Infinite Mario AI - Long Level
Robin Baumgarten3 HN recommenders · recommended in 3 months, 2009–2012
demo · general · evergreen · 2009–2012 on HN
A bot built by Robin Baumgarten plays Infinite Mario using A* path-finding, shown as a red planned path that updates as new obstacles appear. It won the Mario AI competitions and uses only on-screen information.
HN points to it as a vivid demo that plain A* search can beat fancier machine-learning approaches in game AI.
Up next
4 mo on HNKeep your goals to yourself | Derek Sivers
TED
2 HN recommender · 2010–2023
8 mo on HNTech Talk: Linus Torvalds on git
Google
9 HN recommender · 2007–2023
7 mo on HNBlackHat USA 2011: SSL And The Future Of Authenticity
Christiaan008
9 HN recommender · 2011–2017
3 mo on HNPlaying to Lose: AI and "Civilization" (Soren Johnson)
Google TechTalks
2 HN recommender · 2011–2026
11 mo on HNThe Loving Trap
Ben Woodhams
15 HN recommender · 2012–2026
7 mo on HN27C3: Adventures in analyzing Stuxnet (Bruce Dang from Microsoft)
basti0007
8 HN recommender · 2011–2024
4 mo on HNHans Rosling: Debunking third-world myths with the best stats you've ever seen
TED
4 HN recommender · 2011–2017
1 mo on HN28c3: Print Me If You Dare
28c3
1 HN recommend · 2012–2018
10 mo on HNAlphaGo - The Movie | Full award-winning documentary
Google DeepMind
10 HN recommender · 2020–2025
10 mo on HNComputer program that learns to play classic NES games
suckerpinch
9 HN recommender · 2015–2025
5 mo on HNThe Expert Myth
Veritasium
6 HN recommender · 2022–2026
6 mo on HN30 Weird Chess Algorithms: Elo World
suckerpinch
8 HN recommender · 2019–2026
5 comments from Hacker News
The original java infinite mario was used in the fascinating Mario AI competition. Details and papers here: link Video of last years winning entry at work here: ▶ this video It's pretty interesting, a lot of people wrote/are writing high falutin' machine learning algorithms, but they are consistently trounced by good old A* techniques.
Reminds me of the AI competition on infinite mario, recently: Here is a video of the winning bot (A* based approach), cool to watch, as they display its trajectory calculations. ▶ this video GA is cool and everything, but I am sometimes sceptical about its merits. Yes, it gives you a natural problem representation, but would it really have done a better job in this particular domain than using a technique like simulated annealing to move through the Mario AI parameter space? We just don't know without seeing benchmark comparisons. Maybe its much worse. So, its a cool application, but we should be cautious about how quickly we decide GA was a particularly suitable search algorithm.
If you are impressed with this one follow the link in the video to here: ▶ this video For all my hours as a kid I was never this good, I never even realized I could kill the pipe fish by jumping from below.
There was even an Infinite Mario AI contest in which you had to write a player which gets through as many levels of increasing difficulty as possible. The winning solution? A* of course: ▶ this video (I had an also-ran entry in this contest that took 3rd place, but my solution wasn't significantly different than Robin's -- his was able to anticipate bullets firing but mine wasn't) Writeup: link
I seem to miss such contests most of the time and only hear about them when they ended. I would love to attend to some, just for the fun (because I don't know if I would have any chance to actually win). Where would be a good source for such competitions? Another fun competition I read about was the Super Mario AI competition in 2009 where the winner has succeeded with an A* algorithm which he presented in a nice video: ▶ this video The 2010 Mario AI championship also seems to be over already. Another fun competition I have heard about is the International Obfuscated C Code Contest. However, while reading this is very interesting and quite fun, I am not sure if I would have fun to do this myself. (Somehow I am more focused on the output/result of the code and I like to have clean, elegant and intelligent code, as less obfuscated as possible. :))