Formation RF Sensor
To develop and demonstrate a RF-based integrated 4p-coverage range
and bearing sensor that would enable multiple spacecraft to perform
lost in space recovery, coarse formation flying, and collision avoidance.
Terrestrial Planet Finder (artist's concept)
NASA is currently investigating several missions that
would require high-precision formation flying with instruments distributed
among several spacecraft. One of these missions is the formation
flying interferometer (FFI) version of the Terrestrial Planet Finder
(TPF) mission. The FFI is envisioned to consist of up to seven spacecraft
(as many as six “collector” with IR telescopes, and a “combiner”)
flying in precise formation within ±1 cm of pre-determined
trajectories for synchronized observations. The spacecraft-to-spacecraft
separations are variable between 20 m and 100m during observations
to support various interferometer configurations in the planet-finding
mode. The challenges involved with TPF autonomous operations, ranging
from formation acquisition and formation maneuvering, to high precision
formation flying during science observations are unprecedented for
deep space missions.
To meet these challenges, the development of a suite of sensors is
required to enable formation acquisition, stabilization, and precise
control to stay within the operating range of the optical system. For
this purpose, the formation sensor task area will develop and demonstrate
the key technology of the acquisition sensor. Key performance targets
for the acquisition sensor are an instantaneous 4p-steradian field
of view and simultaneous range and bearing-angle measurements for multiple
spacecraft with accuracy better than 50 cm and 1 degree, respectively.
Prototype Ka band AFF sensor hardware
JPL Mesa Ka band AFF sensor testbed
Main technical challenges to meet these requirements
1. Robustness to spacecraft accommodation and the interference
of indirect signal bounced off from the spacecraft structure.
2. Simultaneous multiple spacecraft operation in a distributed
3. Auto-calibration without requiring complex spacecraft rotation
4. Collision avoidance in case of temporary single spacecraft
5. Hardware accommodation on the spacecraft.
To mitigate these challenges, a new signal structure will be
developed to enable 1) an order of magnitude reduction of range
to what was demonstrated on the StarLight AFF sensor, 2) fine
bearing angle measurement without the need for spacecraft rotation
maneuver, 3) simultaneous operation for more than two spacecraft,
4) integrated radar capability for collision avoidance for multiple
spacecraft, and 5) integrated formation sensor with inter-spacecraft
communication. The acquisition sensor will operate at S-band.
The existing baseband processor as well as software will be
to allow prototype system demonstration.
Provide the FCT robots with an attitude determination system and relative position measurements.
FCT Blue Robot
The Celestial Sensor was developed to provide the FCT robots with measurements of their pose. The pose information, consisting of both attitude and position, is given relative to a coordinate frame defined in the room in which they operate. The Celestial Sensor consists of an analog camera that is used to image IR beacons placed on the ceiling and walls of the test facility. Due to the close proximity of these beacons, the camera direction measurements are coupled to both translation and attitude of the robot. This allows unique determination of each quantity, provided enough beacons are in the camera FOV.
The beacons are turned on in a sequential manner. This sequence is repeated at a 5 Hz rate. Identification of a given beacon is done based on its location within a given cycle of the strobe. An Extended Kalman Filter is executed on a peripheral DSP that takes the set of beacon bearing measurements and solves for the robotís pose. This algorithm uses a catalog of beacon positions in much the same way as a flight star tracker would use a star catalog.
The attitude portion of the pose measurement is then mixed with gyro measurements using a traditional attitude estimator. This mixing provided a 3X reduction in the standard deviation of the attitude estimates. One sigma estimation errors, after mixing, were on the order of 1.0 arcminute for each axis. Accuracy of the raw attitude, prior to mixing, was 15.0 arcminutes.
The accuracy of the raw position was determined through testing to be less than 3.0 cm.