that would be great.
I feel like this way of teaching AI means they can only learn from memory. They have no instinct. AI drivers will never replace human skill in a 'first time run' challenge.
Fun video, very well put together for a human. Good work mate
what are you talking about? Even tool assisted runs are closer to human runs in every way 🤣 Still cool though but that take is beyond nuts.
I was here
now make it noseboost
My man thought his code the Shadoe Clone Jutsu
19:00 is not being close the given path is kind of hack?
All AI should be banned.
Amazing work!
anybody know the name of the song at 25:40?
Shitsubo Shita, by Josh Lis
So you need to make a line for every map for the AI to play them ? Or it's just for the training step, and we can then try to use it on brand new map ?
Why do you need machine learning to solve a track? This should be straight forward math problem, the route should be well defined ahead of time.
I hope you aren’t actually taking WR titles with a bot. That would be really stupid and rude.
Elon better watch out... Next self driving update we got Teslas hitting uberbugs on I-90 West
I've never played Trackmania, and for some reason it's all over my TH-cam Channel
If this also works on the new trackmania, you HAVE to train it for the new Platform mode :)
Your videos and those of other greats like Code Bullet have made the concept more clear to me, to the point that I now want to train an AI to make construction drawings of existing architectural models.
AI 24 hours - humanity 10 years.... bro AI took all the informations from 10 years humanity
is this better than Yosh's?
What are your pc specs?
Have you Heard of Nitro circus? The X-Games ? humans jumping motorcycles in to foam pits, is where Travis Pastrana learned to and taught others to do a backflip on a motorcycle. What can AI achieve?
This is impressive work. There's a lot of lessons to be learned here.
mate forget the ai, show us how you make these videos!
pinned comment has exactly 621 replies 😏
gonna be wild when it comes out that there is no "ai", it's just you. you're like the next stig.
I'm just bored of people trying to automate their recreation. And I really don't see a useful future for the automation models, except for malicious applications. Let me download your program, train it, then enter some cash prize tournaments.
They'll never take A-01 from us🙏
Awesome vid ! Fascinating.
I'm a profane so i may be saying a bs. But to make AI learn shortcuts why not try and train another AI to adjust the gamma in real time? Maybe taking more risk will help it find more alternatives.
And so now what? You drop this stupid thing into every competitive racing space and make everyones hard work and fun irrelevant. Wow good job, so cool. All AI is ever practically used for is to lie.
Humanity will always be at a disadvantage in the digital world vs ai. AI cant even come close to matching us in our world.
Have you considered using some form of pretraining/transfer learning so that the model already knows how to drive generally and then trains specifically on that track?
My lag be like: 16:16
Hm... How about using AI to find the best track first and the second model to run close to the best track? I mean finding track by other rules like by maximizing kinetic energy or something. It seems to be really challenging since it is a whole new topic to cover but have potential to severely improve algorithm and make it more human independent.
Ai breaking human records of a decade in a day is truely impressive. Lets see how huge this thing goes
I love how the cars controlled by humans are colorful and the A.I. car is all black.
Have you allready thoght about first training it on the best runs, like with the input data and then letting it train on the map on it's own so that it like knows all the glitches and then can perfect em.
how about you compare your AI to the best TAS you can get
trackmania TAS is too insane. AI doesnt get anywhere close
i dont even know this game but god ths video is insanely good!
Open source A.I. for Track Mania..... Skynet is born....
@Blento0404otnelB is right, its not an A.I. People using this for marketing and it works, as always. No, machine learning is not an intelligence. these are just programs, nothing more. Funny how salty everyone gets and what fake facts they come up with just to not give in the fact they were wrong
amazing video, the best I got AI to do was La Joconde as a cat..
what a way to end it 👏👏
Human are doomed !
22:58 surely you can compress it a lot so it doesnt use that much