DIY 1cm accuracy RTK-GPS
This project claiming to provide 1cm GPS accuracy is getting a lot of heckling over on the Make Blog. A geophysical survey engineer claims this has little practical purpose without lots of time, pro survey skills and equipment. What does the mp3car community think about this?
Link to Make blog:
i really don't see a advantage to this from a car enthusiasts perspective--i am driving a car that is roughly 6 foot wide, and 10 foot long, and last i heard, a standard gps was accurate to about 3ft-- which is still in the path of my car...
i can see a use for it in though for unmanned vehicles/small gps guided robots--the 3ft deviation would be enough for the unmanned vehicle to side swipe other vehicles, or to run off course enough to get stuck
Perhaps not the correct link, but some linkes either way....
This is amazing. 1 CM is the accuracy you need in order to be able to use it for the purpose of "auto-piloting your car". 3Feet obviously is not gonna do it. Accuracy to 3 feet means 3 feet in either direction which actually means accuracy to 6 feet. Really GPS is accurate to 3 meters which is about 9 feet in either direction. Which makes it impossible to autopilot a vehicle on the road. This was an Xprize or DARPA competition not too long ago and if they had GPS accuracy like this it may have been possible for someone to achieve the goal with less complications.
Really though, I have trouble believing that type of accuracy exists.:)
Speaking from experience, even with 1cm accuracy (which is the accuracy of commercial DGPS systems like Novatel and the military PLGR systems) you will experience "pops" of gps, especially in lower coverage areas. This is where GPS will shift by 1-2 meters, and then shift back to accuracy after a few seconds. Many of the competitors in the last two DARPA challenges used DGPS, however without decent obstacle sensing and avoidance and in the case of the last one, the Urban Challenge, without some kind of camera based road detection you are pretty much guaranteed to have issues. Highly accurate GPS is nice to have, but certainly not entirely necessary unless you are trying to avoid having to go the world-sensing route.
On a definition note, technically DGPS doesn't count as GPS being super accurate. It's regular GPS, with extra information from a base station (who knows it's EXACT location) telling you the shift of the satellites based on it's own GPS device. So while GPS is not accurate to 1cm, if you have a fast enough computer and good enough antenna at a base station you can get much greater accuracy.
someday they will figure out how to get virtually perfect clock timing into a handheld chip, and all gps devices will be accurate to withing a few inches. thats probably a couple decades away still though.
Well, there is still the atmospheric effects...
Originally Posted by trader007
Error is inserted in to gps to reduce it's usability. << Disregard this, former president Clinton allegedly had SA turned off.
I had a former job working on DGPS (differential GPS) units for a shipping company. 3 meters of error could be millions of dollars of damage. Differential GPS uses radio broadcasting shore stations that know their exact position. Their broadcast is the correction for the error in the raw GPS data. It was accurate to <1m 99.9(maybe a few more 9s)% of the time.
Without being able reconcile clocks or use broadcasts from very well surveyed points I don't think there's any chance we'll see ~1cm accuracy. Furthermore, what would you use it for?
1) Using RTK for a more accurate GPS solution requires real-time data from a very accurately surveyed base station (source of the reference data). In order to use this type of system for vehicle navigation, many base stations would be required, as well as the ability to seamlessly hand off between stations.
2) Dual frequency receivers are generally more accurate than single frequency receivers, even using RTK, due to the ability to largely eliminate atmospheric and ionospheric effects. Dual frequency receivers are also much more expensive than their more simple counterparts.
3) Accuracy in a differential GPS setup (which includes RTK) is a function of the distance from the remote to the base station, i.e., the further the remote is from the base, the less accurate the solution. Differential systems state these errors in terms of a base error in addition to a certain error in parts-per-million related to distance.
4) Unlike errors in inertial systems due to drift, GPS errors are step functions, i.e., large instantaneous change, due to the rising or setting of satellites used in the solution set. This can be reduced by filtering the outputs, but still remain an issue even in professional setups.
5) Accuracy in a differential setup is still governed by the geometry of the constellation currently in view. Sources of error are reduced, sometimes almost to zero, but a poor constellation geometry can still lead to a larger positional error.
The company I work for has the most accurate GPS technology in the world, we can lay a CD case on a table and tell the different between one corner and the other. So this guy claiming accuracy of 1 centimeter is pure garble.