SISIFOS: Specialized Illumination SImulator for Orbiting Spacecraft

Despite notable advances in computer vision and deep learning, the full potential of these fields remains largely unexplored in the context of spacecraft rendezvous and proximity operations. One of the main reasons is the scarcity of annotated visual data. Although synthetic datasets have emerged as a practical alternative to the complex and costly real-world data collection process, they often lack flexibility, realism, or the capacity to model diverse scenarios in practice. To address these challenges, we introduce SISIFOS (Specialized Illumination SImulator for Orbiting Spacecraft), open-source software for realistic image generation of spacecraft in orbit. SISIFOS enables the mass generation of high-fidelity images under realistic space-environment conditions and is integrated with a comprehensive trajectory generation module, allowing users to simulate complex rendezvous and proximity operations under controlled, yet diverse configurations. The images are also accompanied by an abundance of ground-truth maps, such as depth and surface normals, emulating a complete sensor suite and enabling the use of multimodal inputs for downstream (e.g. relative navigation) tasks. The SISIFOS platform represents a powerful tool for evaluating perception and planning algorithms in orbital scenarios and contributes significantly to the development of robust autonomous systems in support of future space missions. The code and database can be found through https://github.com/JasonVel6/SISIFOS.