Carnegie Mellon University proposes a five-year program to develop telerobotic technologies to deploy elements of a lunar optical interferometer. The program will demonstrate rover deployment of optical telescopes within a 1 kilometer diameter circle. The telescopes will each point at the same star, and relay the gathered light to a simulated beam combiner on a central lander mock-up.
Key research issues include the following:
The concept that has received the most attention is an optical/IR imaging interferometer [1][3][8]. This concept may represent the largest advance in technology, resolution, and science potention of any of the proposed lunar telescopes.
The Moon is particularly well-suited as the location for an optical interferometric array: the vacuum environment improves observation sensitivity and allows observations at all spectral bands; the large, stable, slowly rotating platform offers kilometer-scale viewing baselines; the knowledge of its sidereal motion allows for microarcsecond correction of stellar aberration; and the 100 K nighttime temperature passively cools optics permitting high sensitivity.
Two lunar optical interferometer designs have been proposed. The first [1], modeled after the National Radio Astronomy Observatory's Very Large Array in New Mexico, calls for a Y-shaped distribution of elements with 27 four-meter aperture elements distributed with a maximum baseline of 10 km. The second [3] calls for the interferometer elements to be distributed along a non-redundant Cornwell circle with an outer diameter of 10 km.
A number of precursor designs have also been proposed to provide the necessary engineering data for a more informed design study of a future lunar interferometer. McGraw [6] suggested a 1 m lunar transit telescope with few moving parts that would perform a repeated transit survey of strips of the sky. Genet [5] proposed a fully-steerable 1 m class telescope that would perform the first all-sky CCD imaging survey in the far-UV.
We select optical interferometry, rather than other telescope concepts, to serve as the focus of this technology push because technically it presents the most challenging problems and because scientifically it promises the greatest potential for revolutionary breakthrough.
Deployment. Rovers will explore the vicinity of the landing site, gathering information about favorable and unfavorable terrain areas. Based on this reconnaissance data, a deployment pattern will be selected and uplinked. Rovers will carry telescopes to designated locations, and a positioning procedure will determine with centimeter accuracy the telescope position and orientation with respect to a central station (here called the lander, although it may itself be deployed rather than landed).
Calibration. The telescopes will point at reference stars and determine the kinematic and dynamic parameters of the motorized mounts. Since many motion stages are involved, the parameter estimation will consume a significant amount of time. Meanwhile, the lander will point its acquisition cameras at the telescopes and determine the kinematic and dynamic parameters of its motors, as well as a variety of geometric and radiometric camera parameters.
Pointing. Each telescope will acquire and track the same star. The light gathered by each unit will be relayed to the central station on the lander.
Phasing. At the lander, the optical delay lines will be adjusted to equalize path lengths and produce white light fringes. This data will be recorded and downlinked for analysis and interpretation.
Iterate. Repeat the pointing and phasing process for each star of interest.
To protect the optics from lunar dust, and to reduce interference from stray light sources, concertina baffles will extend around both the main and exit apertures. They will remain extended throughout the life of the telescope. To provide further protection against dust, flaps will be closed whenever the telescope moves.
Communications between the central station and rovers will take place over an omni-directional antenna and a directional VHF antenna attached to the same gimbal system that drives the mirror that directs the starlight beam to the base station.
The entire structure stands freely, and need not be especially stiff. Thermal effects and the changing gravitational load as the telescope tracks celestial objects will cause optical path lengths to change slightly over time. Laser metrology and high-bandwidth variable optical delay elements in the base station are essential to compensate for this so that interference fringes can be maintained over the course of an observation.
During transit the telescope will be locked into the most stable position and the dust flaps closed. Upon reaching the destination the manipulator disengages the tow bar and activates wheel locks. Conceivably it could also open the dust flaps as well as performing other tasks, such as removing lunar dust from the structure with brushes, blowers, or ion beams.
Once grasped, the forklift raises the telescope and moves to a more stable position for transit. The telescope itself may change orientation to facilitate this.
In the first year, we will concentrate on the two latter tasks.
First, we will investigate computer vision-based triangulation. One approach involves pointing the telescopes at the central station, and triangulating from markers at known locations on it. Another approach requires a camera on the central station to point at a rover, and to triangulate from markers on it or markers that it has deployed nearby. With an estimated marker position estimation precision of one-half arc second, a marker separation of 0.5 m produces an accuracy of 4.9 m, a separation of 1 m produces an accuracy of 2.5 m, and a separation of 10 m produces an accuracy of about 1 cm.
Second, we will investigate brightness-based ranging techniques that capitalize on the inverse square relationship of intensity and distance. The basic idea is to use a beacon with a known intensity (units of candles) and observe it from a distance; the observed intensity will vary as the inverse of the distance. A 1 cm accuracy over 1 km requires intensity measurement precision of 2x10^-5 times the beacon intensity, which will require a high-precision detector.
Third, we wil investigate laser or microwave ranging devices, such as the electronic distance measuring equipment used in the surveying industry. These devices are capable of measurement accuracy of plus or minus 5 mm plus 2 ppm; over 1 km, the accuracy is thus 7 mm. We will study adapting such equipment for operation under lunar conditions, in particular, lunar temperature ranges. Our customization will use the beam steering and CCD camera setup already being used by the telescopes and lander, respectively.
Fourth, we wil develop multi-sensor fusion techniques that improve both the accuracy and the precision of range and angle measurements. We will apply these techniques to the measurements from the triangulation, inverse square, and laser-based approaches.
We will develop a control architecture that supports decision-making both by ground-based operators and elements of the system on the lunar surface. This architecture will extend the Computer-Aided Remote Driving approach from JPL and the Safeguarded Teleoperation approach from CMU. The fundamental concept is that two command streams, one generated on the ground and the other on the lunar surface, are arbitrated by a decision-making module.
One key issue is the coordination of multiple rovers and the central lander station. Previous work, for example at MIT, has addressed multi-robot coordination, but the herding and swarming concepts that have emerged are not appropriate for small teams that must operate efficiently. We will develop schemes for rover-lander communications, rover-rover communications, and remote-local communications.
We will develop an operator interface to run on a single workstation. This interface will enable safeguarded teleoperation of the rovers, the science instruments, and the lander.
| Item | FY96 ($K) | FY97 ($K) | Total ($K) |
|---|---|---|---|
| Equipment | 40 | 30 | 55 |
| Total Cost | 250 | 250 | 500 |
| Element | FY96 | FY97 | FY98 | FY99 | FY00 |
|---|---|---|---|---|---|
| System Type | Breadboard | Field-Portable | Single-Rover | Multi-Rover | Supervised Autonomy |
| Telescope Mounted On | Lab Bench | Hand Cart | Rover | Rover Team | |
| Camera Mounted On | Lab Bench | Tripod | Lander Mock-Up | N-camera Lander | |
| Viewing Baseline | 10 m | 100 m | 1,000 m | 100 m diameter | 1,000 m diameter |
| Pointing/Alignment | Manual | Supervised | Autonomous Beam Steer | Autonomous Camera Align | Full Autonomy |
| Metrology Method | Manual | Triangulation; Inverse Square Brightness | Laser Ranging | Multi-Sensor Fusion | |
The Year 1 effort will produce a breadboard system consisting of a telescope with a 30 cm aperture, a steerable mirror, and a steerable camera, all under computer motion control. As the Year 1 milestone, the camera will "see" a ceiling-mounted LED through the telescope/mirror system with a viewing baseline (telescope/camera separation) of 10 m. Researchers will measure all distances and angles; the system will perform its own pointing and alignment.
The Year 2 effort will produce a field-portable system with the telescope/mirror unit on a two-wheeled dolly, with the camera on a tripod, and with self-sufficient computing, power, and other resources. This system will perform stellar imaging with a viewing baseline of 100 m at a remote site in Pennsylvania. The pointing and alignment process will be automated, and metrology will be implemented with triangulation-based computer vision.
The Year 3 effort will develop a single-rover system to carry the telescope/mirror, and a single-camera lander mock-up. This system will perform stellar imaging with a viewing baseline of 1,000 m. The rover will be a modified version of the CMU Lunar Rover, where the modifications involve installing the deployment mechanism developed in Year 2. The rover will travel under safeguarded teleoperation control to two designated sites, each 500 m from the lander. At each site, the rover will calibrate the telescope and steerable mirror mounts, acquire and track a reference star, and relay the gathered light to the lander. At each site, the lander will calibrate its camera and mount, and acquire the rover optical feed. A single operator control station will control both the rover and the lander, sending commands and receiving status and data.
The Year 4 effort will develop a multiple-rover system and a multi-camera lander mock-up. This system will perform synthesis imaging (simulate interferometry) over a viewing area of diamter 100 m. The rovers will be new vehicles configured to meet astronomical instrumentation requirements including power, mass, angular articulation range, stabilization, and vibration suppression. The rovers will deploy their telescopes to designated locations and simultaneously relay images of the same star to the lander, which will simulate beam combination.
The Year 5 effort will demonstrate supervised autonomous operation of the multiple-rover system developed in Year 4. Remote operators will use a single control station to designate the telescope deployment pattern--spanning a viewing area of diameter 1,000 m--and supervise rover navigation. All other command sequences will be generated by computer and uplinked automatically to perform wide-baseline synthesis imaging (simulate interferometry).
The table below depicts the Level Two milestones in the Lunar Telescope Deployment task in FY96. The first two quarters will concentrate on requirements and specifications, so as to make the optical instrumentation as realistic as reasonably possible, and the demonstrations as convincing as possible to the astronomy community. We will consult with experts in the interferometry group at JPL, including Dr. Mark Colavita and Dr. Jeffrey Yu, and lunar scientists on the Lunar Exploration Science Working Group. The second two quarters will concentrate on designing, procuring, implementing, and demonstrating a breadboard system. The centerpiece of this system will be a commercial telescope with a 12.5 inch clear aperture and motorized mount. Testing will take place in a highbay environment, with ceiling-mounted LEDs simulating stars. The breadboard system will point the telescope at designated targets, and steer the gathered light to a CCD camera located 10 meters from the telescope.
| 96Q1 | 96Q2 | 96Q3 | 96Q4 |
|---|---|---|---|
| Requirements Derived | Specifications Derived | Breadboard System Design | Breadboard System Demonstration |