The Formation Control Testbed Celestial Sensor: Overview, Modelling, and Calibrated Performance

The formation control testbed (FCT) is a ground based multiple robot testbed for simulating the dynamic interaction of spacecraft formations in a representative 6-DOF environment. Linear and spherical air bearings are used to mimic the drag free space environment. Each robot is fully autonomous with a self contained supply of float gas, integrated batter)\´ power, and a complete suite of onboard avionics including IMU (3-axis gyro), wireless interspacecraft and ground communication links, cold-gas thrusters, and reaction wheels. For attitude determination each robot uses an analog camera to image 1R beacons fixed to the walls and ceiling of the test facility. These navigation beacons act as an artificial star field. Due to the close proximity of these beacons, the camera direction measurements are coupled to both translation and attitude maneuvers of the robot. This allows unique determination of each quantity, provided enough beacons are in the camera FOV. We have come to refer to this sensing scheme as the "celestial sensor". In this paper, each subsystem of the celestial sensor is discussed with emphasis given to the filtering algorithms. The celestial sensor software processes sequential frame based bearing measurements on a peripheral CPU specifically designed for this application. A frame preprocessor is used to normalize each bearing measurement and apply a number of accept/reject rules. The accepted set of measurements is then passed to an extended Kalman Filter (EKF) that is tuned to track the motion of the robot within the room. A detailed sensor model is described in this paper that is used to predict the performance of the integrated system. Frame based simulations using this sensor model are presented that predict 1-sigma errors on the order of 3.0 arc minutes in attitude (per axis) and 4.0 millimeters in position (per cartesian coordinate). Preliminary results from the production system are given that demonstrate similar resolution statistics and decou- pling of the attitude and position estimates