Bearing calculation out of 2 nmea strings

[RFsolution]

Did anyone ever tried to calculate bearing and distance between
set of 2 cordinates ?

1 fixed (lat, lon), 1 comming in from a GPS NMEA 4800,8N1 GMRC string ?

thansk you

[mackrackit]

I have tried LatLon to UTM and gave up, I thought I could do it when LONGs came out, but not quite. Pretty much the same math, at least the same type.

I could not keep track of the decimal point.

Do you know the formula for this? If not, here it is
http://www.codeguru.com/Cpp/Cpp/algo...cle.php/c5115/

[ScaleRobotics]

Did anyone ever tried to calculate bearing and distance between
set of 2 cordinates ?

1 fixed (lat, lon), 1 comming in from a GPS NMEA 4800,8N1 GMRC string ?

thansk you

I used the Cordic to get the heading and distance. To get this to work, you need to calculate the difference between the two waypoints

lathome_hi = the 37 part of the 37.12345 degrees
lathome_low = the 0.1234 (as a whole number 1234) part of 37.1234 degrees

This allows you to have a 3 degree difference in distance, or about 180 miles, although you could add some math division to allow further distances.

GPSMath:
    latlow_dif = lathome_low - latdest_low   'find difference in present loc and destination
    lathi_dif = lathome_hi - latdest_hi
    
    lathi_dif = lathi_dif * 10000        'multiply integer degree difference by 10000 
    lat_dif = lathi_dif + latlow_dif     'so we can add decimal degrees to 16 bit number
    
    lonlow_dif = lonhome_low - londest_low
    lonhi_dif = lonhome_hi - londest_hi
    
    lonhi_dif = lonhi_dif * 10000
    lon_dif = lonhi_dif + lonlow_dif    
          
    if west = 0 then toggle lon_dif.15       'inverse negative bit for east europe
    if north = 0 then toggle lat_dif.15      'inverse negative bit for southern hemisphere
    quadrant = 1   'if lat_dif.15 = 1 and lon_dif.15 = 0 then quadrant =1
    if lat_dif.15 = 0 and lon_dif.15 = 0 then quadrant =2  'see which results are negative
    if lat_dif.15 = 0 and lon_dif.15 = 1 then quadrant =3
    if lat_dif.15 = 1 and lon_dif.15 = 1 then quadrant =4
    if north = 0 then toggle lat_dif.15      'get rid of falsing for S hem so we can do math
        
        
        gosub do_cordic   'perform cos and atan functions to determine bearing and distance to waypoint    

         
   'adds angle of triange to quadrant angle. (Yeah, my quads don't match what you learned in school)
        if quadrant = 1 then b2dest = 900 - b2dest                  
        if quadrant = 2 then b2dest = 900 + b2dest                         
        if quadrant = 3 then b2dest = 2700 - b2dest             
        if quadrant = 4 then b2dest = 2700 + b2dest     
     
'    my quadrants
'            N          
'           |            
'       4   |   1
    'W______|________E    
'           |
'       3   |   2
'           |
'            S

The above code was for the cordic made for a Pic16f device. The PIC18 cordic allows positive and negative numbers, so it figures the quadrant for you.

See the cordic and the last post here:

http://www.picbasic.co.uk/forum/show...ghlight=cordic

[ScaleRobotics]

Here is the part that corrects for the length of longitude, as latitude increases. This was written for the PIC16, and was used in addition to above code.

do_cordic:                              
    Z_ = lathome_hi      'input whole # latitude 37.xxx for longitude  
                         'correction (lon smaller at higher latitude) in Z_ for calcs
    i = Z_ * 128 / 45   '256 / 90   'changes Z_ to the right format 
    if (Z_*128)//45 >= 23 then i = i +1  'round number if remainder is =>0.5 
    Z_ = i                          'put result in Z_
    call sincos                    'get cos(Z_) and sin(Z_)
 
    i = X_ * lon_dif                ' lon_ratio = cos(lat_current)
    lon_dif = div32 10000  'get lon_dif * the fraction for lon correction 

    X_ = lon_dif           'load new lon_dif into X_ , prepare for ATAN function
    Y_ = lat_dif           'load lat_dif into Y_ , prepare for ATAN function

    call atan          'perform ATAN assembly function get angle of coordinates.

    b2dest = Z_ * 225       '*900 / 256 = 225/64    
    b2dest = div32 64       'puts angle in degrees into the variable  in dd.d

[Megahertz]

I have just started a similar project. I was thinking of few ideas. One of them was to trigger something in the system ( alert/alarm or whatever) if the latitude / longitude seconds differ by 2. For example, if the location is 000.000.2 & 00.00.5 >- let us assume that, the next location which comes from GPS are 000.00.4 & 00.00.5 or where 5 becomes a 7.

Will this approach be a good approach to know when the stationary vehicle is moved? or there is any flaw in this which I am missing!

[ScaleRobotics]

Sounds like a fine approach to me. The more complicated stuff is for if you want to know which direction and distance it is to something. If you just want to see if its been moved, and don't care which direction it's been moved to, it seems nice and simple.

[Art]

You should be able to cycle a list of waypoints,
measure the distance to all of them,
find the closest, calculate the bearing, and draw an arrow pointing to it.
Was that the idea?

I finally got PBP working with MPLab again today,
The trig functions should be fine to rotate points around a screen
(2D screen with X - Y coordinates) yes ?

This is how I'd rotate points about an origin in C, where 160x240 is the origin
this was to connect the points for a 3D cube, so has a faked z axis as well:

    // do pre-calculations for rotation
        theta = rotation;		// pre calculations
	thetab = 360 - theta;
	thetab = (3.141592 / 180) * theta;
	ztheta = rotationz;
	zthetab = 360 - ztheta;
	zthetab = (3.141592 / 180) * ztheta;    
    
    
	for(cnt=0; cnt<8; cnt++) {	// 3D rotate points
        xx = xpoints[cnt];
        yy = ypoints[cnt];
        x = xx - 160;
        y = yy - 240;
        xnew = cos(thetab)*x - sin(thetab)*y;
        ynew = sin(thetab)*x + cos(thetab)*y;
        xnew = xnew + 160;
        ynew = ynew + 240;
        if (cnt < 4) {zxx = 160 + 76;} else {zxx = 160 - 76;}
        zyy = ynew;
        zx = zxx - 160;
        zy = zyy - 240;
        zxnew = cos(zthetab)*zx - sin(zthetab)*zy;
        zynew = sin(zthetab)*zx + cos(zthetab)*zy;
        zxnew = zxnew + 160;
        zynew = zynew + 240;
        xcoords[cnt+1] = xnew;
        ycoords[cnt+1] = zynew;
	}

The source and return data is coordinate integers, but it appears the trick you do
is multiplying out the fractions in the formula is to occupy integers and divide later?

[Demon]

Art, I see you're catching up on some reading.

The last post was from August, 2 years ago.

Robert (being a smart-azz)

[Art]

It's cool, cordic trig would still be a good thing to get working.
I tried last time, and had no luck, but it's worth another shot.

Art, I see you're catching up on some reading.

The last post was from August, 2 years ago.

Robert (being a smart-azz)

[Art]

double post :O