The initial experiment involved an Arduino, Nootropics Video Experimenter board, and an Amani GT plus some code written in Processing.
Living south of downtown San Jose gives us a fantastic and regular view of aircraft on approach. Being a radar-systems engineer my natural reaction was “I must track them.” After months of searching for an affordable gimbal/servo-based positioning system, a night of star-gazing made the solution obvious. Log onto Ebay, wait four days and my $100 Meade antenna positioner arrived, complete with integrated servos and positional encoders.
I should state here, due to legal reasons, I have no intention of tracking the aircraft via radio waves. Instead I would like to test a video tracking system, which would augment a radar when the target object is low enough in elevation to cause multi-path issues for the radar.
To be honest the Meade Autostar is better suited for tracking slow
moving deep space objects instead of that which is considered more
“local.” At least I can only assume it to be true as likewise I have
yet to have success with the unit in the former arena. However the
hardware is there: two-axis motor control with positional-encoder
feedback, all via UART. The problem is, however, that the hardware
interface is via a UART to a controller that is well documented to have buffering issues which ultimately leads to processor hangs and motor runaway.
More problematic is that there are only a few speed settings
available via the uart; FAST! slooow, stop, reverse time-travel. The
video below shows attempt at putting aircraft in track via a simple PI
loop. The tracker attempt to lock, overshoots the target then hangs.
Subsequent attempts to bring the target back into track succeed
temporarily and ultimately results in a runaway gimbal.
The next step is to open the Autostar, tap into the DC motor wiring,
and provide a nice dynamic range of control speeds via the Amani GT and
an H-Bridge PMOD. The UART connection will be maintained for positional encoder data.
It should be noted the control-loop bandwidth will then be limited by
the Arduino/Nootripics assembly. While the video experimenter is a
fantastic proof-of-concept tool, the design will eventually be migrated
to an FPGA-based system where data can be collected and processed in
real-time and passed to the host machine via ethernet. Moreover the
control-loop itself will reside in the electronics package; the host PC
will only provide a user control interface and data display.
The electronics package consists of the Nootropic shield interfaced
with and Arduino. To control the telescope positioner, the electronics
package must communicate with the Autostar via RS232. An Amani GT CPLD shield is used to bridge between the Arduino UART and a Digilent Pmod232.
The Amani is also being used to handle an issue I am having with the
video shield/Arduino stopping the video tracking arbitrarily. Serving
as a watchdog, the Amani looks for a heartbeat signal and resets the
Arduino when loss of heartbeat is detected.
Of course this configuration is just a method of rapid prototyping,
for proof-of-concept and experimentation. The VTE issue must be ironed
out eventually, however to get the tracking algorithm development going
the Amani-reset will serve as a patch in the short-term.
Progress thus far has involved sitting the unit in the window and
observing the video tracker’s ability to resolve and track aircraft
while I steer the gimbal manually. With minor code revisions to the
sketch provided at Nootripic Design, I was able to develop an error
vector, seen below, which will serve as the input error to the tracking
Currently I am ironing out the Autostar commands I found online, some
if not most of which do not seem to work. If anything I’ll need to
sniff the comms between the Astrofinder software on the PC and the
Autostar. It should be noted that the PC software is not involved in
this project and is just used for troubleshooting and communications
study. A PC will be included in this system to provide graphical data
to the user as well as a general control interface. The biggest issue I
have is not being able to control speed. If necessary I’ll open the
gimbal itself, cut into the motor controllers, and take control via the
I do have feed back from the positioner in the form of encoder
counts, which is vital for closing the tracking loop. Currently I use
the building across the way for boresiting the unit and calibrating the
The Meade ETX-60AT itself is a beginner’s telescope, better suited
for terrestrial viewing. As it is integrated into the controller base, I
cannot remove it without compromising the internal optics, thus it
serves as platform for the camera to piggy-back on. It does open
possibilities of integrating the telescope into the system, for
Even more interesting, and a definite DO NOT ATTEMPT, would be to
integrate a laser range-finding system into the telescope while the
camera provides the video tracking. This solution would provide
fully-integrated TSPI data for under $400.