Automated Playing of Survival Video Games with Commonsense Reasoning

Watch a 22-minute conference talk from the 27th International Symposium on Practical Aspects of Declarative Languages where researchers from the University of Texas at Dallas present their work on automating gameplay in the survival game "Don't Starve." Learn how they developed an agent using answer set programming (ASP) framework and commonsense reasoning rules to navigate the challenging, randomly generated environments of the game. The presentation explains how their system translates environmental knowledge into predicates, applies logical rules to determine optimal actions, and interfaces between the s(CASP) goal-directed predicate ASP engine and a custom Lua modification program. Discover how their preliminary automated system already outperforms novice human players, with potential for further improvements to extend survival time. This talk was presented at POPL25, sponsored by ACM SIGPLAN.