Artificial Intelligence basically covers all of the cognitive abilities necessary for artificial systems to exhibit the behaviors exhibited by intelligent creatures. Representing and presenting information, logical inference, planning and learning can be counted among these abilities. Robots are at the forefront of systems where these abilities can best be demonstrated and related artificial intelligence algorithms can be tested in the real world. These cognitive abilities are built on lower layer algorithms such as image processing for robots to interact with objects and the environment, mapping for safe navigation in the environment, and path planning. Video games also offer advanced test environments for the development and testing of artificial intelligence methods. Especially for bot systems that play the game automatically in many popular video games, it has recently become possible for these bots to beat the grandmasters of the game with advanced deep learning methods.
Assoc. Prof. Dr. Sanem Sarıel has been working on these subjects since 2007. She and her team are developing artificial intelligence methods for robot systems and video games in ITU Artificial Intelligence and Robotics Laboratory. Among their most important works are the development of cognitive abilities such as high-level planning and learning with experience to perform daily tasks in cognitive robots, logical inference in games, creating player model, bot design, game content and level design.
In their work with robots, they developed active vision and world behavior modeling, planning and learning from experience methods for autonomous systems. They carried out studies on the ability of robots to sense their own mistakes and learn not to make mistakes and not harm their environment by gaining experience from these mistakes. Recently, they have been carrying out Tübitak-supported projects to make these methods more advanced for safe task execution in humanoid robot systems.
In their study of logical inference in games, they have been developed a system that automatically solves puzzles in video games. This study has been published in the most prestigious journal on Artificial Intelligence in Games.
In another study on games, using machine learning methods, they developed methods for modeling the behavior of the players according to the event and action selection, order and habits in the game stories and improving their experiences. This project was deemed worthy of the "Best Product" award at the 13th Technology Awards given by TÜBİTAK, TTGV and TÜSİAD as part of a project carried out jointly with the industry.
Prof. Dr. Hatice Köse and her team continue to work on gamification and serious games. These studies are especially developed as part of the education and therapy of children with hearing impairment or autism. They carry out 1 Tübitak project and 1 European Union (Erasmus+) project in this area. The data received from the users participating in the game through sensors are processed with artificial intelligence methods and fed to the adaptation and feedback mechanisms of the system. Social Robots are also used as a part of this system.