I am returning from a very good trip to UMN and wanted to send out a quick precis of what we worked on and talked about. Minnesota has long been one of the most advanced and demanding of all the NEES equipment sites. They have what's called the Multi-Axis Substructure Test facility (MAST, http://nees.umn.edu/) a large structural test lab. They have an upcoming test (PI is Cathy French from UMN) where they'll be testing an asymmetric 4-story reinforcing column; it'll be the first experiment of a new laboratory. Currently scheduled for May, though that is uncertain due to many dependencies. (See http://nees.umn.edu/projects/nsf_ms-0324504_project_summary.pdf for more on the experiment) For this experiment, they have: DAQ: - 256 strain gages - 192 voltage inputs (mainly string pots and LVDTs for position measurements) - Between 40 and 200 Krypton inputs (depends on MRI funding for hardware) that return a 3D position for each sensor - 8 4 megapixel still-image cameras on custom robotic platforms - 10 video cameras, all with PTZ. 8 on robots, and 2 permanently mounted with views of the lab floor. All the cameras will be streamed, but only the 8 robotics will be archived. - 10 channels of audio from the floor Sample rates are low thankfully, typically less than 5Hz. Tests are 2-3 days long, continuous. The upcoming experiment has collaborators at Iowa state and Puerto Rico (not sure about the second), so they're working very hard on collaboration tools, control systems, telepresence, and client interfaces. Iowa has prototyped their interface, I've attached a powerpoint file from Jian Zhao that shows what they want to do: They want to run a data turbine locally, mirrored from the ones at UMN, and pull data from that. I was able to get mirroring working, but network problems at UMN prevented a thorough test. (Turns out that the campus net admins have run the MAST lab through the border packet shaper, and it wreaks havoc. This will be fixed. Kudos to Todd Shechter for figuring this out!) Questions from the visit: 1) MPEG-4 work at UMich - status/deliverables/ETA? Chuck? 2) Krypton support - I've just reopened communications with Luc Wens at Krypton, and we're arranging a video conference. 3) Viewing old data - are we proposing to use Chef, or RBNB? 4) Audio - what format should we support on disk? (I think WAV, for simplicity/portability) 5) Davis and UMN both want some sort of e-notebook that supports offline use. We could implement this with Nestor's notebook, since its file-based, but resynchronization is a hard problem. A good idea, though. 6) Might want to ask Creare to reimplement the C interface to RBNB; much interest at UMN. 7) Will gridauth also integrate with the TPM? (I think so, flexTPS will support, yes?) 8) We need to ask Cherry at ORST how much of Ken's time we can get; I want to take and use his applets for RBNB data and video. Can we use him as a developer? 9) Need to get cracking on metadata-in-rbnb. Directions/observations: 1) jfreechart is too CPU intensive (see http://www.object-refinery.com/jfreechart/applet.html and watch your CPU uses), so I think we should look at SGT instead. See http://www.epic.noaa.gov/java/sgt/JRealTimeDemo.shtml 2) We did some quick tests with data turbine mirroring, parent/child and shortcuts. We need to do more, and make sure we understand how this works and when each method is appropriate. For example, datavideogather did not behave as expected when we tried to extract parent/child video. 3) We did a couple hours' testing with the audio/video code from Creare, and it worked better than expected. More testing here, and an always-running source for coding, are required. 4) I went over the idea of event markers, and they like it. 5) Once we're happy with the audio code from creare, we'll need to write code to support it, and maybe pay creare to integrate it into the plot program. 6) For every rbnb feed, we need programs that will save and reload the data, without loss or translation. Programs we/they need, with notes as to developers: 1) Algebra/offsets. We need a program, preferably an applet that supports algebraic manipulation. For example, to be able to say 'a=b+c-0.7' where a,b, and c are RBNB channels. This would enable live offset correction, math for simple calculations such as rotation, shear, etc, etc. This is probably an SDSC task. 2) RBNB applets for data+video, in sync. KenF at ORST has a first pass at this that he's willing to share, so we should start there, and maybe try the SGT toolkit above. Also need XvsY and units. This is a big task, and I'm primary on it. Help appreciated. 3) JPG archiver. Right now, we have no way to save video out of the turbine without converting it to MOV. This is a super-simple program to iterate over a time range, saving all images as discrete JPGs, with the filename equal to the RBNB timestamp. 4) JPG reloader. Given a time range, take the output from #3 and reload it into the turbine. Pretty simple code. This and #3 are probably Terry. 5) Units. I've sent a proposal to Terry, Chuck and Drew. Think we implement and test this very quickly. Once we pick a method, we need to add support to all consumer code as well, and ask Creare to add it to their plotter. 6) EXIF scriber. Chen at UMN is going to write code that writes metadata into the standard JPG header, aka EXIF. Timestamp, exposure info, etc. He's also going to write the camera location there as well, since their cameras are on robots. I proposed that they use the comment text field, and put in there X=value Y=value Z=value P=value T=value Z=value Simple, easily parsed by existing EXIF tools, and also bound into the file itself. He's going to try and do this for both still and live video feeds. Other work to do: 1) Collaboration. I need to get a bulletin board/forum up ASAP, and also something like sourceforge/Savane/freshmeat up ASAP. We must be able to easily collaborate with remote developers, and the current system of CVS+mailing list isn't sufficient. 2) E-notebook to repository. Needs to work! 3) Help UMN with ELN. They need to start using one, and have no experience in doing so. I think the nestor notebook will work well for them, but they may need help in setup and sample data entry.