Hybrid CNN and RNN Approach for Space Debris Detection and Classification

Terreros R.S.; Torres Y.A.; Silva-Cuadros R.; Saico C.D.; Mucha-Mateo J.; Roman-Gonzalez A. · Proceedings of the International Astronautical Congress, IAC (2025) · DOI: 10.52202/083079-0146

Resumen

The increase in satellites and space missions has led to a significant rise in space debris in orbit, posing risks to the safety and sustainability of space operations. Traditional detection methods, based on radars and ground-based telescopes, have limitations in terms of accuracy and coverage and are costly, which hinders effective detection and monitoring. Our goal is to develop an innovative machine learning-based tool for the detection and classification of space debris from satellite images, integrating convolutional neural networks (CNN) and recurrent neural networks (RNN). The CNN architecture extracts relevant spatial features, while the RNN captures temporal and sequential patterns, improving the identification and tracking of moving objects. The model is evaluated using metrics such as accuracy, recall, F1-score, and inference time, ensuring its operational viability. To train and validate the model, we used the Space Debris Detection Dataset, used in the AICrowd AI Blitz 7 – Debris Detection competition and publicly available on Kaggle. The database consists of 20,000 images for training, 2,000 for validation, and 5,000 for testing. The proposed model is expected to outperform traditional methods and significantly improve accurate and rapid classification in the identification of space debris, which would contribute to reducing the risk of collisions and strengthening mitigation strategies. This study contributes to the preservation of orbital space and the development of more efficient technologies for the monitoring and management of space debris, promoting the safety and sustainability of the space environment. Copyright

Instituciones: Universidad Nacional Mayor de San Marcos; Universidad Nacional Tecnológica de Lima Sur; Universidad Nacional del Centro del Perú