What was the most difficult video game you’ve ever played? If it was not QWOP so let me tell you right you know that you don’t know how really difficult a game can be. The seemingly simple racing game is so challenging to master that even a AI trained using machine learning still only managed a top 10 score instead of breaking the record.
If you never reproduced QWOP before, you owe it to yourself for Take a chance and see if you can get your sprinter off the starting line. Developed by Bennett Foddy in 2008, QWOP was inspired by an 80s arcade game called Running track it requires players for Pushing buttons without thinking to win a race. QWOP takes a different approach and, instead, players use four keys to control the individual movements of a runner thighs and calves-one corridor that behaves like a rag doll and is subject toworld physics, including the effects of gravity. It may seem simple, but mastering the time and cadence of keystrokes needed to get the sprinter to go awkwardly can be incredibly frustrating.
Wesley Liao was curious to know how a tool like AI, which has was trained to do things like animate realistically Old pictures of deceased loved ones, would play the game QWOP. After creating a Javascript adapter that would allow an AI tool to actually play and interact with the game, Liao’s first attempt at machine learning simply had AI playing the game alone and learning which actions resulted in positive results (the sprinter advancing) and increasing your speed) and which resulted in negative results (bending of the sprinter’s torso very close to the ground.) Through this approach, the AI learned a “knee scraping” technique that would be able to cross the 100-meter finish line, but not at record speeds.
Liao’s next attempt to train an AI model involved recording gameplay videos of them trying to succeed in the game, including using longer strides that are crucial to increasing speed and crossing the finish line with decent timing. The approach was a little more successful, but AI was not able to master a special technique used by strikers QWOP players that involve upward and forward leg movement to generate additional momentum.
G / O Media can receive a commission
Eventually Liao reached out to a veteran player known as Kurodo (@cld_el on Twitter), one of the main QWOP speed runners around the world, who have recorded 50 videos of themselves playing the game at an expert level. But even with access to the best possible game techniques, Liao found that the best results came from a machine learning training regime that involved 25 hours of AI playing alone, 15 hours of learning from racing data collected from Kurodo experts. and another 25 hours of autoplay.
But even with all that effort, the QWOP– playing the top 100 in AIthe result of the meter dash crossed the finish line in 1 minute and 8 seconds—a top 10 Finalize. According Speedrun.com, the current world record of 100 meters is just 48 seconds, set just a month ago. Liao is confident with more training and a different reward system (as AI learns that something was done correctly), defining a QWOP The world record may eventually happen, although, being a computer playing the game, the record may never be officially recognized.