GEORyder Close Range Spacecraft Rendezvous Dataset for Vision-Based Navigation in Geostationary Transfer Orbits

This dataset has been generated within the framework of the Horizon Europe 2023 GEORyder project (Grant Agreement No. 101135095) and contains image sequences acquired during an experimental campaign addressing close range spacecraft rendezvous scenarios in Geostationary Transfer Orbit (GTO). In particular, the experiments reproduce the terminal segment of the rendezvous phase under controlled laboratory conditions, covering the operational interval from 15 m to 2 m relative distance. To preserve geometric and optical representativeness within laboratory constraints, the close range domain is divided into two segments (15–5 m and 5–2 m), each replicated using a geometrically scaled 3D-printed target equipped with eight ArUco markers. Image acquisition is performed with a calibrated monocular industrial camera rigidly mounted on a robotic manipulator executing predefined approach trajectories generated by a guidance module. The camera motion reproduces the relative dynamics of the terminal rendezvous phase. For each trajectory, image sequences are recorded under two exposure configurations to emulate illumination variability and assess perception robustness. The dataset is organized by target configuration, trajectory index, and exposure setting, enabling systematic evaluation of marker detection, pose estimation, and filtering algorithms under controlled yet representative visual conditions. The dataset enables quantitative benchmarking of close range visual navigation algorithms, including marker-based pose estimation and nonlinear filtering approaches for autonomous rendezvous applications. ---------------------------------------------------------------------- The experimental campaign used to generate this dataset was conducted at the RAFFAELLO (Robotic Arm Facility For Autonomous cubesat ExpLoration in cLose proximity Operations) testbed developed by the DART (Deep-space Astrodynamics Research & Technology) Laboratory of Politecnico di Milano. RAFFAELLO is a robotic kinematic emulation facility designed for hardware-in-the-loop validation of vision-based Guidance, Navigation, and Control (GNC) algorithms for close proximity space operations. The facility reproduces spacecraft–target relative motion using industrial robotic manipulators, enabling controlled six-degree-of-freedom relative pose generation between the camera sensor and the target mock-up. The experiments are performed in a dedicated darkened environment equipped with controllable illumination sources designed to emulate space-like lighting conditions. The authors gratefully acknowledge Fabio Ornati, Alban Beshaj, Elena Pilo, Pietro Califano, Umberto Mondini, Prof. Paolo Panicucci, Prof. Paolo Lunghi, and Prof. Francesco Topputo from the DART Laboratory (Politecnico di Milano) for providing access to the RAFFAELLO facility and for their valuable support during the tests that enabled the acquisition of the dataset.