Pytrack GPS API
-
Sounds really interesting, though we'd have to re-write some of the code to run on a PyTrack & LoPy setup :D
challange-accepted
ALSO, has anyone else found some documentation of how other GPS vendors do the "Accuracy" calculation?
-
@combaindeft I ran the library on a pytrack / lopy combination, but gave up, to many out of memory errors.
-
Hello all,
I received today one Pytrack board for testing, and found it very easy to get more data from NMEA sentences. Just need to parse the correct NMEA sentence for what you need. Original library parse $GNGLL, however for what i need (lat/long, gps time, altitude, fix validation) i just changed the lib to parse $GPGGA and get that values. If anyone need i can show my basic test code.
Some time ago i tried the micropyGPS lib with an external ublox gps module, but also had that memory errors. I think they could be solved, however since i didn't need gps by that time, i never tried to fix anything.
Best regards
-
@neuromystix said in Pytrack GPS API:
If anyone need i can show my basic test code.
Yes, please. I had sadly no time to work on my implementation. You run your code on a 4MB version, if you never had any memory problems, right? I can share my code as well, if it is ready. I used preallocated buffers to get around the memory problem, but my parsing of numbers is quite awefull.
For me it is hard to get one of the 4MB *Pys. The resellers in Germany seems to have only the old versions or the newer ones are sold out really fast.
-
@crumble i followed the original lib from Pycom, just parse a different NMEA message. I've reading Quectel L76 GNSS protocol specification and i find they have only the following message types:
- xxRMC, where we can get UTC Time, validation, lat, long, speed, COG, date and other non important things;
- xxVTG, mostly for track course and speed data;
- xxGGA, where GPS fix data are. UTC time, lat, long, fix status, number of satellites, HDOP, altitude, height of geoid, DGPS info(if any) are all in this sentence;
- xxGSA, infos about satellites;
- xxGSV, more info about satellites;
- xxGLL, a basic sentence with geographic info like lat, long, UTC time, data validation and position mode.
So, depending on what things you need, you just have to parse the exact sentence. In my case, i changed the original library which parsed $GNGLL (few information) to a new function which parse $GPGGA (have more data!). The ideia was the same, reading sentences from i2c till a get a complete GNGLL sentence (searching for beggining and ending of that sentence >> nmea.find(b'GPGGA') and e_idx = gpgga.find(b'\r\n') >> It's a NMEA standard so it always start and end this way), then parse all the values i need. Today it has been working for nearly two hours, reading each second, and i don't get any errors.
For use it just call l76.coordinates1() which will return (lat, long, time(not formatted), altitude (M), valid fix (true ou false), number of satellites).
L76GNSS.py -
@crumble I added a function where i can get time and date. If you want, try it and check if you have any problem.
Code:
## Test code for read date/time from gps and update RTC import machine import math import network import os import time import utime from machine import RTC from machine import SD from machine import Timer from L76GNSS import L76GNSS from pytrack import Pytrack import struct # setup as a station import gc time.sleep(2) gc.enable() #Start GPS py = Pytrack() l76 = L76GNSS(py, timeout=600) #start rtc rtc = machine.RTC() print('Aquiring GPS signal....') #try to get gps date to config rtc while (True): gps_datetime = l76.get_datetime() #case valid readings if gps_datetime[3]: day = int(gps_datetime[4][0] + gps_datetime[4][1] ) month = int(gps_datetime[4][2] + gps_datetime[4][3] ) year = int('20' + gps_datetime[4][4] + gps_datetime[4][5] ) hour = int(gps_datetime[2][0] + gps_datetime[2][1] ) minute = int(gps_datetime[2][2] + gps_datetime[2][3] ) second = int(gps_datetime[2][4] + gps_datetime[2][5] ) print("Current location: {} {} ; Date: {}/{}/{} ; Time: {}:{}:{}".format(gps_datetime[0],gps_datetime[1], day, month, year, hour, minute, second)) rtc.init( (year, month, day, hour, minute, second, 0, 0)) break print('RTC Set from GPS to UTC:', rtc.now()) chrono = Timer.Chrono() chrono.start() #sd = SD() #os.mount(sd, '/sd') #f = open('/sd/gps-record.txt', 'w') while (True): print("RTC time : {}".format(rtc.now())) coord = l76.coordinates() #f.write("{} - {}\n".format(coord, rtc.now())) print("$GPGLL>> {} - Free Mem: {}".format(coord, gc.mem_free())) coord1 = l76.coordinates1() print("$GPGGA>> {} - Free Mem: {}".format(coord1, gc.mem_free())) coord2 = l76.get_datetime() print("$G_RMC>> {} - Free Mem: {}".format(coord2, gc.mem_free()))The L76GNSS.py code:
from machine import Timer import time import gc import binascii class L76GNSS: GPS_I2CADDR = const(0x10) def __init__(self, pytrack=None, sda='P22', scl='P21', timeout=None): if pytrack is not None: self.i2c = pytrack.i2c else: from machine import I2C self.i2c = I2C(0, mode=I2C.MASTER, pins=(sda, scl)) self.chrono = Timer.Chrono() self.timeout = timeout self.timeout_status = True self.reg = bytearray(1) self.i2c.writeto(GPS_I2CADDR, self.reg) def _read(self): self.reg = self.i2c.readfrom(GPS_I2CADDR, 128) #Changed from 64 to 128 - I2C L76 says it can read till 255 bytes return self.reg def _convert_coords(self, gngll_s): lat = gngll_s[1] lat_d = (float(lat) // 100) + ((float(lat) % 100) / 60) lon = gngll_s[3] lon_d = (float(lon) // 100) + ((float(lon) % 100) / 60) if gngll_s[2] == 'S': lat_d *= -1 if gngll_s[4] == 'W': lon_d *= -1 return(lat_d, lon_d) #diff indexes from original - Using GGA sentence def _convert_coords1(self, gngga_s): lat = gngga_s[2] lat_d = (float(lat) // 100) + ((float(lat) % 100) / 60) lon = gngga_s[4] lon_d = (float(lon) // 100) + ((float(lon) % 100) / 60) if gngga_s[3] == 'S': lat_d *= -1 if gngga_s[5] == 'W': lon_d *= -1 return(lat_d, lon_d) def _get_time(self, gngga_s): gps_time = gngga_s[1] return(gps_time) def _get_altitude(self, gngga_s): gps_altitude = gngga_s[9] return(gps_altitude) def _get_satellites(self, gngga_s): num_satellites = gngga_s[7] return(num_satellites) def _fix_quality(self, gngga_s): valid = gngga_s[6] if valid == '1': return True else: return False #Using RMC sentence def _get_time_rmc(self, gnrmc_s): gps_time = gnrmc_s[1] return(gps_time) def _data_valid_rmc(self, gnrmc_s): valid = gnrmc_s[2] if valid == 'A': return True else: return False def _get_date_rmc(self, gnrmc_s): gps_date = gnrmc_s[9] return(gps_date) def coordinates(self, debug=False): lat_d, lon_d, debug_timeout = None, None, False if self.timeout is not None: self.chrono.reset() self.chrono.start() nmea = b'' while True: if self.timeout is not None and self.chrono.read() >= self.timeout: self.chrono.stop() chrono_timeout = self.chrono.read() self.chrono.reset() self.timeout_status = False debug_timeout = True if not self.timeout_status: gc.collect() break nmea += self._read().lstrip(b'\n\n').rstrip(b'\n\n') gngll_idx = nmea.find(b'GNGLL') if gngll_idx >= 0: gngll = nmea[gngll_idx:] e_idx = gngll.find(b'\r\n') if e_idx >= 0: try: gngll = gngll[:e_idx].decode('ascii') gngll_s = gngll.split(',') lat_d, lon_d = self._convert_coords(gngll_s) except Exception: pass finally: nmea = nmea[(gngll_idx + e_idx):] gc.collect() break else: gc.collect() if len(nmea) > 410: # i suppose it can be safely changed to 82, which is longest NMEA frame nmea = nmea[-5:] # $GNGL without last L time.sleep(0.1) self.timeout_status = True if debug and debug_timeout: print('GPS timed out after %f seconds' % (chrono_timeout)) return(None, None) else: return(lat_d, lon_d) #TEST functions #Parser for GPGGA def coordinates1(self, debug=False): lat_d, lon_d, gps_time, valid, gps_altitude, num_satellites, debug_timeout = None, None, None, None, None, False, False if self.timeout is not None: self.chrono.reset() self.chrono.start() nmea = b'' while True: if self.timeout is not None and self.chrono.read() >= self.timeout: self.chrono.stop() chrono_timeout = self.chrono.read() self.chrono.reset() self.timeout_status = False debug_timeout = True if not self.timeout_status: gc.collect() break nmea += self._read().lstrip(b'\n\n').rstrip(b'\n\n') gpgga_idx = nmea.find(b'GPGGA') if gpgga_idx >= 0: gpgga = nmea[gpgga_idx:] gpgga_e_idx = gpgga.find(b'\r\n') if gpgga_e_idx >= 0: try: gpgga = gpgga[:gpgga_e_idx].decode('ascii') gpgga_s = gpgga.split(',') lat_d, lon_d = self._convert_coords1(gpgga_s) gps_time = self._get_time(gpgga_s) valid = self._fix_quality(gpgga_s) gps_altitude = self._get_altitude(gpgga_s) num_satellites = self._get_satellites(gpgga_s) except Exception: pass finally: nmea = nmea[(gpgga_idx + gpgga_e_idx):] gc.collect() break else: gc.collect() if len(nmea) > 410: # i suppose it can be safely changed to 82, which is longest NMEA frame nmea = nmea[-5:] # $GNGL without last L time.sleep(0.1) self.timeout_status = True if debug and debug_timeout: print('GPS timed out after %f seconds' % (chrono_timeout)) return(None, None, None, None, False, None) else: return(lat_d, lon_d, gps_time, gps_altitude, valid, num_satellites) #parser for UTC time and date >> Reads GPRMC def get_datetime(self, debug=False): lat_d, lon_d, gps_time, valid, gps_date, rmc_idx, debug_timeout = None, None, None, None, None, -1, False if self.timeout is not None: self.chrono.reset() self.chrono.start() nmea = b'' while True: if self.timeout is not None and self.chrono.read() >= self.timeout: self.chrono.stop() chrono_timeout = self.chrono.read() self.chrono.reset() self.timeout_status = False debug_timeout = True if not self.timeout_status: gc.collect() break nmea += self._read().lstrip(b'\n\n').rstrip(b'\n\n') #Since or spg or glonass could give date see which one is present -SEE page 10 GNSS protocol #GPS only - GPRMC GPGGA #Glonass only - GNRMC GPGGA #GPS+GLON - GNRMC GPGGA #No station - GPRMC GPGGA gprmc_idx = nmea.find(b'GPRMC') gnrmc_idx = nmea.find(b'GNRMC') if gprmc_idx >= 0: rmc_idx = gprmc_idx if gnrmc_idx >= 0: rmc_idx = gnrmc_idx if rmc_idx >= 0: rmc = nmea[rmc_idx:] rmc_e_idx = rmc.find(b'\r\n') if rmc_e_idx >= 0: try: rmc = rmc[:rmc_e_idx].decode('ascii') rmc_s = rmc.split(',') lat_d, lon_d = self._convert_coords1(rmc_s[1:]) gps_time = self._get_time_rmc(rmc_s) valid = self._data_valid_rmc(rmc_s) gps_date = self._get_date_rmc(rmc_s) except Exception: pass finally: nmea = nmea[(rmc_idx + rmc_e_idx):] gc.collect() break else: gc.collect() if len(nmea) > 512: # i suppose it can be safely changed to 82, which is longest NMEA frame --CHANGED to 512 nmea = nmea[-5:] # $GNGL without last L time.sleep(0.1) self.timeout_status = True if debug and debug_timeout: print('GPS timed out after %f seconds' % (chrono_timeout)) return(None, None, None, False, None) else: return(lat_d, lon_d, gps_time, valid, gps_date) -
@andrethemac Yeah, ran into that issue aswell.
why we are re-doing the gps library that comes from pycom :)
and selectively adding more and more features ... so we won't run out of memory
-
@neuromystix I used to search the nmea messages with _read looking for xxGGA until a fix was found. Only then I use the l76.coordinates to find the exact position.
a bit of a work around.
The new library you posted works as a charm, no more memory errors.
The Quectel is a bit "slow" but with a timeout of 180 (3 minutes), I manage to get a fix (almost) every time the wipy/lopy comes awake (after deep sleep). Not so good for the battery, 3 minutes on +- 70 mA vs 390 uA in deep sleep with the pytrack. The 390 uA are probably from the accelerator, I use the accelerator to wake up the unit when it gets more than 2g for over a second. -
@andrethemac i followed the original lib which use gc to clean memory. But haven't tried any other options.
Same problem here with the starting time. Sometimes i just unplug the cable from my computer and plug it to a powerbank and have to wait 5 minutes for it to get fix. For my project this will not be a problem but for a battery operated device it's not a good thing... -
@neuromystix Works like a champ - it would be good if this was added to the main lib. I added HDOP (GPGGA) as that is a good indication of accuracy.
-
@neuromystix your work has been awesome. Thank you so much for your efforts. In running some test logs earlier today, I found that I got a series of fixes logged that were "invalid". Upon reviewing you code the protocol documents, I have one update to propose
def _fix_quality(self, gngga_s): valid = gngga_s[6] if valid == '0': return False else: return TrueYour initial fix quality function only returned true if gngga_s[6] was 1, however it seems that values such as 2 or 3 may be valid as well.
Per v2.2 of the protocol spec:
Value Description
0 Fix not available or invalid
1 GPS SPS Mode, fix valid
2 Differential GPS, SPS Mode, fix valid
3 GPS PPS Mode, fix validOnce again, thanks for your efforts. Just wanted to share my findings.
-
@mwtidd When i wrote it i was reading the "Quectel l76 Series GNSS protocol Specification V3.3" which says (@page 12):
Fix Status
"0"=Invalid
"1"=GNSS fix
"2"=DGPS fix
"6"=Estimated (dead reckoning) modeThat's why i only used "1". But even with that, we can use the value "2" for the DGPS fix. I havenĀ“t checked if i had some fixes "2" ou another values. If it worked for you, nice ;)
Best regards
Nelson Pinto -
@bmarkus said in Pytrack GPS API:
umber of satellites visible /used for fi
- number of satellites visible /used for fix - Is this possible yet?!
-
@stefan85e If you need the number of satellites used to fix, it's already parsed in l76.coordinates1(). The "Satellites in view" i haven't done it, but it's easy to implement, just parse any $__GSV sentence, the third field is "Satellites in view".
-
-
Hi all. I've taken the original librarya and @Neuromystix addtions and made my own library. It parses aml the sentences from the L76 gps and returns them as a dict. you can get the coordinates parsed, the time, ...
https://github.com/andrethemac/L76GLNSV4I've noticed one (strange?) thing. the RMC (required minimum data to get a fix) message only pass once (just after starting the gps) and does not reappear. Any idea's ?
best regards -
@neuromystix said in Pytrack GPS API:
For use it just call l76.coordinates1() which will return (lat, long, time(not formatted), altitude (M), valid fix (true ou false), number of satellites).
L76GNSS.py
So what did you do to get the nimber of satellites? We are trying this and we cant get it to work. -
@stefan85e Hello,
You have to do something like this inside your code, and you'll get the number of satellites used for fix:
... gps_data = l76.coordinates1() n_sat = gps_data[5] ...Or you can use the @andrethemac library which will get you all the data ;)
-
@andrethemac
(2018, 4, 17, 11, 29, 12, 564887, None)(None(53.33632, 12.53262)
(None, None) Got when I add this:gps_data = l76.coordinates1()
n_sat = gps_data[5]Did also try to use this lib:
https://github.com/andrethemac/L76GLNSV4 -
@neuromystix said in Pytrack GPS API:
num_satellites
Does this work:
return(lat_d, lon_d, gps_time, gps_altitude, valid, num_satellites)
I want num_satellites...
Hello! It looks like you're interested in this conversation, but you don't have an account yet.
Getting fed up of having to scroll through the same posts each visit? When you register for an account, you'll always come back to exactly where you were before, and choose to be notified of new replies (either via email, or push notification). You'll also be able to save bookmarks and upvote posts to show your appreciation to other community members.
With your input, this post could be even better š
Register Login