I got to spend some time at Hanson Robotics a while back, talking to David Hanson and his staff and shooting photos for Robot Magazine (see the Jul/Aug issue for more). My hope is that the photos will give you an idea what a typical day working at Hanson Robotics is like. The day I was there, everyone was preparing androids and other robots for an upcoming TED conference. In the photo above Bill Hicks is integrating an eye assembly into the head of Hanson’s newest android, known as Bina. Several other projects were underway at the same time, including preparation of a Zeno prototype and fabrication of a new mini Einstein toy robot.
To see the full set of photos visit my Hanson Robotics Gallery on flickr. Also, don’t forget to pick up the Jul/Aug issue of Robot Magazine at your local newstand if you haven’t already got it!
These are just a few of the photos so be sure to check out the full Hanson Robotics gallery over on Flickr for wider shots of the lab, more photos of Zino, Einstein, and Bina androids, and lots of other cool stuff from my day at Hanson Robotics!
What free time this month wasn’t sucked up by the Advogato migration was spent working on Tankbot GTR, the current DPRG group robot project. We now have the Mini-ITX mother board that Via donated mounted on the robot. We were doing our initial testing with an old 800MB laptop hard drive but it really sucked the batteries dry quickly. So I picked up an IDE to CF adapter and Martin donated a 1GB CF card. For the moment, I just used dd to move the entire hard disk content to the CF card. This is working surpisingly well considering we were running an old Redhat 9 distro intended for the desktop.
While most distros offer bootable CD images of one sort or another, almost none offer bootable CF card images. Many provide overly complex instruction on how to get their distro to boot from a CF card but few provide something as easy to use as a simple image file that you can copy and boot. Once exception is Flash Puppy, so I’ll probably be experimenting with that later this week. I’m begining to think there might be a real need for an embedded linux distro targeted at robot applications. And one that’s as easy to install as copying an image to CF card, sticking it in a motherboard, and booting.
Yesterday Susan and I braved the scattered rain to attend Dallas Artfest 2005. It turned out to be overcast and cool but only a tiny bit of rain fell. Susan found some interesting jewellery and it made a nice break from work.
I’ve been working with David Anderson in porting his robotics library for the MRM board to gcc 3.4.3. I’ve added a little code of my own in the process and I’m using Doxygen to produce pretty web-based documentation for the whole thing. We still have a ways to go but I’m hoping to release the completed library as a DPRG project in a month or two (under GPL of course).
We saw Hitchhiker’s earlier this month and found it sadly disappointing. It appears to have been adapted to the big screen by people who had no clear understanding of Douglas Adams brand of humor (or British humor in general). It almost appears they didn’t understand it was supposed to be a funny story and tried very hard to turn it into an action movie by excising anything remotely amusing from the script. Even worse, they frequently removed the setup for jokes but left in the punch line or left in the setup and removed the punchline, making the story incomprehensible (or at least very non-funny) for those who hadn’t read the book, heard the radio version, or seen the TV adaptation. The one redeeming moment was when the real Marvin made brief cameo appearance. I suspect everyone reading this has seen the movie by now but if you haven’t, save your money and buy the very nice DVD of the BBC television version which is a lot more fun.
We also saw Revenge of the Sith. What can I say. At least it’s finally over. It was better than the last two but I’m afraid the original version of the first movie is the only one that was really fun as a stand-alone story.
I’ve had some time to start working on mod_virgule again. My highest priority is porting it to the Apache 2 module API so I can finally ditch Red Hat 7.3 on the robots.net box. I’ve almost got a clean compile but it’s going to take a little while to get it debugged and stable before I switch robots.net over.
I’ve been doing some C programming lately on the MRM board, which uses a Motorola 68332 CPU. It’s not exactly a new chip but I think the last Motorola I did much development on was an HC11, so it’s definitely a step up from that. The board ships with an ancient GNU gcc 2.95.3 and equally old versions of binutils and newlib. So, my first problem was building a new cross tool chain using gcc 3.4.3. That done, I’m now begining to work on code that uses the TPU to generate pulses suitable for controlling R/C servos. Next up may be talking to one of those Logitech cordless PS2 controllers.
Between the recent office move and keeping up with herds of cats, I’ve fallen a bit behind in reporting news in other areas. Work is still progressing on the Ultracap power supply. After some further testing of the cell-balancer circuit based on the TI TLV4112IP high-output-drive op-amp, I decided that, while working flawlessly otherwise, it was drawing too much current. I’ve been rethinking the need for dynamic cell balancing and have re-read a lot of the Maxwell technical documents. My new theory is that balancing may not be as essential as I originally believed in a very low current application such as a robot power supply. I now think it may be possible to get by without dynamic balancing provided I protect the cells from over-voltage conditions during the charging cycle with something simple like some zener diodes. Finding very low voltage zener diodes proved a bit of challenge but I finally found some 2.7v 1/2 watt devices made by Fairchild that may work. I’ve ordered a few and will post results of the new tests soon.
Meanwhile, the DPRG received word from the IRS that our application for 501(c)(3) status has been approved, so we’re an official not-for-profit corporation now. And it only took 20 years to get to this point. I missed a couple of weekly meetings because of the move and during that time, Ron and some other members have been working on the A/C compressor at the warehouse and managed to getting it running, so we should have cool air for the next few meetings. We may have heat too shortly, which will come in handy in a few more months.
After prototyping a cell-balancing circuit for my robot power supply based on Maxwell 350F Ultracaps, I discovered the original choice of the TI TLC25L4CN low-voltage op-amp was not a good one. The little chip just couldn’t deliver enough current to balance the cap voltages in any reasonable amount of time. The peak output of the prototype was about 4 ma. So after spending a couple of hours searching for an op-amp that could operate at low voltage and put out a substantial amount current, I came up with the TI TLV4112IP high-output-drive op-amp. At last week’s RBNO, I built a second prototype and… it works! The new op-amp outputs up to 300 ma easily. Using the test circuit, I set it up with one Ultracap at 1v and the other at 2v. In little over a minute, the system balanced with both caps at about 1.6v. Next Tuesday, I’ll put together a more complete prototype with four Ultracaps and three cell balancers.
The cell-balancer has also proven to be a good way of trying out the GPL Electronic Design Automation (GEDA) package. I’ve been using the schematic capture program gschem primarily (screenshots) and have been asking lots of stupid, new-user type questions on the mailing list. They’ve been very patient with me so far and I’m begining to get the hang of it. It turns out good-looking schematics. Once I finalize the power supply, I’ll post a link to the GEDA files for anyone who might want them.
Cats in the garage
The abandoned mother cat and two kittens we brought home a few weeks ago are still with us. We know a lot more about them now. The kittens were much younger than we first thought. We’ll probably hang on to them until the kittens are a bit older. Susan has settled on names. The half-siamese mother cat is Sophie, the black kitten with the missing toe is Zippy (though she insists on spelling it “zippie”), and the tailless calico kitten is Callie. No luck finding a home for any of them yet. Callie is probably too emotionally disturbed to make a good pet but we have located a group called Barn Cats International that assists with finding homes for problem cats on farms and ranches where they can live with minimal contact with humans. Hopefully we’ll be able to find good homes for the other two. If anyone in the Dallas/Ft. Worth area needs a cat, let me know!
Nigritude Ultramarine contest outcome
The contest ended on July 8th. Thanks to everyone who linked to my site, especially, Bram. His link resulted in more hits on my site during the contest period than anything except Google itself. Feel free to remove those links now. I’ll probably leave the page up until the domain expires. I ended up in position 6 in the final results. The winner was a blogger who apparently won primarily through old fashioned Google bombing. Oh well, my site received a sort of honorable mention prize, the Judge’s Choice Award. My prize is one of those teeny, tiny “James Bond Stealth digital cameras” like you see on ThinkGeek. I’ll post a photo as soon as I get it talking to my Linux box (it only comes with Windows software but it has a USB cable so I’m hoping I can just mount it like a little USB drive).
We’ve been pretty busy at the office doing website design jobs. The lease on our office space is up soon and we’re deciding whether to stay put or move to new space. Even if we move, we’ll stay in Dallas and probably in the same general area.