FiPy with other GPS sensors

[securigy]

As I recently tried and found out the GPS capabilities provided by PyTrack are not sufficient to even think about any commercial application with it due its TTFF (Tome To First Fix) that can range from few minutes to 12.5 and even 20 minutes according to the responses I received and the stuff that I read... and although this fix can be preserved - it is only can be done for 3 hours; so 3 hours off and you are waiting again for long minutes for a fix...

So my question is whether anyone on this forum tried connecting FiPy to NEO-M6x - or even better - to NEO-M8x series of GPS sensors from u-blox ? The latter has Assisted GPS (Online Assisted, Offline Assisted, and Autonomous Assisted) that provide not only 25-30 sec cold startup (TTFF) but some other features making it capable for applications of fast moving objects like cars, drones, etc. with accuracy of 2-3 cm versus 2.5-5 m as I have seen with other vendors.

So if anybody tried I would really like to hear about it, especially about electronic/physical connections and issues.

[robert-hh]

@securigy said in FiPy with other GPS sensors:
I tried the NEO-6M and it simply works. It connects to a UART port. You may use UART1 for that. It fixes after power on within 10-20 seconds.
On my breakout board is 4K EEPROM and a small battery, which is however empty. It is possible that it supports storing the satellite config data between power outages. The current consumption I measure is about 33 mA at 3.3V

[andyh13]

I am also experimenting with methods to increase GPS accuracy. I am interested in whether you've had success.

With regards to startup time: to what extent have you explored the various hot/warm start options? I believe the pytrack's GPS has assisted GPS as well. The library from Pycom doesn't seem to touch any of the GPS's features, you will likely need to delve into the module's datasheet.

With regards to accuracy:

If you are expecting cm level accuracy, you're talking about RTK mode, correct? With a base and rover? Some models also seem to support a moving base, but presumably that would decrease your accuracy.

I am currently experimenting with a couple different inertial measurement units. Unfortunately, an accelerometer alone (which is included in the pytrack) isn't enough to augment position information. A gyroscope and magnetometer are also needed, to provide accurate absolute orientation information. Combined with acceleration data (which I am sourcing from the integrated accelerometer in the IMU, with the hopes of it being more correlated to the other sensor data), one can of course double-integrate the acceleration to find the position, and combine the information with the GPS estimate. I have taken a shot at writing a Kalman filter for this, although it currently relies on numpy and thus probably can't run onboard. I'm going to continue to explore this.