by Dimitrios Arampatzis (FORTH-ICS), Maria Doulgeraki (FORTH-ICS), Michail Giannoulis (Univ. of Crete), Evropi Stefanidi (FORTH-ICS), and Theodore Patkos (FORTH-ICS)
EduBAI (Educational Basketball playing using Artificial Intelligence) is an educational platform that helps users familiarise themselves with the main tenets of common-sense reasoning in dynamic, causal domains, by means of an interactive entertaining environment. This article discusses the design and features of the platform, along with the rationale of sample game tactics of diverse modelling complexity.
Formal action languages are well-established logical theories for reasoning about the dynamics of changing worlds, contributing solutions to domains as diverse as high-level robot cognition, ambient intelligence and complex event detection. The heterogeneous application areas of these formalisms attract the interest of researchers from a variety of disciplines and with diverse backgrounds. Given the quite advanced technical training needed to master logic-based reasoning, the educational process of understanding both the theoretical and practical aspects of common-sense reasoning is a challenging task.
EduBAI  (Educational Basketball playing using Artificial Intelligence) aims to facilitate the learning process of logic-based reasoning. EduBAI is a game platform that assists students and researchers having basic knowledge of the underlying formalisms in writing, executing and evaluating logic-based axiomatizations. It offers a testbed for experimenting with a rich range of reasoning features, structured around a game scenario involving three-player teams competing against each other in a basketball match. The gameplay gives the ability to explore aspects related to causality-based reasoning, geospatial reasoning, multi-player coordination, temporal reasoning, probabilistic reasoning, function optimisation and others.
The platform was developed in the context of a post-graduate knowledge representation and reasoning course at the Computer Science Department of the University of Crete and has been co-developed by students. The gamification nature of the approach aims to offer a more entertaining way of introducing logic-based programming, motivated by the observation that toy problems in the form of grid settings and logic puzzles have proven to be an effective way of understanding insights of AI programming. The platform is free-to-use and available online for testing and play [L1].
Users of the platform model the intelligence of their team using formal languages, such as the Event Calculus and Answer Set Programming (ASP), in the context of a dynamic, non-deterministic domain. The Event Calculus is a narrative-based many-sorted first-order language for reasoning about action and change, while ASP is a declarative problem-solving paradigm oriented towards complex combinatorial search problems.
Each of the six players can move within the premises of the EduBAI basketball court, represented as a 5 × 7 grid (Figure 1). The players can perform only a limited set of actions, namely shoot, move (up, down right, left) and pass. Player actions are contingent on appropriate preconditions; for instance, a player can perform a pass only if she is in possession of the ball. Moreover, the outcome of each action is decided probabilistically, e.g., a shoot action has 95% probability of being successful if it occurs in the same cell as the basket; distance from the basket reduces the prior probability by 17% per cell, whereas each opponent in the same cell reduces the probability by 25%. The game proceeds in rounds. At the beginning of each round, the human users decide the starting position of their players, as well as the defending and attacking tactic that their team will follow during the round; these tactics cannot change in the course of the given round, therefore once this initialisation step is completed, the game proceeds without any human user intervention.
Figure 1: Snapshot of the EduBAI UI.
One of the objectives of the EduBAI platform is to help users understand the main principles of causality-based reasoning, through the axiomatization of their own attacking and defending team tactic intelligence. EduBAI enables the user to experiment with features of increasing complexity, still presenting an operational team every time, permitting a gentle introduction to the modelling concepts. Tactic encodings can start by ignoring the opponent positions, for instance, and only follow specific, pre-defined patterns; or they may consider spatial and topological relations among players, probabilistic reasoning on the effects of actions, cost function optimisations etc.; they may even take into account past actions, in order to learn patterns and predict future moves.
Future plans involve extending the platform with friendly interactive features that will further enhance the educational experience - for example, an embedded editor to enable users to encode their tactics or a tutorial-style assistant to walk a new user through the main functionalities. In addition, we plan to offer a more advanced competition environment, where the users can schedule matches or upload their own AI-enabled bots and keep track of online statistics.
 D. Arampatzis, et al.: “EduBAI: An Educational Platform for Logic-Based Reasoning”, in: Artificial Intelligence Applications and Innovations, AIAI 2019, IFIP Advances in Information and Communication Technology, pp. 464-472 vol 559, 2019.
Theodore Patkos, Dimitrios Arampatzis, FORTH-ICS, Greece