Through the use of reinforcement learning to train AI, researchers from the Singapore University of Technology and Design (SUTD) created an AI, dubbed Agent Ryu, capable of beating professional Street Fighter players. According to the associated study, the researchers’ new approach to training artificial intelligence could innovate movement science and affect everything from aerial drones to self-driving cars.
“The results were astonishing,” Desmond Loke, associate professor at SUTD and principal investigator of the study told Interesting Engineering , “Our findings demonstrate that reinforcement learning can do more than just master simple board games. The program excelled in creating more complex movements when trained to address long-standing challenges in movement science.”
Agent Ryu AI Beats Brandon Chia At Street Fighter
The researchers asked one of the top Street Fighter players, Brandon Chia, to compete against the AI in a time-trial evaluation. Chia took first place in the Capcom Pro Tour 2022 and 1st position in the Intel World Open 2021. SUTD’s artificial intelligence defeated the six in-built opponents faster than Chia. However, Agent Ryu took longer than 10 seconds to defeat in-built opponents Dhalsim and Blanka, while the human Street Fighter player did not defeat these two.
At first, the AI and Chia were at a draw with a score of 3-3. However, Agent Ryu soon gained a lead in damage dealt, after which Chia could not catch up for most of the remaining matches. Chia did manage to defeat the AI while playing as Dhalsim, though both characters suffered a tremendous amount of damage before the match ended.
“Agent Ryu is more difficult than the in-built opponent, countering a lot of my inputs with Shouryuken to the point that it is almost unplayable. It would be more useful for tournament practice if there is a way to reduce the AI’s sensitivity to human inputs. Nevertheless, it is still very fun,” said Chia of the first event.
In a second event, the AI won matches with a score of 4-2, taking 10% less time to defeat the Street Fighter player. SUTD’s artificial intelligence was able to defeat the player in less time due to a new combination of moves that the AI itself created. The agent still lost to Blanka. However, the AI dealt substantially more damage to the character in the second event.
Jack Brassell is a freelance journalist and aspiring novelist. Jack is a self-proclaimed nerd with a lifelong passion for storytelling. As an author, Jack writes mostly horror and young adult fantasy. Also an avid gamer, she works as the lead news editor at Hardcore Droid. When she isn't writing or playing games, she can often be found binge-watching Parks & Rec or The Office, proudly considering herself to be a cross between Leslie Knope and Pam Beasley.
