Space-based debris trajectory estimation using vision sensors and track-based data fusion techniques

Resident Space Objects (RSO) are human-made objects in orbit around Earth and can remain there for an extended period. These objects can include active satellites, rockets, and space stations, as well as debris caused by previous space endeavours. Debris originated as a consequential outcome of activities such as space launches, orbital missions and collision events, pose a formidable threat to currently operational space assets. To reduce the risk of on-orbit collisions, it is imperative that spacecraft operators enhance their situational awareness concerning potential threats posed by RSO. This necessitates comprehensive tracking of the total number of objects in space and the continuous estimation of the probability of accidental collisions. Effective Collision Avoidance (CA) manoeuvres rely on accurate tracking and characterization of RSO. Currently, RSO are monitored and catalogued using ground-based observational systems. However, Space-Based Space Surveillance (SBSS) presents a viable solution for tracking the RSO, providing superior sensor resolution, tracking accuracy, and independence from weather conditions. Accurate and continuous orbit determination of RSO is critical for developing a robust framework that enables accurate prediction of RSO dynamics. This capability is essential for applications such as Interplanetary space exploration, space tourism and Point-To-Point Suborbital Transport (PPST), which are anticipated in the future. The current study proposes a multi-sensor data fusion strategy designed to integrate angular measurements extracted from image sequences obtained by multiple cost-effective Electro-Optical Sensors (EOS) sensors deployed in SBSS missions. The main contribution of this study lies in the development of data fusion frameworks tailored for constrained computational environments, ensuring seamless real-time implementation on intelligent Distributed Satellite Systems (iDSS). This study proposes and rigorously compares three distinct data fusion methodologies—Measurement Fusion-1 (MF-1), Measurement Fusion-2 (MF-2), and Track-to-Track (T2T) fusion—examining their impact on tracking accuracy across varying sensor-to-target geometries. Additionally, the data fusion framework is validated under diverse operational conditions, including Ground-Based Space Surveillance (GBSS), SBSS, and the synergistic integration of GBSS and SBSS. A validation case study is conducted on an iDSS constellation executing a SBSS mission. The results indicate that MF-1 outperforms other algorithms in the SBSS scenario in terms of tracking accuracy. In contrast, T2T fusion demonstrates superior performance in terms of computational time. Notably, the integration of SBSS and GBSS data surpasses the performance of GBSS across all evaluated data fusion methodologies.