Thanks Brian. I agree that there is a lot of potential for open source projects like AOG to enable autonomous functions for the mass market. I believe this will especially have a large impact on emerging markets with limited resources.
I’m still getting up to speed on the technical details of AOG but am looking forward to understanding your challenges and how we can help.
I need to do some research on this and will follow up with my thoughts.
@PotatoFarmer Thank you
@torriem
More than 2 user mentions are limited on my account - sorry.
Regarding several comments on the ZED-F9R, we have recently made optimizations in the firmware to adapt to slower moving heavy machines and robotics systems. So it is now possible to use the integrated sensor fusion system, which can significantly improve performance in high multipath environments such as being close to trees or buildings. The IMU data is accessible by the external host, so it may be possible to use this directly with the AOG System.
It’s not clear to me what you mean by calculating virtual reference point for terrain compensation. I’ll look into this in the docs but more detail would help me understand.
The ZED-F9R reference points for the antenna and IMU are both configurable.
We’re keeping an eye on market potential for a dual antenna module. This could especially be a good fit for PCB size constrained like drones. Generally ground based vehicles have been ok with using two modules, but this is good feedback.
giotoma I will look into this
Kevin I would make sure you are running the latest firmware on your ZED-F9 as SBAS was added in an update about a year ago. Also ensure you disable SBAS in other continents as you can sometimes pick up satellites in other regions, degrading your performance.
ZED-F9R gives you the advantage of integrated dead reckoning, but ultimately either ZED-F9P/R standalone will both give you meter-level accuracy with SBAS. RTK is required for applications requiring cm level accuracy. Generally, our commercial applications leveraging ZED-F9 are using RTK corrections using either a local base or our PointPerfect SSR-RTK service that is delivered via either L- band satellite (using a NEO-D9S module) or IP (via a hotspot, cellular module, ISM forwarding from a gateway, etc). This is a continental stream that eliminates the need for a local base: https://www.u-blox.com/en/product/pointperfect.This can get you sub 5cm CEP95 in all of the US and EU with a good antenna. Disclaimer: It is a paid service, but I think it’s very competitively priced. Pricing details are on the web here: Thingstream
You are very good at outdoor positioning, but we expect indoor positioning devices from you 
best regards.
I don’t know the proper terminology. In the farming world we call it Terrain Compensation. Basically the antenna is on the roof, or otherwise above the ground on the vehicle and is subject to pitching and rolling as the tractor moves over uneven ground. In order to steer the vehicle we need to know what the GPS position is on the ground under the axle. In other words translated from the roof position to the ground position, with a lateral offset added that is proportional to the tilt of the tractor. This is what I referred to as a “virtual reference point.” That was a terminology that came from the F9R documentation. In AOG we don’t have a huge need for dead reckoning. Rather we require RTK fixes at all times. That’s why the F9R is not sufficient presently. RTK Float just isn’t good enough.
If the receiver could do the terrain compensation internally and output the GPS coordinates as if they were on the ground (translated by user-defined variables such as distance from axle and height above ground which vary from machine to machine), then AOG wouldn’t have to do any of that kind of calculation itself, or try to synchronize the IMU readings with the GPS fixes, as the receiver would do that internally all by itself.
Thanks for the explanation. This makes sense. I am meeting with the product team next week and will discuss this topic. Will follow up with any updates.
We have indoor positioning products now too! Using Bluetooth 5.1 Angle of Arrival, you can achieve ~1m accuracy with “anchors” placed in a grid about 10 meters apart.
For more precision, anchors can be placed closer, but there are diminishing returns. Applications could include indoor drone landing, vehicle parking, and general asset tracking.
I checked out your 5cm correction pricing, but its not very clear how it is priced. This could be a great service for many who do not want to maintain a base.
But what is the cost in real terms like $/month or $/hour. Commercial corrections are usually priced in day/ month/ hour/ increments.
“0.00001cents per event” Im not sure how to price check that.
RTK is awesome, but this would greatly simplify things for broad acre applications and really hilly land.
Our pricing is available on the web here in the Location-as-Service - PointPerfect tab: Thingstream
We have plans where you purchase a fixed number of hours of streaming, or unlimited plans.
Ex) For corrections streaming via IP, the unlimited plan is $19 per device per month.
Hourly plans start at $3.90 for 60 hours of operation.
We also have pooled plans but these are targeted to large fleets of vehicles in the tens of thousands of hours of operation per month
Another possibility, given how well AOG already does terrain compensation with good IMU data, would be just a receiver with an onboard IMU (which automatically calibrates based on GPS heading, and fuses gyro and accelerometer readings) that outputs NMEA where the IMU pitch, roll, and yaw rate numbers are synced to the time of the GPS fix. This is what the Panda arduino sketch is trying to do.
If its IP I think the second F9P works best already, though that is a great service and will help indeed.
More interested in the L-band delivered correction, for NEO. If only F9 could also get this service, It would rank with the offerings major Ag brands.
What does this L band service cost on the NEO?
Nice to meet you, I’m a Japanese user.
I am satisfied with the performance of F9P, but it is a pity that QZSS cannot be used with the RTCM output of the reference station (RTCM 1117 MSM7).
I don’t know why it can’t be used, but please consider it.
So after doing some reading the Neo D9s is an add on chip just for the premium L5 correction for the f9p.
So if it can get 5cm, please direct me to a breakout board that I can use or buy now.
Let us know all costs involved
Unlimited L band access is $30 per device per month
The C101-D9S is our eval board for the NEO-D9S which can be purchased from our distributors:
DigiKey, Mouser, RFPD, Future Electronics, Alpha Micro
I recommend contacting your preferred distributor to inquire on price and availability.
We also have several partners working on standalone D9S boards similar to what is available from 3rd parties with the F9P today. These will likely be available around October.
Very cool, premium L band correction in adverse terrain is a great option.
Do you happen to know the convergence time using this system?
What is the time drift / positional repeatability using these corrections. Seasonal, same year, year to year?
What datum system and epoch are used as the reference frame for this system, If one was to use this system as a fall back or sync base for jumpless transfer?
Hi Sir - Have one system running with a C099-F9P board. Was using it to spray burn-down this weekend. I have been casually working with the C102-F9R board. (you mentioned several improvements recently, what firmware version has these updates?)
I’m interested in the potential. Have been able to get data and look at how it’s working through U-Center. I’m working to try and take the UBX-PVAT sentence and convert it to an AgOpen PANDA or PAOGI sentence. For me, it would be very helpful if Ublox had an NMEA sentence that gave heading, roll and pitch that we could feed directly into AgOpen. HDT is standard for heading, I haven’t found a standard sentence for roll & pitch. PASHR is the closest I’ve found…if going there, just go to PANDA.
I am using the recent firmware. I will have to check on the other continent corrections. I was able to do more testing Saturday. It worked much better than the last time I tried. I seemed to have an offset. If I reset the ab line after turning and getting lined up, it held position well to other end of field. There were trees at end of field. I would like to do more testing away from trees. We use WAAS corrections for most of our operations. The 1m drift hadn’t been a big issue.
The first link wouldn’t work for me.
Thanks
Hi Nabeel
I have heard, that the 1.30 firmware has worse performance than the 1.13 firmware version
@PotatoFarmer I believe it is currently around 30 seconds, and will improve to under 15 seconds with a firmware update later this year. Reference frame is ITRF2014 current epoch. Accuracy over time will be within our CEP95 spec of 3-6cm.
@87yj HPS1.21 has these optimizations under the robotic lawn mower model
@Peter_S Any details you can share would be appreciated so we can review. I heard that some are having an issue with lower update rates. With 4 constellations in RTK mode, the rate decreased from 8Hz to 7Hz with 1.30 FW, but GPS+GLO+GAL should remain at 10Hz.