Feature

Robot Astronomer

Eyes peering through a telescope to gain astronomical knowledge may not be human.

Ed Budding

Within the last ten years, small computer-controlled telescopes have been providing a useful approach in astronomy. A striking example is the Automatic Photometric Telescope (APT), which allows the painstaking collection of measurements of changes in the nature of radiation emitted by celestial objects. One such "robot astronomer" is a 14-inch APT, the only one of its kind in the country, which has been in use at the Kotipu Place Observatory in Pukerua Bay.

Information of basic astronomical significance can be gathered via photometric techniques. One useful application is in the monitoring of closely orbiting pairs of stars which periodically eclipse each other. There are hundreds of these bright enough to observe with smaller telescopes. Using relatively modest procedures, photometric data can lead to a knowledge of the sizes, masses, surface temperatures and even some internal structural properties of the stars involved.

The APT is one of the simplest kinds of robotic data-collecting device which has direct practical applications in this field. Their use is on the increase, especially for variable star photometry.

The APT placed at the privately owned Kotipu Place Observatory, near Wellington, was built a few years ago by combining a commercially available Compustar and a PC-based control system. Early support for the work came via the Lottery Grants Board and further assistance from the Carter Observatory in Wellington [Automated Astronomy, April 1995].

Motivations for developing an APT can be summarized around the points of automatisation enabling a greater degree of observational precision, coverage, and networking facility. Networking is a feature of multi-component observational campaigns, allowing data to be taken and analysed from observations made around the world.

The New Zealand setting for the APT is also useful, as it allows us to contribute to continuous observational coverage when that is desired. In addition, it allows access to specialities of the southern sky, such as the nearby companion galaxies of the Magellanic Clouds.

Building a Robot
Astronomer

In moving from a conventional, human-operated photometric procedure to an automated one, the main thing to ensure is that the telescope progresses from a directly set, but approximate, initial positioning to a sure centring on the target star. This requires an information feedback loop.

In our case, this means the master PC has to send the required information through a driver box (referred to as the CCS). On receiving the appropriate start signal, the CCS energizes stepper motors, moving the telescope about either of its two axes (polar and declination) in its search for the designated star. The telescope is thus pointed, and its main light collector sends a direction-dependent light signal to the attached photometer, from which a corresponding voltage is returned to the PC's joystick port. If and when this return voltage corresponds with pre-set parameters, the PC registers that it has found the sought star and starts recording its light signal. Otherwise, the PC continues to activate the stepper motors to "hunt" for the object in an ordered search procedure.

One expects the initial targeting of the CCS to be in error, for various reasons, so this star-finding procedure is a key operation. The combination of initial proximity and relative brightness of the target star normally results in a fast, unambiguous location of the correct object. This might suggest some limitation, with a simple search strategy, to brighter, more isolated objects. In practice, for the 250,000 or so stars brighter than tenth magnitude, the nearest star to the coarse initial setting probably is the right object. If not, setting criteria may have to be adjusted more stringently. A progressive search for fainter objects in the vicinity of more easily identifiable bright ones could be devised; though this has not yet been necessary.

Difficulties in getting the search steps just right have been encountered, however, introducing progressive drifts, or non-uniformities of movement. It can be a significant hazard to efficient centering but, fortunately, absolute centering is not too crucial. Our photometer has a "Fabry lens" which stabilizes a not-exactly-central light flux input location on the detector.

The "black box" philosophy, which we were tied into with our ready-made components, inevitably raises uncertainties about performance limitations. It accents the issue of what has been fixed by manufacturers versus what can be safely rearranged by users. What is achievable in this context tends to become clear later rather than beforehand, as would be the case in a "from scratch" design.

There are advantages in having built-in command codes, and significant time savings in software and interfacing development. These are balanced by having the control software inaccessibly embedded in the microcontroller chip, making it impossible to modify, and having the software tied to existing stepper motor gearing arrangements.

The manual option is useful, allowing the CCS to be used like an "intelligent" handset, as is the use of an external "joystick" movement option.

From Small Things

Astronomy is remarkable among the fields of science in that there many more active amateurs than professionals. Alongside this, there is a pressing problem in matching supply to demand for good data. It is very hard to get enough time on large professional facilities, particularly to do the necessary, more routine observational checks, such as observing photometric standards to ensure the most accurate results. High-flying "big-science" projects are also often enhanced by supportive data from smaller telescopes, taken over a more extended period around the large facility's time allocation.

Programmes involving more routine or supportive tasks could be efficiently carried out by combining data from many smaller aperture telescopes rather than allocating large aperture devices. There could be appreciable cost savings to such an approach. Modern design features, such as efficient detectors, computer control and excellent engineering, are considerably extending the capabilities of smaller telescopes. Equipped with one of the newer CCD-type, aerial imaging photometers, a new generation of small telescopes can become more capable than the famous 200-inch Palomar Telescope was when it was built.

Combining these various developments in the context of the Internet, we can envisage a network of communicating, remotely controllable APTs, filling a very important role in extending coverage, and hence precision, of data supply. As well, there can be exciting prospects of participation in research to a new community of spare-time enthusiasts. The components of these developments are all with us -- they have not yet been fully assembled, though there exists in the US a proto-organization: the Global Network of Automated Telescopes (Inc.) which appears poised to do this job.

We have been looking at the first effective APT in New Zealand. It is likely that Wellington's APT will have an upgraded drive assembly in the near future. It may later become part of a global network of such electronically linked devices. Its smallness and relatively low overheads means that it offers outreach to a wide context of low-budget participants. With this robot astronomer we may be witnessing the embryonic form of a whole new species of providers to basic science.

Ed Budding is a professorial research fellow at the Central Institute of Technology, Upper Hutt, and an honorary research fellow with Carter Observatory.