Close-proximity testing with real hardware
RAFFAELLO is DART's robotic hardware-in-the-loop (HIL) facility for autonomous Guidance, Navigation, and Control (GNC). Its primary focus is vision-based navigation (VBN) around small bodies. A flight-like camera acquires real images under controlled physical, optical, and geometric conditions.
Robotic architecture
The facility operates two industrial robotic manipulators in a darkened range. A rail-mounted arm carries the camera through an extended workspace, while a ground-fixed arm carries the target. Scaled asteroids, spacecraft mock-ups, docking interfaces, and fiducial calibration patterns can be exchanged to suit the experiment. Controlled illumination reproduces the viewing and lighting geometry specified by the test scenario.
The two-arm configuration makes the relative camera-target motion programmable while keeping the experiment repeatable and observable. The same environment can be used for camera characterization, navigation-data generation, closed-loop GNC experiments, and comparisons between synthetic and acquired imagery.
Calibration and safe motion
An opto-geometric calibration framework estimates the relative poses of the robots, camera, and target together with the camera's intrinsic parameters. A dedicated correction accounts for robot axial drift. An external time-of-flight sensor can be used to calibrate the target mounting, and the illumination geometry is characterized separately.
A mapping layer converts desired relative poses and trajectories into feasible joint commands. Before the robots move, it applies kinematic constraints, workspace limits, and collision-avoidance requirements. This mapping connects a mission scenario to safe laboratory commands.
Applications
RAFFAELLO is used to generate datasets around scaled small-body targets, evaluate fiducial-marker navigation, validate image-processing and VBN algorithms, and exercise autonomous guidance during close-proximity operations. Its combination of controlled image acquisition and programmable motion brings camera behavior and laboratory geometry into verification and validation (V&V) experiments.