Scenario Generation for Robot Simulation from Public Data

Robotic simulation environments require realistic scenarios to test navigation algorithms and sensor fusion. Traditional manual scenario creation is resource-intensive and limits testing diversity. This paper presents a system for automated generation of realistic 3D scenarios from publicly available geospatial data for robotic simulation. The system enables selection of geographical areas through an intuitive interface, processes terrain data at multiple detail levels, and generates simulationready models compatible with major robotics platforms. Our approach addresses scalability, realism, and computational efficiency challenges. Performance analysis reveals optimal balance between visual fidelity and real-time simulation requirements. The system integrates with ROS 2 systems in Gazebo and Unity environments, demonstrating significant reduction in scenario creation time from weeks to minutes while maintaining geometric accuracy and texture quality.