Localization training video

Well, I watched it, it's the same thing Trimble does for a calibration.

MightyMoe, post: 455438, member: 700 wrote: Well, I watched it, it's the same thing Trimble does for a calibration.

That's why I shared it here. I'm not familiar with the Trimble interface, but I suspect the tools are there to identify problems with a localization/calibration and to address them. There is a great deal of ignorance regarding these transformations yet many professionals are using them. What's even worse is that they are encouraging subordinates who are even more ignorant to use them. But there is hope, as ignorance is curable. I do think there are many occasions that localizations/calibrations are very useful tools. However, there is a lot going on and users need to be aware.

Same with Leica although I can't include pairs and not hold them to see how they compare when different combinations of other pairs are held. That feature is cool.

Shawn Billings, post: 455441, member: 6521 wrote: That's why I shared it here. I'm not familiar with the Trimble interface, but I suspect the tools are there to identify problems with a localization/calibration and to address them. There is a great deal of ignorance regarding these transformations yet many professionals are using them. What's even worse is that they are encouraging subordinates who are even more ignorant to use them. But there is hope, as ignorance is curable. I do think there are many occasions that localizations/calibrations are very useful tools. However, there is a lot going on and users need to be aware.

Using Trimble you can walk through and back out of using points, look at the residuals.

Basically it's the same process.

Good info (and presentation) Shawn.

Hopefully this will reduce the MISUse of calibrations/localizations, and improve the overall understanding of just what is going on behind the curtain.

Due to my avoidance of all things construction related, I don't have any need for these techniques, and find their use in Land Surveys (retracements), to be problematical at best. But I don't do residential subdivisions either, so my 2 bits isn't worth much.

Loyal

MightyMoe, post: 455450, member: 700 wrote: Using Trimble you can walk through and back out of using points, look at the residuals.

Basically it's the same process.

I'm sure Trimble provides more than just residuals. As I mentioned several times in the video, residuals do not tell the whole story. You have to evaluate the actual parameters and make sure they are realistic.

Shawn Billings, post: 455470, member: 6521 wrote: I'm sure Trimble provides more than just residuals. As I mentioned several times in the video, residuals do not tell the whole story. You have to evaluate the actual parameters and make sure they are realistic.

Yes, you can drill down in the data if you want, what I saw in the video basically confirmed my opinion that it's a thing to do only when there are no other options. I always thought "localization" was some kind of one step thing the way I've heard it described.

Loyal, post: 455464, member: 228 wrote: Good info (and presentation) Shawn.

Hopefully this will reduce the MISUse of calibrations/localizations, and improve the overall understanding of just what is going on behind the curtain.

Loyal

Loyal, I wouldn't count on it 😮

MightyMoe, post: 455475, member: 700 wrote: Yes, you can drill down in the data if you want, what I saw in the video basically confirmed my opinion that it's a thing to do only when there are no other options. I always thought "localization" was some kind of one step thing the way I've heard it described.

In my opinion there is no better method for determining the relationship between two static Cartesian coordinate systems than the Helmert 7 parameter transformation. Far better than holding a single point for translation and a second single point for rotation. With the 7 parameter transformation, each matching pair of points contributes to the overall solution.

Now, having said that, I also believe it is quite dangerous using Helmert 7 parameter transformations (call them localizations or calibrations) without a working knowledge of what is taking place. You shouldn't "drill down in the data if you want", you should "drill down in the data because you must". It's imperative for good, consistent results to evaluate the parameters. This is a main focus of the video (or at least that was the intent).

Shawn Billings, post: 455492, member: 6521 wrote: In my opinion there is no better method for determining the relationship between two static Cartesian coordinate systems than the Helmert 7 parameter transformation. Far better than holding a single point for translation and a second single point for rotation. With the 7 parameter transformation, each matching pair of points contributes to the overall solution.

Now, having said that, I also believe it is quite dangerous using Helmert 7 parameter transformations (call them localizations or calibrations) without a working knowledge of what is taking place. You shouldn't "drill down in the data if you want", you should "drill down in the data because you must". It's imperative for good, consistent results to evaluate the parameters. This is a main focus of the video (or at least that was the intent).

There were a number of things in the video that I found unsettling. He glossed over changing the numbers for the base point, why would he do that?

And because he didn't have very good elevation control he put a setting at 0 which he said would allow him to extend outside of his calibration with good elevations. Really?

I didn't understand that at all, is he saying he's attaching a geoid model to an elevation point, or is he saying he's making the rise over run 0?

Trimble used to have in their calibrations a ft/ft number (probably still do) to tilt their plane, of course a 0 would mean 0ft/1ft.
0 would not work well at all for that number, at least not where I am.

I went through the calibration wars back in the day, nothing is different now that I can see.
Using a Helmert 7 parameter transformation doesn't help when the base data is inferior.
Nor does LS make inferior superior.

The few times I've needed to do one of these I looked at all of the parameters of the transformation, but primarily the scale factor and the direction and percent grade of the vertical plane; I think those tell you more about the overall fit.

Shawn, I thought the video was well put together. I also think it will help future surveyors understand what is happening in the magic box (regardless of brand) so they can make informed decisions.

MightyMoe, post: 455515, member: 700 wrote: There were a number of things in the video that I found unsettling. He glossed over changing the numbers for the base point, why would he do that?

And because he didn't have very good elevation control he put a setting at 0 which he said would allow him to extend outside of his calibration with good elevations. Really?

I didn't understand that at all, is he saying he's attaching a geoid model to an elevation point, or is he saying he's making the rise over run 0?

Trimble used to have in their calibrations a ft/ft number (probably still do) to tilt their plane, of course a 0 would mean 0ft/1ft.
0 would not work well at all for that number, at least not where I am.

I went through the calibration wars back in the day, nothing is different now that I can see.
Using a Helmert 7 parameter transformation doesn't help when the base data is inferior.
Nor does LS make inferior superior.

I also missed the logic in that. The first thing the Trimble Sitech guys will tell you is to make sure your control extends past the limits of the site, make sure all of the points are good (relative to each other), and don't work outside of that.

Same process with Topcon Magnet Tools. Not sure why this is so eye oping to some.

Maybe old apps like TDs Pro Carlson can't do this?

MightyMoe, post: 455515, member: 700 wrote: And because he didn't have very good elevation control he put a setting at 0 which he said would allow him to extend outside of his calibration with good elevations.

I didn't re-watch the video, but as I recall Shawn elected to hold the geoid model instead of any of the surveyed point elevations, which would provide consistent vertical beyond the localization boundary.

MightyMoe, post: 455515, member: 700 wrote: There were a number of things in the video that I found unsettling. He glossed over changing the numbers for the base point, why would he do that?

The base was started from an autonomous position. I processed the base to CORS using DPOS. This caused the base position to move in my local system. I then rectified it by opening localization again and the parameters were recalculated using the new geodetic positions of the points shot from the base.

MightyMoe, post: 455515, member: 700 wrote: And because he didn't have very good elevation control he put a setting at 0 which he said would allow him to extend outside of his calibration with good elevations. Really? I didn't understand that at all, is he saying he's attaching a geoid model to an elevation point, or is he saying he's making the rise over run 0?

No not really. You should re-watch this part. I zeroed the inclined plane. This effectively translated vertically using the average difference of all points held in vertical. Since there was no additional tilt, the vertical datum is based on the Geoid used for the surveyed points (in this case Geoid12B). The units for the tilt in this application are degrees, minutes, seconds. 0 is zero seconds. Inclined planes are the reason you are typically encouraged to stay "in the box". If you aren't tilting the plane though, you're on the Geoid and you are free to work beyond the limits of the control.

MightyMoe, post: 455515, member: 700 wrote: I went through the calibration wars back in the day, nothing is different now that I can see.
Using a Helmert 7 parameter transformation doesn't help when the base data is inferior.
Nor does LS make inferior superior.

Nothing helps when the base data is bad. You'll notice, the video says one of the keys to success is reliable coordinates.

eddycreek, post: 455792, member: 501 wrote: I also missed the logic in that. The first thing the Trimble Sitech guys will tell you is to make sure your control extends past the limits of the site, make sure all of the points are good (relative to each other), and don't work outside of that.

If the plane is not tilted, you do not have to stay in the box for elevations to remain valid. You'll note that in the video a point is intentionally thrown in that is wrong vertically. The residuals look good, but the tilt is unreasonable. Staying in the box would not have been good. Zeroing the tilt, in most cases, particularly with a good geoid model is the better approach.

leegreen, post: 455797, member: 2332 wrote: Same process with Topcon Magnet Tools. Not sure why this is so eye oping to some.

Maybe old apps like TDs Pro Carlson can't do this?

I didn't share it to discuss the virtues of the software, I shared it to discuss the methodology of localizations/calibrations in general. As can be seen in several recent threads and from personal experience, there is still a lot of ignorance regarding localizations/calibrations in the profession.

Lee D, post: 455517, member: 7971 wrote: The few times I've needed to do one of these I looked at all of the parameters of the transformation, but primarily the scale factor and the direction and percent grade of the vertical plane; I think those tell you more about the overall fit.

Exactly Lee. The video points out cases in which the residuals did not provide enough information to isolate a blunder, particularly regarding scale and inclination. Users should do more than review residuals when using a localization/calibration and review the parameters as well. They will tell the story more completely.