Small Robotic Observatories

Small Robotic Observatories: Operations, Deployment, Future Developments

Jeffrey S. Medkeff
David H. Healy

(This is a web-version of the oral outline that was presented to MPAPW III in the form of a PowerPoint presentation. I've expanded on the outline where additional comments are necessary to make the meaning of the outline clear.)

701 Junk Bond Observatory

• 16” Meade LX-200
• ACP/PinPoint/MaximDL/CAA
• Robotic control script defeats telescope
• Reference implementation for Rich Guy's Telescope:
  .8 meter (32”) R-C f/7. Rich Guy's Detector = 2k thinned?

Definitions

• Script: Software program in a high level interpreted programming language, which runs in its own operating system space or within a host application, and whichuses OS resources.
• List Processor/FIFO Event Generator/etc: A program that steps through a sequential set of commands or task routines. E.g., MPO Connections (“best”).
• ASCOM: An implementation of the Component Object Model, by which software may accept external calls.
• Unintelligent Robot: Runs autonomously but without environmental self-awareness.
• Intelligent Robot: Runs autonomously and analyses its environment with its sensors.
• Windows: Refers only to modern Win32 operating systems such as Windows 2000. (This definition is adopted for purposes of this talk.)
• Robotic:

Steve Larson: “Robotic” means you get some sleep.

Jeff Medkeff: “Robotic” means you occasionally crash in new and sometimes highly interesting ways.

Really means hardware is under the influence of a programmable automation controller.

Reasons for its Increasing Adoption up to 1 meter

• Device independent: 22 camera lines; LX-200 & PC-TCS and other ACL, et al.
• To some extent, server application independent.
• Cheap - Medkeff’s control script is freeware, as are many others.
• Robust & Powerful - inclusion of logical operations, mathematical functions, etc.
• Relief of necessity to low-level program (e.g., device drivers, detection code, etc).
• Language independence (Perl, TCL, Python, JavaScript, JScript, VBScript, etc ad nauseum).
• Feature independence from developer, but with technical support. (Some developers have asked me "why do you want to do that?" as though they drive the design requirements. ASCOM releases the observatory operator from that kind of attitude.)
• Developing nations have greater Windows expertise than Unix expertise (although nothing prevents ASCOM on Unix except for limited Unix implementations of the standard).
• Amateur and self-funded programs tend also to have greater Windows than Unix familiarity.

The 701/933 Reference Implementation

• Goal 1: Create a powerful data acquisition script for asteroid observing; must be easy to use; must not be time consuming to set up observations.
• Goal 2: Create a reference version for use on RGT; propose this reference as a current state of the art to MPAPW. (Rather brash, don’t you think?)

Optional Operations

• Unpark, sync, focus (robotic or manual).
• Explicit target selection (robotic and manual).
• Live reductions during observation

Operational Loops

• Process target list; schedule explicitly selected targets.
• Must achieve a time domain separation between observations, and must catch objects widely spaced in RA without doing anything silly.
• Schedule, buffering with automatically selected targets or surveys.
• Process target lists and buffers:

  Slew
  Closed Loop Pointing (optional)
  If exptime > n, select guide star
  Expose data image
  Begin slew to next target
  Download image (closed or open loop) while slewing

• Buffer fields are selected based on one of three criteria:

  Ecliptic-area survey fields at the RA of the last explicitly scheduled target; “random.”
  Orbital search for TOO - typically high ephemeris uncertainty MBO’s, also NEOCP/NRO page hits.
  Low elongation surveys.

• Pre-slew target tests (simultaneous):

  low altitude west (skip)
  low altitude east (park wait/buffer restart)
  low declination (skip)
  sanity scheduling (alt < -25, skip)
  obs per - only image if needed
  position exists (skip)
  Force triangular noise (slew offset +120d PA to last)

• After targets and buffers are depleted, automated fields are selected on following criteria:
  tds/exptime = images/3; ((images * exptime) / 2) + meridian = RA of survey fields. (Plain old survey fields start east of the meridian, and when we are done observing them, the scope is pointed an equal distance west of the meridian; hence we get the target fields while transiting.)
  TOO search. (Targets of opportunity can be searched for in a database; e.g., high ephemeris uncertainty objects can be culled from ASTORB.)
  Low elongation. (Targets of low solar elongation may be observed, since these are not well observed by the surveys.)

• Modifications to automatic scheduling:
  Keep fields away from galactic plane (user-defined tolerances for both summer and winter zones)
  Keep fields away from stationary points, if a survey field (move fields toward opposition point).
  Keep away from the moon.

• These tests are not looped but modify one another sequentially (no lockup in case there are contradictory requirements; simple to code).

Guiding Subloop

Two Methods:

• Medkeff (with Warner): Hunt around for guide star using exptime, then exptime * 2, then shift in a 3x3 grid. 99.6% acquisition rate; avg. time to acquire 0.86 minutes.
• McClusky (adopted by Medkeff): Pick a guide star from a star catalog. 99.91% acquisition rate avg. 0.2 minutes.

Both subroutines are free. The McClusky routine is now used in 701's Acquire script, with Warner's hunting method as a fail-safe backup in case the catalog star doesn't exist.

Night-End Activities

Park & Dark

• Generally 9 darks, all of them taken during twilight or daylight
• Turn off CCD cooler - crash or ramp down

Operational log has been made

• Pointing error for each slew, time of each hardware device event, results of loops and tests, etc

Additional Capabilities

• Cloud sensor: No stars in pointing images = clouds.
• Inter-observation clock updates w/Dimension 4 or AboutTime
• Delay exposure start to integral second for platforms that require this capability
• ???

Effects:

• Of robotics in general, increased observation:

  670 Camarillo:
  Before Robot: 512 (1998)
  After Robot: 1067 (2000)

  701 Junk Bond Observatory:
  Before Robot: 72 (part 1998)
  After Robot: 1068 (2000)

• Of additional automation to reduction system: (see Canonical List of Stupid Astrometry Mistakes)
  2 Errors Pre-Robot: 2.6% observations
  2 Errors Post-Robot: .0018 % observations

Observer vs. Script “smarts” plotted against length of run

A whimsical slide meant to show that the robot does not get tired and begin to make stupid errors, while the lecturer, at least, does.

Typical scripted image

This is a more or less "typical" image taken by 701's control script, in a barren area of the sky. The image quality from the robot is as good or better than can be consistently managed by human operation.

Exceptionally Noisy Image

Occasionally, things go wrong. In this case, I set the robot to use a temperature of 20C for the camera; I meant to set it to -20C instead. The result is an image with a lot of dark noise. Nevertheless, our target is present and PinPoint was able to reduce the image.

Exceptionally Noisy, Moonstruck image

We were going for a second night on a new asteroid that happened to be fairly close to the moon at the time. The asymmetrical haze and the loop on the left are caused by inadequate baffling in 701's telescope. Still, the asteroid is present and PinPoint was able to reduce the image.

Current & Future Deployment:

• At least 16 telescopes from .3 to 1 meter based on this reference; all but one continent
• Up from 4 one year ago
• Have fielded one inquiry per day since August 2000 Sky & Telescope article; perhaps 40 of them serious and still outstanding.

Users not at 701/933 (as of May, 2001):

• Nine are professional installations in developing nations.
• Four are educational observatories in US and elsewhere.
• Three are amateur or self-funded installations.
• In use for asteroids and for variable stars.

Emerging Trends

• Further deployment of this general technology in small professional installations, especially in developing nations.
• Further deployment in amateur installations as “no coding needed” realization grows.
• Additional erosion of Medkeff’s resources providing freeware, free support, and free dogmatic advice. (But it is fun!)
• Further dissipation of Medkeff’s time adding all the new features you will shortly suggest this “needs.”
• Further Medkeff dismay over the bugs that will continue to be “found.”

Resources:

• Robotic Observatory mailing list & web site

Robotic Observatory home page.