Ecosynth - Comments

Erle Ellis on Rising Popularity of the R Programming Language

Sun, 13 Feb 2011 17:13:33 -0500

Did you notice we have a full site reviewing R-Books at : http://ecotope.org/blogs/page/R-Book-Review.aspx

Garrett on Easystar clones- which is best for Ecosynth?

Tue, 25 Jan 2011 01:30:26 -0500

I see no problems with using the Hawksky as an alternative to the EasyStar. We could just tape the ailerons into place so they don't move during flight. I know that the ArduPilot board has a fourth port for aileron control but from what I've seen so far I don't think we really need ailerons. I'd only consider using them if it seemed as though the plane was unable to make sharp turns. The only thing that I can see us gaining from the addition of ailerons is the chance for one more thing to go wrong, I can't really think of any real practical differences.

I also think the AXN Floater has far too little interior room to be considered as an alternative. Even with the generous amount of space in the EasyStar's nose the electronics are a tight fit.

Chris on Easystar clones- which is best for Ecosynth?

Mon, 24 Jan 2011 18:37:51 -0500

Both the Hawksky and the AXN Floater(see comments) look better for us than the EasyStar...

Has anyone spotted advice on whether a "full house" plane has significant advantages for UAV stability? I understand that for beginner RC usage it's hell because nobody can keep track of 3 DoF, but I havn't seen much on the practical differences between autopilot with aileron-rudder-elevator and autopilot with rudder-elevator.

Chris on New Website for Micro Aerial Projects

Mon, 24 Jan 2011 18:18:53 -0500

IMHO:
It would certainly be worthwhile to look at a heavier-lift machine (up to a maximum of 1kg non-battery payload or so) for multi-camera IR-visual sensor fusion, for wide-angle lenses on mirrorless cameras, and for camera stabilization methods. This price, though, is obscene.

One can scale up a multirotor for relatively little money, since so much is spent on the guidance. Longer arms and bigger props & motors might even work on a MikroKopter. In theory, large slow-rotating props should be one of the few things that might result in longer endurance for a given weight:battery ratio. I think an Arducopter prototype or three, a mapping workflow for the EasyStar/clone Ardupilot, and a larger glider Ardupilot would all be viable platforms to experiment with this year as a supplement to the Mikrokopter. We could do all that and a whole lot of other expensive projects for the price of a Microdrone.

Jonathan on From Wired UK: Kinect hack builds 3D maps of the real world

Mon, 24 Jan 2011 17:34:51 -0500

I saw this this weekend too off another tech blog. Here is the creator's original post:
blog.decoratorpattern.com/.../

My first thought, was "Oh look, Meshlab." I also saw out there a blog about a stripped down Kinect box on the way that will be for the DIY "hacker" crowd, can't find it now, and there is talk of an official Kinect SDK: www.itproportal.com/.../

How does a kinect vision dataset compare to a terrestrial LiDAR or a walking - upward LiDAR for measuring canopy structure, stem diameters and stand density? The specs on this equipment put it at < $1000 and the builder here states: "These points are all in real-world coordinates and could be put, say on Google Earth and appear in the right place. "

It would be fun to talk to Noah about this w.r.t. PhotoCity.

Jonathan

Chris on From Wired UK: Kinect hack builds 3D maps of the real world

Mon, 24 Jan 2011 17:28:43 -0500

Or several. Kinects have a very narrow field of view, and you can fit at least six into a single hemisphere with no overlap, or you can rotate them, say, on the end of a walking stick. With only a 3m range, they would be a limited way of doing ground observations, but a potent one.

The challenge is in the coding, and in the bandwidth. A single Kinect basically saturates a USB 2.0 bus, and most motherboards have only two buses (regardless of the number of ports). I did locate a PCIE card which has 4 true buses, rather than a hub that is only capable of 480mbps total: www.newegg.com/.../Product.aspx , but I'm still looking for a laptop version.

These would be interesting to use for forest ground observations in groups, coupled with a bunch of ultrabudget netbooks.

Here are some other mapping projects:
www.ros.org/.../RGBD-6D-SLAM
http://www.ros.org/wiki/ScaViSLAM

There are also specialized aerial applications being investigated:

www.ros.org/.../Quadrotor%20Altitude%20Control%20and%20Obstacle%20Avoidance

For now, it's probably worth it to get one retail Kinect and one PrimeSense Developer's Kit, to see if the software is compatible and if the higher resolution camera helps with anything.
http://www.primesense.com/?p=1132
diydrones.com/.../kinect-sensor-devkit-is-200

Jonathan on DC police using drones

Mon, 24 Jan 2011 17:21:07 -0500

I think it is police in Texas that are doing this. So when do we go from using toys, even expensive toys, to something more, or does that distinction even matter? I think that even the MK's can still be considered hobbyist aircraft. Good thing we stay so low!

Jonathan

Garrett on Easystar clones- which is best for Ecosynth?

Thu, 20 Jan 2011 23:05:13 -0500

Yeah, I've extended the rudder by around 3/4 of an inch using a small piece of Depron foam prior to any of the test flights. The ArduPilot uses a proportional–integral–derivative or PID controller to determine how much rudder is required to keep the plane level and on track. In a PID controller there are constant gain values for each of the proportional, integral or derivative terms, the values for which can be changed in the code as needed. In short the proportional gain is multiplied by the input (amount of error) to create a value for output, which is related to the amount of rudder deflection. In order to get straight flight paths we just have to change the gain values and do many test flights to see how the plane responds.

Jonathan on Easystar clones- which is best for Ecosynth?

Mon, 17 Jan 2011 18:57:25 -0500

Fixed DIYDrones link.

I think that the unit Garrett is flying is the Multiplex EasyStar described first on this list. One thing that struck me was the comment about the small rudder. I wonder if a larger control surface might be necessary for helping the autonomous system make tight turns or u-turns when flying a pattern? Based on the telemetry that Garrett posted about last week, it makes me think that the system is trying to make tighter turns than it is able to, resulting in overcompensation or over-shoots.

I can't wait to catch up with Garrett in the field and see his machine fly some longer legs.

Jonathan

Erle Ellis on First Autonomous Flight w/ EasyStar

Sat, 15 Jan 2011 20:27:15 -0500

Very cool! Congrats Garrett!