Ahoj vespolek, trochu jsem hledal, jeslti by se nejakymi rozumne dostupnymi prostredky nedala zlepsit presnost tracklogu, zejmena ziskanych na kole, ale klidne i pesky a autem. Zlobi mne to hlavne pri rychlejsi jizde na kole spojene s rycheljsimi zmenami smeru (kdy Garmin 60CSx "nestiha zatacet") a v lese, kde je presnost vubec na pytel (60CSx ani Holux M-1000C). Zacal jsem uvazovat o nejakem modulu, ktery by sbiral raw GPS data na dostatecne frekvenci, na nichz by se dal udelat postprocesing - ale ne v cenove kategorii 30.000CZK a vic (tedy profi GPS zarizeni). Narazil jsem dve firmicky, ktere neco takoveho delaji (a do jiste miry spolu i spolupracuji): Optimalsystem.de a OneTalent-GNSS. Navrhy desek dela Michele Bavaro pro obe firmy a primarne pracuje pro ten OneTalent-GNSS - nize najdete jak moje dotazy na neho, tak i jeho odpovedi. Prestoze vyjadruje skepsi kolem dosazeni presnosti 1m v lese (pricem 1m povazuji pro potreby OSM naprosto dostacujici), nicmene porad by to mohlo by podstatne zlepseni proti tomu, co mam ted. Uvazuji o jednom z nasledujicich modulu: http://www.onetalent-gnss.com/ideas/wireless/denga10logbt http://www.onetalent-gnss.com/ideas/wireless/rappen10logbt Jinak data pro postprocesing by napr. pro okoli Brna mely byt zdarma dostupne ze site stanic EUREF: http://www.epncb.oma.be/_networkdata/stationlist.php pro Brno napr. tady: http://www.epncb.oma.be/_networkdata/siteinfo4onestation.php?station=TUBO_11503M001 Moje dotazy do fora: - Hrali jste si nekdo s necim podobnym? - Meli byste chut a cas se nekdo do toho taky pustit? Diky, Petr From: Michele Bavaro [mailto:mbavaro na onetalent-gnss.com] Sent: Sunday, October 07, 2012 10:19 PM To: Petr Holub Subject: Re: a module for accurate mapping for OSM Hello Petr, Thanks for getting in touch! It will be very difficult to guarantee 1m accuracy (standalone real-time or differential post-processed) under dense foliage. Trees are well known to disrupt carrier phase information which is essential to achieve such accuracies. There is something about it on the German forum Kowoma: http://www.kowoma.de/gpsforum/viewtopic.php?f=1&t=3248&p=14992#p14992 Denga10 and Rappen10 are not suitable. But their _LogBt or _LogWi version instead might work: they have a battery which will last a longer than 6hr, logging, and Bluetooth - or pretty much any other wireless communication system. Of course both have raw measurements but their carrier phase will be impaired under foliage as I wrote above. I am finalising the firmware in these days (today I had logging with fatfs and USB MSC running simultaneously finally) but -as it is quite experimental- I cannot really claim that is a "set and forget" kind of solution. Firmware is open though and if you have programming skills you may modify it to your needs. I am also working on enclosures... I have the bare boxes but I need to hand-tool them. If you are happy to do it by yourself I can make a discount: http://www.onetalent-gnss.com/announcements/contributorsgetafreereceiver Unfortunately in any case I cannot make you a lower price than 250EUR+VAT+shipping as I hand assemble the boards and I don't sell many of them so I cannot reduce the price by producing volumes. I would love to help you but I understand that compared to a 100EUR logger it's a lot of money to invest. By the way, all new Mediatek receivers have 10Hz and very high sensitivity... the new MTK3333 also has Glonass. ST Microelectronics has the STA8088FG which performs very well and it's very low cost. Not to mention of course the usual CSR GSD4E and Skytraq Venus.. Have you looked at Fastrax, Locosys, and GlobalTop? I am very happy to help an enthusiastic mapper, please let me know if I can do something else for you. Best regards, Michele On 7 October 2012 18:52, Petr Holub <hopet na ics.muni.cz> wrote: Hello, I'm looking for a GNSS module that could improve precision of my mapping efforts for Openstreetmap - mainly mountainbike trails and hiking paths in the Czech Republic (http://www.mtbmap.cz/). Up to now, I'm using Sirf3-based Garmin 60CSx mounted on the MTB handlebar, and MTK3329-based Holux M-1000C logger switched to 10Hz, mounted to my MTB helmet. While Garmin 60CSx produces reasonably precise results while hiking open-air, it has significantly worse behavior in hilly forests and 1Hz operation is not sufficient for accurately tracking shape of paths when riding fast on my bike. Holux provides improves tracking of fast motion, but it has usually significantly worse accuracy (i.e., wanders more) despite being mounted on top of my helmet and thus having good sky view. Thus I'm looking for a reasonably priced (<400 EUR, ideally 200-300EUR) solution for more accurate logging of MTB trails, ideally at 4Hz or better and approx. 1m accuracy - obviously I don't need 1cm accuracy, but I'd like to have 1m accuracy even under forest foliage. I don't need precise real-time output (Garmin will work fine for real-time navigation; having, e.g., additional Bluetooth NMEA output to verify operation, would be nice for the module, but it is not conditio sine qua non). I assume to put the device on my helmet again, replacing the Holux logger - so I shouldn't be excessively large or heavy. An alternative solution would be to have the device in my backpack and only put the antenna onto the helmet. The device should keep working on a battery for at least 6hrs. The ideal workflow for me is to keep the device logging into its internal memory, come back home, download the raw log and post-process it using RTKLIB to achieve 1m log accuracy, without need for paying recurring costs (such as CZEPOS professional DGPS correction). I'm doing this effort as a hobbyist, thus minimizing operational costs is important. Is there some solution for this in your portfolio? I was looking into Denga10LogBt, but I'm not sure if it meets my requirements - to sum it up: - battery operated for at least 6 hours - internal RAW logging for later processing - 1m accuracy under foliage (50cm accuracy would be absolutely perfect) - 4Hz or higher frequency (about 10Hz is ideal) - either whole device or at least the antenna should be helmet-mountable - helmet-mountable part of the device (or complete device) needs to be weather-sealed and needs to have reasonable weight (I can try to build my own housing, but not sure...) - post-processing should avoid need for paid data (such as CZEPOS) - ideally use open-source software to process the data Maybe my wishes still can't be fulfilled, but I've just stumbled over your website and the devices really look promising... Thanks, Petr -- Michele Bavaro Consultant at OneTalent GNSS Mob. IT: +39 347 009 4377 Skype: onetalent-gnss http://www.onetalent-gnss.com _______________________________________________ Talk-cz mailing list Talk-cz na openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-cz
Ahoj vespolek, trochu jsem hledal, jeslti by se nejakymi rozumne dostupnymi prostredky nedala zlepsit presnost tracklogu, zejmena ziskanych na kole, ale klidne i pesky a autem. Zlobi mne to hlavne pri rychlejsi jizde na kole spojene s rycheljsimi zmenami smeru (kdy Garmin 60CSx "nestiha zatacet") a v lese, kde je presnost vubec na pytel (60CSx ani Holux M-1000C). Zacal jsem uvazovat o nejakem modulu, ktery by sbiral raw GPS data na dostatecne frekvenci, na nichz by se dal udelat postprocesing - ale ne v cenove kategorii 30.000CZK a vic (tedy profi GPS zarizeni). Narazil jsem dve firmicky, ktere neco takoveho delaji (a do jiste miry spolu i spolupracuji): Optimalsystem.de a OneTalent-GNSS. Navrhy desek dela Michele Bavaro pro obe firmy a primarne pracuje pro ten OneTalent-GNSS - nize najdete jak moje dotazy na neho, tak i jeho odpovedi. Prestoze vyjadruje skepsi kolem dosazeni presnosti 1m v lese (pricem 1m povazuji pro potreby OSM naprosto dostacujici), nicmene porad by to mohlo by podstatne zlepseni proti tomu, co mam ted. Uvazuji o jednom z nasledujicich modulu: http://www.onetalent-gnss.com/ideas/wireless/denga10logbt http://www.onetalent-gnss.com/ideas/wireless/rappen10logbt Jinak data pro postprocesing by napr. pro okoli Brna mely byt zdarma dostupne ze site stanic EUREF: http://www.epncb.oma.be/_networkdata/stationlist.php pro Brno napr. tady: http://www.epncb.oma.be/_networkdata/siteinfo4onestation.php?station=TUBO_11503M001 Moje dotazy do fora: - Hrali jste si nekdo s necim podobnym? - Meli byste chut a cas se nekdo do toho taky pustit? Diky, Petr From: Michele Bavaro [mailto:mbavaro na onetalent-gnss.com] Sent: Sunday, October 07, 2012 10:19 PM To: Petr Holub Subject: Re: a module for accurate mapping for OSM Hello Petr, Thanks for getting in touch! It will be very difficult to guarantee 1m accuracy (standalone real-time or differential post-processed) under dense foliage. Trees are well known to disrupt carrier phase information which is essential to achieve such accuracies. There is something about it on the German forum Kowoma: http://www.kowoma.de/gpsforum/viewtopic.php?f=1&t=3248&p=14992#p14992 Denga10 and Rappen10 are not suitable. But their _LogBt or _LogWi version instead might work: they have a battery which will last a longer than 6hr, logging, and Bluetooth - or pretty much any other wireless communication system. Of course both have raw measurements but their carrier phase will be impaired under foliage as I wrote above. I am finalising the firmware in these days (today I had logging with fatfs and USB MSC running simultaneously finally) but -as it is quite experimental- I cannot really claim that is a "set and forget" kind of solution. Firmware is open though and if you have programming skills you may modify it to your needs. I am also working on enclosures... I have the bare boxes but I need to hand-tool them. If you are happy to do it by yourself I can make a discount: http://www.onetalent-gnss.com/announcements/contributorsgetafreereceiver Unfortunately in any case I cannot make you a lower price than 250EUR+VAT+shipping as I hand assemble the boards and I don't sell many of them so I cannot reduce the price by producing volumes. I would love to help you but I understand that compared to a 100EUR logger it's a lot of money to invest. By the way, all new Mediatek receivers have 10Hz and very high sensitivity... the new MTK3333 also has Glonass. ST Microelectronics has the STA8088FG which performs very well and it's very low cost. Not to mention of course the usual CSR GSD4E and Skytraq Venus.. Have you looked at Fastrax, Locosys, and GlobalTop? I am very happy to help an enthusiastic mapper, please let me know if I can do something else for you. Best regards, Michele On 7 October 2012 18:52, Petr Holub <hopet na ics.muni.cz> wrote: Hello, I'm looking for a GNSS module that could improve precision of my mapping efforts for Openstreetmap - mainly mountainbike trails and hiking paths in the Czech Republic (http://www.mtbmap.cz/). Up to now, I'm using Sirf3-based Garmin 60CSx mounted on the MTB handlebar, and MTK3329-based Holux M-1000C logger switched to 10Hz, mounted to my MTB helmet. While Garmin 60CSx produces reasonably precise results while hiking open-air, it has significantly worse behavior in hilly forests and 1Hz operation is not sufficient for accurately tracking shape of paths when riding fast on my bike. Holux provides improves tracking of fast motion, but it has usually significantly worse accuracy (i.e., wanders more) despite being mounted on top of my helmet and thus having good sky view. Thus I'm looking for a reasonably priced (<400 EUR, ideally 200-300EUR) solution for more accurate logging of MTB trails, ideally at 4Hz or better and approx. 1m accuracy - obviously I don't need 1cm accuracy, but I'd like to have 1m accuracy even under forest foliage. I don't need precise real-time output (Garmin will work fine for real-time navigation; having, e.g., additional Bluetooth NMEA output to verify operation, would be nice for the module, but it is not conditio sine qua non). I assume to put the device on my helmet again, replacing the Holux logger - so I shouldn't be excessively large or heavy. An alternative solution would be to have the device in my backpack and only put the antenna onto the helmet. The device should keep working on a battery for at least 6hrs. The ideal workflow for me is to keep the device logging into its internal memory, come back home, download the raw log and post-process it using RTKLIB to achieve 1m log accuracy, without need for paying recurring costs (such as CZEPOS professional DGPS correction). I'm doing this effort as a hobbyist, thus minimizing operational costs is important. Is there some solution for this in your portfolio? I was looking into Denga10LogBt, but I'm not sure if it meets my requirements - to sum it up: - battery operated for at least 6 hours - internal RAW logging for later processing - 1m accuracy under foliage (50cm accuracy would be absolutely perfect) - 4Hz or higher frequency (about 10Hz is ideal) - either whole device or at least the antenna should be helmet-mountable - helmet-mountable part of the device (or complete device) needs to be weather-sealed and needs to have reasonable weight (I can try to build my own housing, but not sure...) - post-processing should avoid need for paid data (such as CZEPOS) - ideally use open-source software to process the data Maybe my wishes still can't be fulfilled, but I've just stumbled over your website and the devices really look promising... Thanks, Petr -- Michele Bavaro Consultant at OneTalent GNSS Mob. IT: +39 347 009 4377 Skype: onetalent-gnss http://www.onetalent-gnss.com _______________________________________________ Talk-cz mailing list Talk-cz na openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-cz
Nicméně pro ostré zatáčky v lese by pomohl jedině GPS v kombinaci s gyroskopem/akcelerometrem.
Nicméně pro ostré zatáčky v lese by pomohl jedině GPS v kombinaci s gyroskopem/akcelerometrem.
Nicméně pro ostré zatáčky v lese by pomohl jedině GPS v kombinacis gyroskopem/akcelerometrem.
nepouziva, vzdyt 3D
Android....
mailing list
Ahoj vespolek, trochu jsem hledal, jeslti by se nejakymi rozumne dostupnymi prostredky nedala zlepsit presnost tracklogu, zejmena ziskanych na kole, ale klidne i pesky a autem. Zlobi mne to hlavne pri rychlejsi jizde na kole spojene s rycheljsimi zmenami smeru (kdy Garmin 60CSx "nestiha zatacet") a v lese, kde je presnost vubec na pytel (60CSx ani Holux M-1000C). Zacal jsem uvazovat o nejakem modulu, ktery by sbiral raw GPS data na dostatecne frekvenci, na nichz by se dal udelat postprocesing - ale ne v cenove kategorii 30.000CZK a vic (tedy profi GPS zarizeni). Narazil jsem dve firmicky, ktere neco takoveho delaji (a do jiste miry spolu i spolupracuji): Optimalsystem.de a OneTalent-GNSS. Navrhy desek dela Michele Bavaro pro obe firmy a primarne pracuje pro ten OneTalent-GNSS - nize najdete jak moje dotazy na neho, tak i jeho odpovedi. Prestoze vyjadruje skepsi kolem dosazeni presnosti 1m v lese (pricem 1m povazuji pro potreby OSM naprosto dostacujici), nicmene porad by to mohlo by podstatne zlepseni proti tomu, co mam ted. Uvazuji o jednom z nasledujicich modulu: http://www.onetalent-gnss.com/ideas/wireless/denga10logbt http://www.onetalent-gnss.com/ideas/wireless/rappen10logbt Jinak data pro postprocesing by napr. pro okoli Brna mely byt zdarma dostupne ze site stanic EUREF: http://www.epncb.oma.be/_networkdata/stationlist.php pro Brno napr. tady: http://www.epncb.oma.be/_networkdata/siteinfo4onestation.php?station=TUBO_11503M001 Moje dotazy do fora: - Hrali jste si nekdo s necim podobnym? - Meli byste chut a cas se nekdo do toho taky pustit? Diky, Petr From: Michele Bavaro [mailto:mbavaro na onetalent-gnss.com] Sent: Sunday, October 07, 2012 10:19 PM To: Petr Holub Subject: Re: a module for accurate mapping for OSM Hello Petr, Thanks for getting in touch! It will be very difficult to guarantee 1m accuracy (standalone real-time or differential post-processed) under dense foliage. Trees are well known to disrupt carrier phase information which is essential to achieve such accuracies. There is something about it on the German forum Kowoma: http://www.kowoma.de/gpsforum/viewtopic.php?f=1&t=3248&p=14992#p14992 Denga10 and Rappen10 are not suitable. But their _LogBt or _LogWi version instead might work: they have a battery which will last a longer than 6hr, logging, and Bluetooth - or pretty much any other wireless communication system. Of course both have raw measurements but their carrier phase will be impaired under foliage as I wrote above. I am finalising the firmware in these days (today I had logging with fatfs and USB MSC running simultaneously finally) but -as it is quite experimental- I cannot really claim that is a "set and forget" kind of solution. Firmware is open though and if you have programming skills you may modify it to your needs. I am also working on enclosures... I have the bare boxes but I need to hand-tool them. If you are happy to do it by yourself I can make a discount: http://www.onetalent-gnss.com/announcements/contributorsgetafreereceiver Unfortunately in any case I cannot make you a lower price than 250EUR+VAT+shipping as I hand assemble the boards and I don't sell many of them so I cannot reduce the price by producing volumes. I would love to help you but I understand that compared to a 100EUR logger it's a lot of money to invest. By the way, all new Mediatek receivers have 10Hz and very high sensitivity... the new MTK3333 also has Glonass. ST Microelectronics has the STA8088FG which performs very well and it's very low cost. Not to mention of course the usual CSR GSD4E and Skytraq Venus.. Have you looked at Fastrax, Locosys, and GlobalTop? I am very happy to help an enthusiastic mapper, please let me know if I can do something else for you. Best regards, Michele On 7 October 2012 18:52, Petr Holub <hopet na ics.muni.cz> wrote: Hello, I'm looking for a GNSS module that could improve precision of my mapping efforts for Openstreetmap - mainly mountainbike trails and hiking paths in the Czech Republic (http://www.mtbmap.cz/). Up to now, I'm using Sirf3-based Garmin 60CSx mounted on the MTB handlebar, and MTK3329-based Holux M-1000C logger switched to 10Hz, mounted to my MTB helmet. While Garmin 60CSx produces reasonably precise results while hiking open-air, it has significantly worse behavior in hilly forests and 1Hz operation is not sufficient for accurately tracking shape of paths when riding fast on my bike. Holux provides improves tracking of fast motion, but it has usually significantly worse accuracy (i.e., wanders more) despite being mounted on top of my helmet and thus having good sky view. Thus I'm looking for a reasonably priced (<400 EUR, ideally 200-300EUR) solution for more accurate logging of MTB trails, ideally at 4Hz or better and approx. 1m accuracy - obviously I don't need 1cm accuracy, but I'd like to have 1m accuracy even under forest foliage. I don't need precise real-time output (Garmin will work fine for real-time navigation; having, e.g., additional Bluetooth NMEA output to verify operation, would be nice for the module, but it is not conditio sine qua non). I assume to put the device on my helmet again, replacing the Holux logger - so I shouldn't be excessively large or heavy. An alternative solution would be to have the device in my backpack and only put the antenna onto the helmet. The device should keep working on a battery for at least 6hrs. The ideal workflow for me is to keep the device logging into its internal memory, come back home, download the raw log and post-process it using RTKLIB to achieve 1m log accuracy, without need for paying recurring costs (such as CZEPOS professional DGPS correction). I'm doing this effort as a hobbyist, thus minimizing operational costs is important. Is there some solution for this in your portfolio? I was looking into Denga10LogBt, but I'm not sure if it meets my requirements - to sum it up: - battery operated for at least 6 hours - internal RAW logging for later processing - 1m accuracy under foliage (50cm accuracy would be absolutely perfect) - 4Hz or higher frequency (about 10Hz is ideal) - either whole device or at least the antenna should be helmet-mountable - helmet-mountable part of the device (or complete device) needs to be weather-sealed and needs to have reasonable weight (I can try to build my own housing, but not sure...) - post-processing should avoid need for paid data (such as CZEPOS) - ideally use open-source software to process the data Maybe my wishes still can't be fulfilled, but I've just stumbled over your website and the devices really look promising... Thanks, Petr
nejsem si jist, zda si chces hrat s nekonvencnim merenim 2) nebo opravdu potrebujes jen lepsi klasicky pristroj 1). Nicmene problem lesa je predevsim vicecestne sireni signalu, utlum, terenni prekazky a voda po desti.
2) kodove mereni * frekvence vzorkovani signalu u kapesnich prijimacu neni problem 1hz, nikdy jsem se zatacenim nemel problem. Samozrejme v terenu vzdy existuje jista mira nejistoty/nepresnosti.
* Garmin Vista co jsem mel byl pro mapovaci trackovani temer nepouzitelny
* to o cem se pise nize v mailu je mereni fazove uplne odlisne od kodoveho, ma to sve vlastni problemy. Je hodne metod jak s tim pracovat... * abys mohl delat postprocessing potrebujes RINEX nebo podobny vystupni format a na korekce kdysi byl nejaky free sw, lze delat nad fazovym i kodovym merenim * samotne fazove mereni je nejednoznacne a ma problemy s prerusenim signalu
PS: jinak lesaci se touto problematikou docela casto bavi, napr: http://www.lesycr.cz/odborne-rady/granty-a-dotace/Documents/optimalizace_tds_pomoci_gps-web.pdf
Já jsem v autě dosáhl slušných výsledků s Qstarz BT-Q1000X na 5 Hz v kombinaci s funkcí Kalman Filter v http://www.efficasoft.com/gpsutilities/windowsmobile_professional.html. Největší gró tam asi dělá ten Kalman filtr. Možná že by šlo použít pro postprocessing toto: https://github.com/lacker/ikalman
co třeba tohle: http://gpspp.sakura.ne.jp/rtklib/rtklib_beagleboard.htm (viz první odkaz na pdf)
nejsem si jist, zda si chces hrat s nekonvencnim merenim 2) nebo opravdu potrebujes jen lepsi klasicky pristroj 1). Nicmene problem lesa je predevsim vicecestne sireni signalu, utlum, terenni prekazky a voda po desti.
2) kodove mereni * frekvence vzorkovani signalu u kapesnich prijimacu neni problem 1hz, nikdy jsem se zatacenim nemel problem. Samozrejme v terenu vzdy existuje jista mira nejistoty/nepresnosti.
* Garmin Vista co jsem mel byl pro mapovaci trackovani temer nepouzitelny
* to o cem se pise nize v mailu je mereni fazove uplne odlisne od kodoveho, ma to sve vlastni problemy. Je hodne metod jak s tim pracovat... * abys mohl delat postprocessing potrebujes RINEX nebo podobny vystupni format a na korekce kdysi byl nejaky free sw, lze delat nad fazovym i kodovym merenim * samotne fazove mereni je nejednoznacne a ma problemy s prerusenim signalu
PS: jinak lesaci se touto problematikou docela casto bavi, napr: http://www.lesycr.cz/odborne-rady/granty-a-dotace/Documents/optimalizace_tds_pomoci_gps-web.pdf
Já jsem v autě dosáhl slušných výsledků s Qstarz BT-Q1000X na 5 Hz v kombinaci s funkcí Kalman Filter v http://www.efficasoft.com/gpsutilities/windowsmobile_professional.html. Největší gró tam asi dělá ten Kalman filtr. Možná že by šlo použít pro postprocessing toto: https://github.com/lacker/ikalman
co třeba tohle: http://gpspp.sakura.ne.jp/rtklib/rtklib_beagleboard.htm (viz první odkaz na pdf)
Já jsem v autě dosáhl slušných výsledků s Qstarz BT-Q1000X na 5 Hz v kombinaci s funkcí KalmanFilterv http://www.efficasoft.com/gpsutilities/windowsmobile_professional.html. Největší gró tam asi dělá ten Kalman filtr. Možná že by šlo použít pro postprocessing toto: https://github.com/lacker/ikalmanDiky za tip - o Kalmanovic filtrovani jsem sice vedel, nicmene jsem si myslel, ze mi to moc nepomuze - u toho Holuxe mam problem v tom, ze on ma jakoby tendenci ustrelovat jakoby systematickou nikoli nahodnou chybou (tj. ne ze by ty body byly rozhazene kolem nejakeho spravneho prumeru, ale jsou ujete o nejak metry tam ci onam). Bohuzel nemam Windows Mobile, s nimiz bych to mohl primo pouzivat, nicmene jsem narazil na NMEA Processor, ktery by to mel taky zvladat: http://www.gpspassion.com/forumsen/topic.asp?TOPIC_ID=106662&whichpage=21 https://sites.google.com/site/victorpomortseff/nmea-processor ale jeste ho budu muset trochu prozkoumat, co to vlastne umi.
Já jsem v autě dosáhl slušných výsledků s Qstarz BT-Q1000X na 5 Hz v kombinaci s funkcí KalmanFilterv http://www.efficasoft.com/gpsutilities/windowsmobile_professional.html. Největší gró tam asi dělá ten Kalman filtr. Možná že by šlo použít pro postprocessing toto: https://github.com/lacker/ikalmanDiky za tip - o Kalmanovic filtrovani jsem sice vedel, nicmene jsem si myslel, ze mi to moc nepomuze - u toho Holuxe mam problem v tom, ze on ma jakoby tendenci ustrelovat jakoby systematickou nikoli nahodnou chybou (tj. ne ze by ty body byly rozhazene kolem nejakeho spravneho prumeru, ale jsou ujete o nejak metry tam ci onam). Bohuzel nemam Windows Mobile, s nimiz bych to mohl primo pouzivat, nicmene jsem narazil na NMEA Processor, ktery by to mel taky zvladat: http://www.gpspassion.com/forumsen/topic.asp?TOPIC_ID=106662&whichpage=21 https://sites.google.com/site/victorpomortseff/nmea-processor ale jeste ho budu muset trochu prozkoumat, co to vlastne umi.
Já jsem v autě dosáhl slušných výsledků s Qstarz BT-Q1000X na 5Hz v kombinaci s funkcí KalmanFiltervhttp://www.efficasoft.com/gpsutilities/windowsmobile_professional.html. Největší gró tam asi děláten Kalman filtr. Možná že by šlo použít pro postprocessing toto: https://github.com/lacker/ikalmanDiky za tip - o Kalmanovic filtrovani jsem sice vedel, nicmene jsem si myslel, ze mi to moc nepomuze - u toho Holuxe mam problem v tom, ze on ma jakoby tendenci ustrelovat jakoby systematickou nikoli nahodnou chybou (tj. ne ze by ty body byly rozhazene kolem nejakeho spravneho prumeru, ale jsou ujete o nejak metry tam ci onam). Bohuzel nemam Windows Mobile, s nimiz bych to mohl primo pouzivat, nicmene jsem narazil na NMEA Processor, ktery by to mel taky zvladat: http://www.gpspassion.com/forumsen/topic.asp?TOPIC_ID=106662&whichpage=21 https://sites.google.com/site/victorpomortseff/nmea-processor ale jeste ho budu muset trochu prozkoumat, co to vlastne umi.Takze - uchodit jsem to uchodil a Kalman filtrovani dela presne to, co bych od neho cekal - odfiltruje tu "nahodnou chybu" (v uvozovkach to pisu proto, ze to, co jsem vyse oznacil jako "systematickou chybu" je ve skutecnosti taky nahodna, ale jina nez to kratkodobe driftovani. Pokud chcete, muzu nekam vystavit priklad, jak to dopadlo. Petr _______________________________________________ Talk-cz mailing list Talk-cz na openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-cz
Určitě, ukaž, zajímá mě to. Já si tenkrát dělal porovnání současného záznamu Mediateku na 5 Hz s Kalmanem a Sirfu na 1 Hz vestavěného v HP RX5910 a ten Mediatek vypadal definitivně lépe.
Určitě, ukaž, zajímá mě to. Já si tenkrát dělal porovnání současného záznamu Mediateku na 5 Hz s Kalmanem a Sirfu na 1 Hz vestavěného v HP RX5910 a ten Mediatek vypadal definitivně lépe.
Takze 2 rychle priklady: - Kalmanovske filtrovani - priklad ze Zamilce, Holux M-1000C (MTK3329), 10Hz na kole; Kalamnuv filtr udelan NMEAprocessorem (predtim zkonvertovano do GPX GPSBabelem, protoze na tom puvodnim NMEA logu ten NMEAprocesor padal): http://arwen.ics.muni.cz/~hopet/tmp/osm-cz/kalman.zip
- Holux M-1000C (MTK3329) vs. Garmin 60CSx (Sirf3); nevybiral jsem nic specialniho, vzal jsem log ze vcerejsi kratke vyjizdky na kole http://arwen.ics.muni.cz/~hopet/tmp/osm-cz/60CSx_vs_Holux.zip Znacna cast te trasy (nikoli cela) je projeta tam i zpet, takze je pekne videt, i jak moc to lita mezi prujezdy.
Takze 2 rychle priklady: - Kalmanovske filtrovani - priklad ze Zamilce, Holux M-1000C (MTK3329), 10Hz na kole; Kalamnuv filtr udelan NMEAprocessorem (predtim zkonvertovano do GPX GPSBabelem, protoze na tom puvodnim NMEA logu ten NMEAprocesor padal): http://arwen.ics.muni.cz/~hopet/tmp/osm-cz/kalman.zip
- Holux M-1000C (MTK3329) vs. Garmin 60CSx (Sirf3); nevybiral jsem nic specialniho, vzal jsem log ze vcerejsi kratke vyjizdky na kole http://arwen.ics.muni.cz/~hopet/tmp/osm-cz/60CSx_vs_Holux.zip Znacna cast te trasy (nikoli cela) je projeta tam i zpet, takze je pekne videt, i jak moc to lita mezi prujezdy.
- 2012-09-08.gpx: puvodni mereni zaznamenane jako NMEA proud na Androidim telefonu, prekonvertovane do GPX GPSBabelem - 2012-09-08_kalman.gpx: data prohnana pres Kalmanovo filtrovani
- 2012-10-13.gpx: Garmin 60CSx - 2012-09-11_09-10-32.nmea: Holux M-1000C
Je videt, ze oba pristroje se s tim popasovaly vicemene velmi dobre a skoro stejne!- 2012-10-13.gpx: Garmin 60CSx - 2012-09-11_09-10-32.nmea: Holux M-1000C
Je videt, ze oba pristroje se s tim popasovaly vicemene velmi dobre a skoro stejne! Dne 14.10.2012 22:34, Petr Holub napsal(a):- 2012-10-13.gpx: Garmin 60CSx - 2012-09-11_09-10-32.nmea: Holux M-1000C
Já jsem v autě dosáhl slušných výsledků s Qstarz BT-Q1000X na 5 Hz v kombinaci s funkcí Kalman Filter v http://www.efficasoft.com/gpsutilities/windowsmobile_professional.html. Největší gró tam asi dělá ten Kalman filtr. Možná že by šlo použít pro postprocessing toto: https://github.com/lacker/ikalman Nicméně pro ostré zatáčky v lese by pomohl jedině GPS v kombinaci s gyroskopem/akcelerometrem.
Já jsem v autě dosáhl slušných výsledků s Qstarz BT-Q1000X na 5 Hz v kombinaci s funkcí Kalman Filter v http://www.efficasoft.com/gpsutilities/windowsmobile_professional.html. Největší gró tam asi dělá ten Kalman filtr. Možná že by šlo použít pro postprocessing toto: https://github.com/lacker/ikalman Nicméně pro ostré zatáčky v lese by pomohl jedině GPS v kombinaci s gyroskopem/akcelerometrem.
Ahoj! Já jsem v autě dosáhl slušných výsledků s Qstarz BT-Q1000X na 5 Hz v kombinaci s funkcí Kalman Filter v http://www.efficasoft.com/gpsutilities/windowsmobile_professional.html. Největší gró tam asi dělá ten Kalman filtr. Možná že by šlo použít pro postprocessing toto: https://github.com/lacker/ikalman Nicméně pro ostré zatáčky v lese by pomohl jedině GPS v kombinaci s gyroskopem/akcelerometrem.
obavam bude
konskym hrbetu...)
mailing list
To by asi fungovalo v aute, ale na kole se obavam bude gyroskop/akcelerometr dost zmateny...
To by asi fungovalo v aute, ale na kole se obavam bude gyroskop/akcelerometr dost zmateny...
Otevře váš e-mailový klient. Odpovědi pak sledujte zde na webu.