Help! I need advice on a UAV for geoscience applications

[gue007]

My needs: I want to build a system that can conveniently disassembled into a backpack or carrying case so that I can carry it up mountainsides and into remote areas easily (I am starting with a GWS slow stick). Then I want to be able to assemble it and launch it to fly out and collect arial photographs and video that can be used for geological mapping. I hope to have two cameras: one looking straight down and a second looking back slightly. Using these two cameras I can generate images that are stereo pairs (assuming I know the aircraft's position, altitude, velocity and orientation for each photo). With stereo pairs I want to make 3D terrane maps (like topographic maps) that are useful for mapping things like earthquake fault scarps, land slide movements etc.

I also want to use the same system but with one camera for taking basic arial photos for various geologic mapping applications.

I want this system to allow me to fly the airplane out of sight around corners etc so it should have some kind of auto pilot set up to where I can specify it's flight path and where I want to start collecting data etc.

As you can see I have a fairly ambitious set of goals but based on what I have read it seems at least partially doable...what do you think??

I would appreciate any practical advice that you guys could give me especially on what I should buy in terms of cameras and the autopilot system etc. I have been building model aircraft all my life so on the airframe side I have some good ideas (especially since I come from the free flight community where light but strong is the way of thinking and low speed stability is important).

My present goal is to have a basic slow stick version flying by the end of July. With your help I can meke this happen...money is not much of an issue.

[macboffin]

I guess that your biggest problem will be always taking pics at the same altitude, so that edges of pics can be matched up.Otherwise you will have problems scaling pics up/down so they can match up to form bigger pics.
You may also have shadow problems unless all pics taken at the same time each day.

[gue007]

Hi Macboffin thanks for the response,
I hope to take care of the altitude problem by utilizing an autopilot that keeps the aircraft fairly level during data collection legs and by collecting several data sets for a given area...my problem is I can't decide which autopilot system to buy...there are several all sounding good but I can't figure out if one is better than the other (if someone reading this can clue me in on which system I should buy this would be helpful).

Shadows could be a problem but they can also be useful...low sun angle arial photography from light full size aircraft has been used for years to ID unmapped fault scarps...as you can imagine the low sun angle in the early morninngs and evening really highlights small scale relief (like fault scarps) in the landscape. If I am not using shadow I will be avoiding it by flying at the appropriate times for specific projects.

B

[danstrider]

If money really is of no issue, the Piccolo is a turn-key solution for flying autonomously. Once you get it in the plane, tuning goes fast. Plus, you can steal the state data onboard so you know exactly how the plane was pitched/rolled/yawed for post-data processing of your imagery. The LT is 110g, so maybe a littttle heavy for a Slow Stick, but if money isn't an issue, changing vehicles is relatively minor.

I've done long legs with a fixed downward looking camera mount and relying on the Piccolo to keep the wings "level" so the images weren't too distorted. That worked just peachy on days that weren't very windy. Since you know the aircraft orientation though, the "wings level" thing shouldn't be an issue, no?

Dan

[gooroo363]

hey, ya, I'm a Cartography and GI Science Major at Salem State in Salem Ma and i'm working on a very similar use.....I also work for a company that does aerial imagery and the problem that you might have is correcting the orthographic photos, there is alot of math and such that the images that need to be corrected by in order to get them orthographically correct.

[mlbco]

If you have research money to buy a complete system, my company has been selling ready-to fly UAV's that do many of the tasks you mentioned. Our aircraft are regularly used to do aerial mapping missions and we recently imaged over 250 sq. miles of Baja Mexico at 3-6 inch/pixel resolution. Our aircraft automatically takes overlapping digital pictures of regions designated by the operator. All the inertial data is recorded for each photo taken (lat/lon, roll, pitch, heading, height, etc.) We are developing software for automated stitching but many of our customers have had success using "PTGUI". Here are some example images taken at 3" per pixel. The larger mosaic is shown at reduced resolution and the cropped image is at full resolution.

Steve Morris
MLB Co.

[Tom Harper]

Very impressive!!

Tom

[mwhazard]

Gooroo points out an important issue. Unless the pictures you take are pointed straight down, you will incur serious perspective distortion. You can correct this in software or simply stabilize the camera mount.

As far as stereo goes, most stereo systems use cameras pointing in the same direction. One pointing down, one pointing back is not the same, especially since most stereo algorithms assume the principal axes of both cameras are parallel. More sophisticated algorithms (look up Hartley and Zisserman) may be able to deal with meshing images with rotations between the cameras, especially if both cameras are calibrated. Additionally, you may be able to use a kind of false stereo with a single camera if there is sufficient overlap between images.

Stereo cameras are typically used for short range observation since the parallax effect that lets you calculate ranges to an object becomes smaller the farther you are from the target. Increasing the baseline (distance between cameras) will help. You may want to consider putting a camera out each wing to spatially separate the camera centers, although this will definitely affect flight characteristics due to the increased moment of inertia. Especially on a vehicle with no aileron control (slow stick?) this effect may be significant.