I was fortunate enough to stumble across a project at school that has
intereseted me, and I've began development on some new robotics projects that
I'd like to discuss here.
One important portion of botanical research involves surveying and collecting
plants from their natural habitat. A herbarium (pl. herbaria), is a collection
of plant samples. They usually consist of a bunch of file cabinets containing large
peices of cardstock that have a sample pressed to them and relevant information
written down. If the plant is in danger, the physical sample can be omitted,
and the information about the plant will still be useful. The field
botanist will write down as much information as possible, such as surrounding
species, physical details of the plant and position (soil composition, grade,
sun quality, location, etc.). These physical files are digitized. Because
herbaria have existed for over a hundred years, not all digital entries
contain a GPS coordinate.
Samples will be dealt with at whatever organazation the collector belongs to
(Cal Poly, Yosemite NP Herbarium, UCLA, etc.) and later combined into a huge
database at the Consortium of California Herbaria. This is hosted at UC
Berkeley. There are a few hoops before getting the entire data set, but you can
search it online at ucjeps.berkeley.edu/consortium/.
There are a few issues in the pipeline, and finding solutions to these Special
Problems will be a goal of mine next quarter. (I even get a faculty advisor and
credit for my work.) Currently, the entire set of Cal Poly's data is in an
Excell spreadsheet. Each herbarium has a different data housing style, which was
designed to give each institution control and resposibility of their data. When
the Consortium recieves a new set of data, they have scripts to convert each
institution's data to the Consortium's database system. Then, if other
institutions comment on the entry, that info never comes back to Cal Poly. There
are entries dating back to 1880, and might be located at "Sierras", not a GPS
Lastly, I am hoping to work with people smarter than myself to search through
the data and use the massive data set to answer questions such as: are there
plants where we didn't expect them to be? How have things changed over time?
Where haven't we looked and should be?
That will be a self-guided class for school. The other project I wanted to write
about will be worked on in free time, and is some new robotics projects. I
recently picked up a Raspberry Pi, and it gives a much more powerful
programming environment in a portable, low cost, low energy computing unit. With
$30 you can have a Linux box with 1080p HDMI output, with a bit more you can get
a keyboard and HDMI dongle and have a full functioning system. The GPU is
cheap, but GUI's are always optional.
I am currently collecting parts for a quadcopter, or possibly a hexcopter.
Multirotors are increasingly popular due to ease of control and stability. There
are relatively cheap flight controllers that use user input as well as
gyro/accelerometer data to stabalize the flight. The user can focus on autonomy
or interesting flight details rather than low-level rotor control.
The projects I want to investigate include filming (and stability),
search/rescue and surveying applications, and speech based expert systems. These
are difficult tasks, so progress will be tedious.
Well, I usually loose my drive to write around now, so have a good day.