mlml logo

nsf logo Powered by Blogger.
Thursday, November 15, 2007: Where we stand
Today is Mindy and Marcus's sad last full day, and one of the first we've had Rusty and BLee along, so we were a bit crowded in a a dive tomato for our video ray deployment.

Nick collected data on the sewage outfall pile for his capstone project, and the rest of us got practice deploying and driving the video ray so we'll know what we're doing after Marcus leaves.

I thought the middle of our time here would be a good time to give an update on SCINI's technical progress and the issues still facing us, so the rest of this post will be a some-what technical run through; You'll either find it RIVITING or BORING.


So far our major successes with SCINI have been:
  • "Medium" (high definition grade, good enough to pull photos out of) video capture
  • Reasonable motor and positioning control and excellent tether dragging/current fighting power
  • No bottle leaks or floods down to ~150ft (50 meters)
  • A full (time saving!) deployment through a 10in diameter hole drilled at New Harbor.
  • A successful dive with SCINI and the video ray in the water
We currently operate with two similar ROVs (SCINI I and SCINI II aka SCINII aka SCINEEEEEEEE). The electrical distribution systems have been running smoothly so far, though we haven't really tested them (they haven't caught on fire or stopped working).

Most of the science video collection and exploration we did with the video
ray at new harbor, both because of manuverability reliability (SCINI had
broken motors and an unexperienced pilot, and it's design doesn't make for
as smooth manuvering yet). video ray is great, very easy to get everything
rolling, but needs a larger hole melted out (long time) and it's video isn't
going to get any better than it is now.

Mechanically, we've had a number of motors fail. We tried filling our
casings with mineral oil, but 50% of those have failed as well. We've been
able to swap out entire thruster couplings (the white tubes in between
bottles) easily so far; probably the worst part of a motor failure is how
long it takes for the pilot to realize they're driving poorly because of
mechanical failure, not just carbon monoxide poisoning.

The tethers haven't been a show stopper, but a few connectors have broken or
leaked, and we're also having some network troubles streaming video over the
longest lengths; maybe the elphel camera could be tweaked to work better in
these conditions? That would be treating the symptoms, not the cause.
Bouyancy in general has been slightly positive, but that's
good because it keeps SCINI slightly nose down which gives us a better
view without leaning on the thrusters.

For controls, we've been sticking with the simple python software I wrote a
while back and the 6 degree of freedom controller. I'd like to complete the
MOOS software system instead of hacking on more features (like dual
joysticks) to the simple program.

The camera situation is one of the top priorities right now. In general it's
not super smooth to playback because it's so much data, i'd like to tweak the
laptops so it plays better. It's also been hard to get it to display super
high resolution video reasonably (fit to screen, almost crashes computer); I
think I can improve this with a couple hours fiddling about.

Only SCINI (not SCINII) has inertial and depth sensors set up right now, and
i'm not really doing anything with that data... it would be really good to
at least get visual feed back of tilt and rotation speed for the pilot, but
that will take some calibration and software writing.

Our collected video has been getting backed up and i'm working right now on
transcoding it to a more workable file format. Some better solution for
anotations and screen grabs would be great, MBARI has software that i'd love
to get running and integrated. We also haven't done any editing and few
clips, I'd like to streamline that process so we can publicize our results
and more importantly start doing science with SCINI.

A growing pain with the camera is that the LED lights are thermally
connected to the sensor board, and after about 5 minutes the sensor gets up
to 55 celsius and the images get noise, errors, artifacts, and eventually
get corrupted. Either cooling or thermal isolation are required (no fan in
that bottle right now).

The LED reflection cone I built is working pretty well, you can see the difference in these two stills, both with lights turned up high:

Without light baffle

With Light Baffle

As is it is a pain to switch camera modes, take snapshots
at a different resolution, change exposure settings (I usually turn off auto exposure because it varies the frame rate at the long exposures we use, and I haven't figured out a way of compensating for this). If I have time I would love to write a MOOS client for the elphel to manage these settings from the controls interface, and log information about the video at the same time (eg, frames with navigation data).


I believe camera improvements are on the way - there are several pieces of software that will be ready soon that will improve video recording. We also have now software for recording on a micro hard drive, within a month or so we'll have an option to record on a CF card - hope it can fit on board SCINI.

"Overheating" of the sensor may also have a software solution - that problem can be caused by changing of the sensor output drivers parameters when hot - that can be compensated by re-adjusting the phase shift in the FPGA. There is a description on our wiki how to change that phase.


You - and your camera - are amazing! We never would have thought there might be a software solution and are excited to give it a try.

Hmm, good 4 ya. Congr.

Thanks you all

This comment has been removed by a blog administrator.

This comment has been removed by a blog administrator.

Post a Comment!

Subscribe to Post Comments [Atom]

<< Home
Subscribe to Posts [Atom]

This material is based on work supported by the National Science Foundation under Grant No. ANT-0619622 ( Any opinions, findings and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation.