Requires accurate parallel validation algorithm for (GPS)

Question

I wrote a delphi program that creates a gpx file as an input for a "poor people management system" for aerosol spraying using an ultralight plane.

By and large, it creates a route (parallel rows) using the gpx file as output.

The route mechanism is based on the Vincenti algorithm, which works fine for any wgs84 calculation, but I can’t get the accuracy of the grid created by ExpertGPS Topografix (requirement).

I assume two-dimensional computation on an ellipsoid:

• 1) From the very beginning rtept (route point), calculate the next rtept taking into account the bearing and arbitrary distance (roll length).

• 2) Calculate the next rtept corresponding to the previous bearing (90 ° rotation) and other arbitrary distance (the distance between the shafts).

• 3) Repeat 1) with the last rtept as a starting point, but in the opposite direction, etc.

What is wrong with him?

Answer 1

You do not describe your Pascal implementation of the Vincenti Earth ellipsoid model, so the following assumption:

• The model uses many geometric trigger functions - ATAN2, COS, SIN, etc. Depending on whether you use the internal functions of Delphi or your own versions, there is a possibility of a lack of accuracy in the calculations. The accuracy of the pi value used in your calculations may affect the required accuracy.
• Floating-point arithmetic can lead to decimal errors. It will matter whether you use single, double or real. I believe that some Delphi internal functions have changed from different versions, so maybe the version of Delphi you use will affect the implementation of the internal function.
• If implemented accurately, the Vincentys formula should be accurate to 0.5 mm. Amazing accuracy. If there is rounding errors or lack of precision in the Delphi implementation, positional errors can be significantly larger.
• Consider the accuracy of your GPS information. Depending on how many satellites are used by the GPS receiver at any given time, the accuracy of the location information changes. Errors on a possible order of 50 feet or more. In addition, updating GPS positional information is not necessarily instantaneous; therefore, if the lane “turns” quickly, you will need to ensure that the GPS updates at the turning point.
• Your template calculation routine seems reasonable, so take a look when implementing the Vincenty algorithm in your Delphi code.
• This list is not exhaustive; I believe that others can dramatically improve it. What I'm talking about is based on my experience with GPS and the various versions of Delphi and what I could remember from my head.
• Something you can try is to compare your distance / bearing calculations using your implementation of the algorithm with the examples presented on the Internet. There are several online calculators. If you have not been there before, the aviation form is a great place to look for examples of other navigation tricks. http://williams.best.vwh.net/avform.htm . Comparison will allow you to gain confidence in the accuracy of Delphi implementation of the Vincenti algorithm with data calculated by mathematics. Simply, your Vincenty implementation may not be accurate. Again, the error may be elsewhere.

Answer 2

I am doing farm GPS guidance similar for ground rig just with Android. Great for second tractor to help follow previous AB tracks especially when they disappear for a bit . 

The function of repeating GPS accuracy from one day to the next will give a greater distance. Expensive use of the dGPS2cm-10cm.5-30metres system is different without dGPS. A simple solution is calibrated in a known place. Cheaper light strips use this method.

Drift As stated above, it concerns movement during operation. Mostly imperceptible <20 cm 3 hours. Can jump 1-2 meters rarely. I think when the satellite will connect or disconnect. Re-calibrate regularly at known coordinates, i.e. spray fill point

GPS accuracy. High speed update phone 1hz. 3? seconds between corrections, say 50 km / h, 41.66 m between corrections. At the ground installation, 18 km an hour, but there will be traces after the first launch. Try checking the update speed of the Bluetooth GPS 10hz, and, as already mentioned, this quickly turns the problem.

The accuracy of the input and whether your management will use dGPS will be of great importance.

As soon as you disconnect from your line, say 5 meters 100 meters to the next point, and then at 50 meters you are still 2.5 meters, unless your guide returns you to the route and not to the following coordinates.

I do not use Vincenty, because I can "lean back" on the line manually and more than 1 km with a difference of <30 cm according to the only link I saw, however I take 2 points and create points of a parabolic point.

Hope these ideas help your situation.