Well, yesterday afternoon found me and the 84 year old "Pop" out at the Clinton flying field where for the second time we flew a 1/5 scale R/C model of one of these. In red and black it is pretty stunning.

http://www.aviation-history.com/travelair/mystery.html

No wind whatsoever and the landing was HOT. Nosed over at the end of the runway and that BIG cowl decided to come loose. Nuts. End of the day for that one.

But, I brought the Topcon robot and second task of the day was to proof those centerline road shots I created with a Locus in "continuous kinematic" mode atop the mag mount on the Subaru!
When I ran it, I merely measured the vertical height to the bottom of the Locus very crudely---eyeballed it in other words. Looks like 1.600 meters to me. After the fact I made a bit more precise measurement by using a 2 Meter pole laid horizontally across the roof rack perpendicular to the roof and measured from the ground up to the bottom of the pole on both sides of the Subaru and took the average. 1.518 meters. I then measured the dimension from the bottom of the pole as it passed next to the mag mount within mm's of the bottom plate of the mount up to the bottom of the Locus. So aside from:

1)My Big 240# frame loading the suspension and

B) Not considering the crown in the pavement

The vertical height works out in the neighborhood of 1.613M. I used 1.600M for the processing.

So how did the robot agree? Well, we can't say much about the horizontal when you try and stake out those "continuous kinematic" points, other than they are pretty close to the centerline of the road. As close as I could drive them anyway. A single mag mount and a Mercedes hood emblem from the junk yard mounted directly in front of me on the hood and putting the Locus directly over me on the roof should improve this ;-)
But the vertical you ask. Glad you asked. Would you believe that staking these points---with the robot and the 360 prism atop a 2 meter pole---gave me anywhere from "Cut 0.02 to Cut 0.12" with the average being 0.05 or so? Meters? NO. FEET! One epoch of data got those results??!! It should also be noted that this was a "radial" kinematic survey with the Locii. I had one on a tripod as the base and came off the bar for the initialization. There were no other receivers used and no adjustment was performed. Merely processed results.

Comments?

Dangerous Dave
Modified By Dave Huff on 10/25/2004 at 12:58 PM

So our EOL friend did as it was supposed to do or better........ is what you are saying I think.....

From my K work with Locus units, it is what I'd expect.....gonna get ya them trailer mount pics taken today....now where is that camera......

Jimbo

Hey Dave,
I promised a test a few months ago of epoch by epoch processing on a stationary point. I did the test but never got around to sharing.

Although the processor (AS 2.6) occassionally showed 0.2' error estimates or more the points were all within about 0.10'. It seemed like, as long as the solution was 'fixed' then you could count on solutions to the nearest 3 cm horizontal and 5 cm or less vertical. A healthy majority were cm level horizontal and vertical.

Again, this was based on single epoch solutions. The only point I see in having multiple epochs in a stop and go session is improved statistics on the site. I don't see that you will get better (more accurate) results with a longer session than a shorter one. This assumes that the site is good (no obstructions, no multipath). The real question when it comes to quality results with stop and go is am I 'fixed' or 'float'. Multiple epochs won't help a float.

That said, with our own work doing stop and go, Dad and I did a topo once where we were both running rovers at the same time. He was using 30 sec observations, I was using 20. I figured that with 300 points or so, that 10 seconds would add up to about an hour. When we got in, downloaded and processed, Dad's statistics were much better. I didn't have any floats, so we felt ok. But many of my shots showed error estimates in the 0.2'-0.25' range, while Dad's were very nice, his worst were in the 0.1-0.15' range. Same constellation, same conditions, different occupation times.

So I guess it boils down to how much do you want in the way of statistical good vibes or are you willing to sacrifice a poor error estimate in the interest of increased productivity. I think both have their place. But I do feel relatively sure that fixed or float is the big issue at hand in a kinematic survey.

Taildraggers are the hardest to land correctly. My last had so much zap holding the front together that by it's end of life the fuel tank was almost in the tail to get the CG correct.

I do have a few comments.

Height of Subaru- I checked the NGS Calibration site and found nothing, however the Subaru site had the height. I would weight this on age and spring sag to get a good approximation.

And you could use slightly stiffer springs on the left side to compensate for the "ballast" in the front seat. Although with the antenna mounted top center, then any shift is minimal at the axis.

The Mercedes emblem is probably not the best choice, unless you can drive like a mailman and sit in the middle of the car.

I would suggest a laser gun site mounted to the bottom of the GPS mag mount that points to the road. Might take a bit of work getting it squared and corners would be problematic, but workable.

Sounds like the results were pretty darned good in any case.

Deral


PS to Shawn-So it sounds like ol' Dad is still showing you up!...Age and treachery...Youth and exhuberance.. Something to be said for that.




Modified By Deral_ Paulk on 10/25/2004 at 5:59 PM

One of the seminar presenters last week has put some real effort into getting good verticals from his truck mounted antenna. Example: The difference between full and empty on the fuel tank is 0.010 meters.

Examples he showed of GPS trajectory from the top of the truck versus RTK on a pole were surprising. He has put some real effort into getting good results while sitting behind the wheel.