Abstract
This dataset provides synthetic event-camera data and ground-truth camera motion for autonomous lunar navigation, covering two mission phases: Low Lunar Orbit (LLO) and the Coasting phase of a descent trajectory. Synthetic image sequences were rendered at 100 Hz using ABRAM, a physics-based radiometric renderer for celestial bodies, using the 2015 SLDEM for the LLO and Chang’e 2 DEMs for the coasting. Event streams were then generated with the same temporal resolution (without frame interpolation) using the v2e emulator, with two empirically-tuned noise/threshold configurations selected according to the sequence’s phase angle (above/below 60$^∘$). The dataset contains 258 sequences: 164 LLO sequences in moon_LLO/ and 94 Coasting sequences in moon_COA/. Each sequence includes an events.aedat4 file and a yml folder with the metadata associated with each rendered synthetic frame. A shared camera.yml file provides the camera intrinsics common to every sequence. Sequences in both LLO and COA folders are divided in tiles, based on the moon region and are denominated as: moon_[TYPE]###_dt10_##[N/S]_### where: [TYPE]### identifies the sequence category and number (C## for coasting and S## for LLO) dt10 indicates the timestamp resolution of synthetic images used to generate the event stream (which coincides with event stream resolution) ##[N/S]_### specifies the latitude (+/- 90) and longitude (0/360) of the first pose (e.g., 58N_123 is at latitude 58 N and longitude 123)
Paolo Lunghi
Associate Professor of Space Systems
Aiming for autonomous Guidance, Navigation, and Control for spacecraft.