Ikaros: Building cognitive models for robots

Abstract

The Ikaros project provides an open infrastructure for system-level brain modeling and real-time robot control. Utilizing a modular architecture, the system enables the construction of complex cognitive models by connecting functional components via an XML-based description language. All inter-module communication is standardized as matrices of floating-point numbers, facilitating compatibility across diverse computational algorithms. The simulation kernel supports discrete-time execution, multi-threading for real-time performance, and platform independence through adherence to ANSI C++ and POSIX standards. An integrated web-based interface leveraging SVG and JSON allows for remote monitoring and decentralized simulation control. In addition to its core processing capabilities, the framework includes tools for validating models against neurobiological databases and an extensive library of experimental data. Since its inception, the infrastructure has supported research in areas such as cognitive development, visual attention, and emotional learning, while providing control systems for robotic platforms ranging from mobile agents to multi-degree-of-freedom hands. This modular and inclusive design offers a transparent environment for large-scale cognitive systems modeling and cooperative research. – AI-generated abstract.