PolyOculus

UCF Astrophotonics Lab · CREOL

Aperture is expensive. The cost of a conventional telescope rises steeply with mirror diameter, which is what makes very large single-aperture instruments so rare. PolyOculus takes the other route: link many low-cost, commercially available telescopes together with optical fibre so the array behaves as one much larger collecting area.

Bar chart of cost per square metre of collecting area against telescope diameter. Standard telescopes range from 1.3 to 1.7 million dollars per square metre across 1 m to 30 m apertures, while PolyOculus ranges from 150,000 dollars at 1 m down to 100,000 dollars at 8 m and 30 m. — Cost per square metre of collecting area, PolyOculus against conventional telescopes. The advantage widens with aperture. Figure: PolyOculus programme, CREOL.

The science case

The array's driver is a redshift drift experiment — watching absorption features in the spectra of high-redshift quasars shift over time, a direct measurement of cosmic acceleration rather than an inferred one. That needs an enormous amount of collected light and an extremely stable spectrograph, fed through photonic linkage and calibrated against a laser frequency comb.

Concept diagram of the redshift drift experiment: light from a high-redshift quasar is collected by a low-cost array of autonomous commercial telescopes, carried by photonic linkage to an ultra-stable spectrograph calibrated with a laser frequency comb, producing a spectrum with hydrogen absorption and emission features and metal absorption lines. — The redshift drift concept — array, photonic linkage, frequency-comb calibration, and the resulting quasar spectrum. Figure: PolyOculus programme, CREOL.

What I work on

I own the telescope-node software. An array only pays off if every telescope in it acquires and tracks its target without someone standing at each one, so each node has to run reliably and hands-off across the whole set.

The nodes

Each of the 8 telescopes is driven by a Jetson Nano that runs its mount, guide camera and fast steering mirror. I brought the Nanos up on mainline Linux, replacing NVIDIA's end-of-life Ubuntu 18.04 image. Every node is rebuildable from one repository with Ansible, so no node's setup lives only on its SD card.

Guiding

Centroid-based autoguiding reads the star's position off the guide camera and turns it into guide pulses for the mount. In simulation it cuts tracking error from 3.01″ to 0.29″ RMS. A closed loop on the piezo fast steering mirror holds a laser spot to 0.2 px.

Numbers

8: Telescopes in the array, each on its own Jetson Nano.
0.29″: Simulated RMS tracking error with guiding, down from 3.01″ unguided.
0.2 px: Laser-spot hold of the closed-loop steering mirror.