Implementação de uma interface para o framework de programação de sistemas multiagentes MASPY

Abstract

Debugging multi-agent systems is one of the current challenges in the field of multi-agent systems, owing to the changing internal states and to the behaviors arising from the interactions between agents. MASPY is a Python framework for developing multi-agent systems, but until now its information could only be visualized through a command-line interface. In view of this limitation, this work aimed to develop and validate MASPY-GUI, a graphical interface for the visualization and debugging of multi-agent systems in MASPY. The interface receives the execution data directly from the framework; after that, it captures and classifies the information and presents it in four tabs: a home page that summarizes the system state and the intentions, an agents page that details beliefs, goals, and the history of intentions, an environment page that shows the percepts and their history over time, and a messages page that displays the communication between agents throughout the simulation, in addition to a tool that allows generating message-exchange diagrams while letting the user select the desired agents. Another essential tool is the timeline navigation system, which makes it possible to move forward and backward through the execution history and to replay the simulation again at different speeds, including after the MASPY execution has finished, thus allowing the user to examine what happened during the simulation. The validation was carried out in two stages; in the first stage, a performance evaluation was conducted with fourteen scenarios, varying the numbers of agents, intentions, and messages exchanged between agents; in these scenarios, execution time, memory usage, and processor usage were assessed. In the second stage, two usability studies were conducted with students from two universities, using the System Usability Scale, which resulted in scores of 88.35 and 73.6 respectively, both considered positive, which guided us in making improvements to the timeline and to the visualization of percepts based on the feedback received in the surveys. It is concluded that MASPY-GUI met the proposed requirements, succeeding in providing a way to visualize the internal information of multi-agent systems in MASPY. As future work, we foresee support for the visualization of learning processes, the possibility of direct modifications to the agents, and further optimizations to handle large-scale systems. This undergraduate thesis summary was based on the article “A Graphical interface for Visualising and Debugging MASPY Agents”, presented in “The 14th International Workshop on Engineering Multi-Agent Systems (EMAS 2026)”