High-Fidelity Serious Games in Simulation Robotics: A Methodological Framework for Mission-Critical Scenarios

The convergence of Simulation Robotics and Serious Games has enabled a new methodological paradigm for the development, evaluation, and training of autonomous robotic systems in safety-critical domains. This paper presents a comprehensive framework for designing high-fidelity, mission-oriented simulation environments that function simultaneously as scientific testbeds and pedagogical Serious Games. The proposed approach leverages Digital Twins implemented in Gazebo to model representative disaster scenarios under GPS-denied and perceptually degraded conditions, enabling repeatable and risk-free experimentation. Within this framework, robotic agents are trained and evaluated using Deep Reinforcement Learning, combining discrete and continuous control policies to reflect heterogeneous robotic platforms. Localization and mapping are treated as core mission capabilities through the integration of RTAB-Map for real-time 3D volumetric reconstruction, while semantic perception is embedded directly into the game loop using a state-of-the-art YOLOv11 detection pipeline. Perception events, mapping quality, and navigation safety jointly determine mission success, ensuring that autonomy is evaluated at the task level rather than through isolated algorithmic metrics. The framework is validated across a hierarchy of progressively challenging simulation environments, including a structured calibration space, an industrial disaster setting with reflective sensory adversaries, and a feature-poor subterranean scenario. Experimental results demonstrate that geometric, LiDAR-centric SLAM is essential for reliable localization in textureless environments, while continuous-action reinforcement learning policies exhibit superior stability and safety in cluttered spaces. Overall, this work establishes a scalable and scientifically grounded methodology for simulation-based robotics research and training, highlighting the role of Serious Games as a unifying construct for autonomous system development and human-robot team preparation.