I am planning an attempt at this technique of GPS surveying in the near future (as an experiment) using LOCUS recievers. Has anyone tried this using the kinematic handheld? From the one source I have read on the topic of Pseudo kinematic, I have gathered that short, uninitialized static observations are made over relatively short baselines 2-3 times and that it will yield the standard kinematic accuracies without the limitation of maintaining lock. The observation times were listed as 5 min and a repeat under a new constellation. I am unsure if this short of a time is for L1/L2 or applies to L1 only. I suppose only testing will tell. Also no mention of vector length is given in the text I have so I will probably hold at about 2000' and less and see what happens.

Any input would be greatly appreciated.

Shawn Billings
SIT Texas

Speaking personally and not as an Ashtech employee, I think you'd have MUCH better luck with dual freq receivers in your experiment. With L1 Locus', I bet most of your baselines will fail. Basically, you're trying to get an L1 OTF fix, which you can do, but I think you need a decent amount of time with clear sky, probably half an hour or so.

Again, not the official party line, hey, I'm on vacation now!

Shawn,

Actually I've done something very similar. I wouldn't recommend cutting the times real short unless you can afford to go back if necessary but.....

After losing lock I have regularly used the equip. "in K mode" and recorded multiple 5 minute intervals (in the HP) for the same unknown point for a total of about 15 minutes, then continued on with the K survey. I'm not trying to set control doing this, just collecting topo features. Usually get pretty tight, mostly .1' or better.

And while there are thesis on the general board that it doesn't really improve the solution, I use 2 bases and one rover, and tend to look at the geometry so my "intersections" are similar to what they would be on a regular triangulation network.

Jimbo

I don't know if the "jeep" processing engine currently in Locus and Solutions handles pseudo kin. occupations the same way as Prism did in years past. There used to be a specific setting in Prism for pseudo, and I always thought that the software tried to process the ambiguity during the first short occ.,to a float solution or better, and then use a kalman filter to predict where the amb. would be when you visited it again later in the session, then refine the final ambiguities to a fixed integer solution...

Since Locus uses a different processing engine with less user intervention and more processing assumptions handled by the software, I wonder if that will work..Take the following case: You do an init on the bar, then do some topo work for a while, then lose lock...You go back to your last known good point and reinit, right? What if your last "known" point was really blown and you didn't know it? You would be then forcing an init on a point with "false" coords, and the following points would be blown because you forced a bad init...

I look at Locus and pseudo kin. in the same light...If you don't get the correct ambiguity fix the first time, Locus may use that bad position to force an init and then your next points will have the same issues.....Help, Bill.....b

I am wondering why you would consider 2 to 3 observations of 5 minutes using an unsure method when the accumulated time on each point is equivalent to a sure-fire static shot in good conditions.

Am I missing something, or are you just having a "xspearmint"?

Nearly Normal,

If you're asking me it's because the K mode software in the HP only counts 300 seconds...I think.

Jimbo
Modified By James Webb on 7/8/2001 at 8:41 AM

I see two possible advantages to this if it works. (I hope to try it next Saturday. Getting the time to play seems to require an act of congress for me.) 1 is that the three observations will be made under very different satellite geometries whereas a single static observation would not have a great deal of change in geometry. 2, we work in areas that seldom (if ever) permit a kinematic survey without frequent loss of lock problems, therefore this would (again, if it works) allow us to use the kinematic kit to survey in hard to reach places without regard to lock issues. I would not trust a single fifteen minute observation to be very accurate and believe that the redundancy in a three observation Psuedo kinematic scheme might offer more confidence in the results. I don't have a whole lot of faith that this will work, but I would like to find some use for the kinematic kit and am willing to at least give it a go.

Shawn

If I understand what you are going to test it sounds similar to something that Ben Remondi proposed a number of years ago. He certainly felt that it would work for the reason you stated, better satellite geometry. Unfortunately I don't remember the references to what he wrote about it at the time, maybe someone else does.

Shawn,
I have used pseudo-kinematic (but not for several years). This method is considered obsolete, and is not supported in current Ashtech s/w (Solutions). It is supported in GPPS, PRISM and WinPrism. Does it work? Yes, quite well. The biggest limit is the logistics of occupying each point twice about an hour between occupations.

Mr. G,
Yes, this was written abouut by Ben Remondi about 15 years ago. It was implemented in Ashtech s/w, and I used it quite a bit. The idea is that the processor treats the two observations as a static occupation with a gap of about one hour. This provides sufficient geometric change to resolve the ambiguities.

Regards,
---Brian

To all,

My question is what happens if you lose lock between the seperate observations ? Does this affect the solution ?

Jimbo

In my experiment I fully intend to force a loss of lock between each observation (which won't be hard). My plan is to do three quasi kinematic observations on several points (hopefully at least 10 to start) all tagged with a unique i.d. Then I will attempt another new trick I have learned and use the standard deviation of each observation on a point and do a manual weighted mean of the three observations. I may be on the wrong path here as there will be no correlation between the 1st 2nd and 3rd to do any sort of ambiguity resolution in my final "adjustment". But I am pretty sure that Congress (my wife) is going to allow me the opporitunity to attempt my test this weekend.

Shawn

Pseudokinematic (better called pseudostatic) is not dependent upon continuous signal lock. The way it works is that two (or more) short (1-5 minute) occupations are made about an hour apart, and processed as if they were a single static obs with a gap in the middle. As I previously explained, Ashtech Solutions (and Locus) do not support this. GPPS, PRISM and WinPrism do. The technique was propopsed by Dr. Ben Remondi in the mid-1980's, and further developed by yours truly (as published in GPS World circa Sept. 1990) and Ashtech. It has now been largely rendered obsolete by dual-frequency OTF. Good stuff, just not as good as what we have today.
Regards,
Brian D. Ewing, PLS

Pseudokinematic (better called pseudostatic) is not dependent upon continuous signal lock. The way it works is that two (or more) short (1-5 minute) occupations are made about an hour apart, and processed as if they were a single static obs with a gap in the middle. As I previously explained, Ashtech Solutions (and Locus) do not support this. GPPS, PRISM and WinPrism do. The technique was propopsed by Dr. Ben Remondi in the mid-1980's, and further developed by yours truly (as published in GPS World circa Sept. 1990) and Ashtech. It has now been largely rendered obsolete by dual-frequency OTF. Good stuff, just not as good as what we have today.
Regards,
Brian D. Ewing, PLS

Obselete??? You say it is considered obselete because of dual frequency OTF. I can see the reasoning behind that, except that the LOCUS is single frequency, so I would think that a viable "short-cut" technique using L1 only would be contemporary function considering that L1 is obviously not obselete (you still sell it and we still buy it). Does that make sense?

Shawn

Shawn,
Pseudokinematic is logistically difficult. Others at Ashtech decided t's obsolete, and very few of us ever used it. L1 (aka single-frequency) equipment is obviously not obsolete.

Clarification: Ben Remondi developed the concept. Kendall Ferguson implemented it in the software. I developed field techniques to make it viable, in conjunction with Ashtech developers to make it usable.

If you and others feel it's viable, I suggest using older s/w to see how practical it really is, then taking your case to marketing personnel.

Regards,
---Brian

Thanks,

I appreciate the info. I just read about this in a surveying text book the other day and thought it sounded interesting. I had heard the term used on occassion but had never really questioned what it was or the mechanics of it.

Shawn

I have a lot more faith in a single static observation than you have got.

I have never found a significant error in a static shot that solved well, but will continue trying.

Shawn,

Sounds real interesting....

I have much faith in my static obs. but I do see situations where the technique could be quite useful.

Brian,

Is there any "paperwork" readily available explaining the techniques you developed ?

And since I'm relatively new to the GPS game, are the old software packages you mentioned readily available over the net ? My initial search didn't show much relevancy...any pointers or links would be appreciated.

Jimbo

Jimbo,

Ellis Veatch and I used to teach pseudokinematic in Ashtech training classes, but most people found it logistically difficult. The basic idea is that with static, the reason the occupations are so long is not that a large quantity of data is required, but you're waiting for the satellite geometry to change. With pseudokinematic, you occupy point #1 for 5 minutes, move to point #2 for 5 minutes, and so on until about an hour has passed, then reoccupy #1 for another 5 minutes, #2, etc. This gives you (for each point) two 5 minute occupations about an hour apart (the same geometrical change as a 1 hour static occupation). The processing software treats it as a static occupation with a gap in the data. Simple enough if you use 1 receiver as a "base" and move the other around. We were using 3 or 4 receivers, all moving, but coordinating the observations via radio so everyone's observations were simultaneous, which let us compute all vectors between all receivers. You can see this makes the logistics a bit more complicated. It's not cost effective if your points are too far apart, though. We had points on paved roads at 1/2 mile intervals. As to where you can get the old s/w, I don't know unless you can find a long-time Ashtech customer or dealer who might have a copy laying around.

Pseudo works well, but it requires good field people who can handle the logistics. Most people would rather do something simpler. It does have the advantage of not needing continuous signal lock (unlike L1 kinematic).

Regards,
---Brian

Shawn Billings wrote "...(you still sell it and we still buy it).

Shawn,

I don't sell anything. Nothing. Zip. Nada. If you handed me a wad of cash for an RTK system, I'd turn you over to someone who sells them. The closest I come to selling anything is updating my resume. ;-}

---Brian

Shawn, Jimbo,

For more info on pseudokinematic, try http://www.ion.org in some of the old proceedings you'll find a few things:

ION GPS 1989
Kinematic and Pseudo-Kinematic Surveying With the Ashtech XII, Kendall Ferguson, Ellis Veatch, Richard Sauve, Mike Evers and Wendy Corcoran 35

ION GPS 1988
Kinematic And Pseudo-Kinematic GPS, Dr. Benjamin W. Remondi 115

I have some of the old stuff I wrote, but it's on my old 486 PC (not currently set up). Ben Remondi's paper from 1988 pretty well sums it up. This paper is what got me started using it.

Regards,
---Brian

Perhaps I should clarify. I have little faith in a 15 min static session. Never has the 5k timer gone off in 15 min or less. Now take that fifteen minutes and spread it around an hour to an hour and a half and you have some redundancy that is meaningful.

Brian, I didn't think you were in sales, but I was more eluding to your association with Ashtech more than your actual title.

Shawn

Brian,

"It does have the advantage of not needing continuous signal lock (unlike L1 kinematic)."

Exactly what I'm thinking about.

I'll start looking for that stuff tonite.

Thanks,

Jimbo

My sentiments exactly.

Shawn

Shawn (and everyone else),

I want to make it crystal clear that my postings on this and other message boards or other web venues are those expressed by me as a licensed surveyor.

I am not in any way a spokesman for, technical or sales representative of Ashtech or any other GPS manufacturer. When I post on GPS topics, especially on the Ashtech message board, I am doing so to express personal opinions, or to keep someone else from repeating my mistakes.

Regards,
---Brian

Brian,

And your comments and help are greatly appreciated.

Jimbo

This is kind of funny. I got on Yahoo and did a search for anything with "Psuedo Kinematic GPS" and about sixth on the list was a link to something that would process such an animal. Upon further review it was an Ashtech link to a product listing for Ashtech V 2.0. Apparently, when our upgrade software comes, the ability to process these Psuedo kinematic obs will be available. I still think it has some merit and am anxious to try it out.

I thought it was funny though that I was prepared to search the e-world over to find some software that could at least process it, and the circle came right back to Ashtech.

Shawn

Not a big surprise, as the pseudokinematic concept came from Ben Remondi, who did much of the early work on Ashtech software.

The new Solutions 2.x will not process pseudo, however. I'd guess you found an old link for PRISM or WinPRISM 2.0.

---Brian

Brian,

Check out the Ashtech site and read the different "release" notes. I believe it was under "specifications".

Qoute: "Designed to meet the needs of all surveyors, Ashtech Solutions supports processing of data from all Ashtech survey systems offered to date. All commonly used modes of survey data collection are supported including Static, Rapid Static, Pseudo-Kinematic, Stop&Go Kinematic and Continuous Kinematic." End Quote.

Was this accidental with it's inclusion of Pseudo Kinematic, or are we confusing terminology and possible application ideas?

I personally don't know. Any thoughts.

J.D.

p.s. I appreciate your contributions to our educations on this forum. Thanks.

Well J.D., you've read the marketing materials, and you've read my statements. I'll leave it to you to decide which is accurate.

Brian, JD,

It's easy........

Bill,

Is WinPrism available for us Locus users that would like to try Psuedo Kinematic ? And compare it to the Psuedo K function listed in the new V2.0 software ??

Jimbo

I think you should go right ahead with your experiment, but I say again you are gaining nothing.

When I have 8 sats, I get 5km alarm in 8 to 9 minutes, no trees.

How in the world is false-kino going to improve on this?

Maybe you arent aware that you can continue a loss-of-lock kinematic session after a single static, logging 20 seconds at the end?

Sometimes while staking a line I lose lock irreparably during kinematic. Then I do a static, re-lock on that point, and continue with kinematic 20 second observations.

How would your psuedo improve on this method, even if completely successful?

Wow!

Who would have thought that the subject of Pseudo-Kinematic mode of data collection would create such a long discussion?

Shawn,
I understand your curiousity in this mode of data collection. In the late '80s, when I was performing GPS surveys for a living using Ashtech equipment, this mode of data collection caught my eye also. Back then, the average data collection period was about 40 minutes or so (L1 only), so the prospect of being able to accurately establish a point with two 5-minute observations, separated by approximately one hour, was appealing. So, I performed a great deal of experimenting with the technique and did eventually incorporate it into a few of our jobs. We used it primarily in situations were we would have liked to use the Kinematic mode but could not due to obstructions on the project site causing frequent loss-of-lock. The data collection procedure was somewhat similar to kinematic. We would observe a point for 5 minutes. Leaving the receiver turned on but not worrying about maintaining lock on satellites, we would then move to the next point and observe for 5 minutes. We continued in this manner, observing as many new points as possible, until about 1 hour passed. We would then go back to the first point and re-trace our steps, re-observing each point for another 5 minutes and assigning the same series of site IDs. When each point was observed twice, we would end the data collection session by closing the data file. If we had more points to do, we would start over again with a new data collection session.

Here is a summary of my experiences:

1. If you are not careful, it is easy to get lost in your data collection process. Re-observing each point with an hour gap and being sure you assign the same site ID can get confusing, especially if you have a number of points going at once.

2. I used two 5-minute observations separated by 1 hour. On average, this provided me with about a 90% success rate, i.e. 9 out of 10 points processed successfully.

Today, using an L1-only GPS receiver like Locus, observation times are closer to 20 minutes. I question if the Pseudo-Kinematic data collection mode is worth the trouble. Taking into account the two 5-minute observations, the two setup times, and the extra time of returning to each point, there probably isn't enough time saved to warrant the use of this method, considering it's drawbacks. It's primarly for this reason that the Pseudo-Kinematic mode of data collection has fallen from grace.

Now, let me clarify a confusing point regarding whether or not Ashtech Solutions can support the processing of data collected using the Pseudo-Kinematic mode of data collection. As you are all aware, Ashtech Solutions was derived from Locus Processor. In essence, Locus Processor v1.2 was built upon to produce Ashtech Solutions v2. Between Locus Processor v1.20 and Ashtech Solutions v2, we made an "improvement" to the processing of Kinematic data that unwittingly resulted in our inability to process Pseudo-Kinematic data. I personally did not notice it until this discussion started here and I tried to process some Pseudo-Kinematic data in Ashtech Solutions. What did we change? Well, in Locus Processor v1.20 or older, if you assign the same site ID to different observations in a data file, when processed, these observations are combined to produce one vector. Some of you felt it would be better if we provided a vector for each observation of a point so that these multiple observations could be examined for repeatability. It was a good idea so we implemented this somewhere between Locus Processor v1.20 and Ashtech Solutions v2. Well, this messed up the Pseudo-Kinematic processing where you need the software to combine the multiple observations into one vector solution. We will need to look at this again to see if we can get the Pseudo-Kinematic processing back.

Well, that's the sum of

Bill,

Your post says everything except that Locus Processor v1.2 does Psuedo K.

Does it ?

Jimbo

Edit: If I put 5 minute obs on points and repeat an hour later using the same point number for each then Processor 1.2 "assumes" an hour long obs ? And that is how you get away with a short reobs. on a known point for a reint after a loss of lock ??! Right ??
Modified By James Webb on 7/12/2001 at 2:09 PM

First I would just like to thank Brian Ewing, Bill Martin and Ashtech for taking such an active interest in this board. It is one thing that you all produce excellent products and provide training, but that you also educate (as in the case of psuedo kinematic techniques) is highly commendable on your parts.

I suspected that Locus processor V 1.2 might be able to handle pk data because of what I have understood to be "simultaneous" processing which I had thought (hoped) might be as Bill indicated, that the various vectors with identical site i.d.'s would be combined in one vector solution. I see several benefits to pk in our work, of course there are obvious considerations that must apply when deciding betwee a strict static or pk. In the case of the Locus processor, since it is a fully automated processor with few procedures for "cleaning" rough data, I would think that getting 4 pk obs would allow the user to easily identify problem sessions, exclude them from the solution, and still have sufficient data to not only get a solution, but to also have redundancy. The 4 obs could be made in a 3 hr block of time, or could (I am guessing) be spread out over any convenient length of time exceeding 1 hr. Conversely, a 20+ min observation with bad data would just have to be eliminated.

Thanks
Shawn

Shawn, Brian and Bill,

Excellent topic and info. Shawn good thoughts about the software also.

I can understand why marketing might believe Pk is not viable. But with the advent of the low cost Locus system it provides a method of operation similar to K in places that continuous lock is not achievable ( and you don't have to spend $45,000 to GPS it). With a unit on a pole and a bipod the setup times become negligible.

And Bill's point about short occupation times for Locus Static is true, in many situations Shawns' point about having more than 1 observation is really the key factor to me.


Thanks for the input and the education.

Jimbo

Jimbo,

Come on up to Stumpwater, East Texas, this Saturday and you can play too. We're gonna stretch these puppies around to 6 or 8 known controls from our base station post and see what happens. After yesterday's little semi test of "sumokimenectic", I think it's worth investing a Saturday to just find out what we can expect.

Our thinking is that on our municipal route work (currently sewer line planning in the usual East Texas creek bottoms) that this procedure may help us to gather an average of 3 good points per hour during optimum sv configuration (and low pdops). Mostly points for control verification. In my little test yesterday I ran 2 sessions of five minutes each, one hour apart. The base station was left up for the entire session (A), the rover with two individual sessions (A and B). The sessions were processed first together, then individually by question marking first the "A" session and processing "B", then the reverse. Different output each time, all within reason. So, we're assuming the Locus Processor (1.2) is doing what we expect. We plan to hit points anywhere from 400 feet from the base up to about 3.5 miles from the base, with at least 2 sessions, maybe 3 if Shawn's other boss will let him play that long.

If anyone else is interested in the results we'll make it available.

Ain't this stuff fun?


J.D.

JD,

I'd love to ...but you remember when you mentioned something about being real busy kinda cuz poor planning..??
I think I'm gonna be cogo and drafting slave cuz it seems I promised a whole lotta things for Monday and Tuesday, not counting the stuff for Friday...today is Thursday right?

Good Luck on the testing and am definitly interested in the results. Wish I had a secure open area for a base, to many trees around the house unless you get about 30-40 ft up.

Stumpwater....isn't their some brewmeister there in them woods ?

Both occupations[sic] sound like fun to me !!

Jimbo