I have a few questions below. I’ll first provide some backstory, but feel free to just gloss over the questions; I cannot understate how much any help would be appreciated…
I’ve been continuously struggling to get a SparkFun GPS RTK (RPi PHat) to work for about 3 weeks now. I have it working somewhat reasonably, but I’m finding that the performance is rather underwhelming. At this point, I have also bought a SparkFun RTK SMA kit to create my own base-station. As such, my base station is no more than 20m away from my robot.
I should also mention that I am not using a caster. I have been sending RTCM3 bytes from the base station to the rover via LoRa or sockets (WiFi).
In some cases, my covariance (as reported by the unit) will be very low, despite the unit being 5 meters away from the reported position. Moreover, I can sometimes see drifting despite my robot being totally stationary. Due to the stated success of so many others, I am very surprised to be seeing the positional drifting.
Questions
-
I have a WiFi antenna and a LoRa radio that are mounted close to both boards. Could there possibly be cross-talk/interference? Could it be an issue of cross-talk from motors (I have relatively powerful motors that are mounted about 0.3m away from my board and about 0.7m away from my receiver)? That being the case I will sometimes see drift when the motors are not active.
-
Is it possible that a couple of bytes dropped here and there in the correction data are causing dramatic drifting. I can see that the number of bytes being received by my robot lags slightly.
-
How important are wheel ticks? Right now I am fusing my GPS output with wheel ticks through ROS. Could it be that I need to feed my RTK board a wheel tick signal in order to have a more stable output?
-
Does the receiver matter too much? I am using the GNSS Multi-Band Magnetic Mount Antenna for both of my modules. I understand that there are fancier receivers, but I was under the impression that the one would be sufficient.
-
Is it possible that I am receiving signals that have bounced off some other surface? I am pretty much out in the open with the exception that both modules are situated slightly close to a pool and that the base station has a glass panel somewhat close to it. In both cases, I am using the metal plate that comes with the modules. The other reason that I ask is I have calibrated my base station with both the survey-in as well as the PPP process (12 hours). When I calibrate with survey-in, my points seem to have drifted ~1m in a specific direction, which causes me to think there is some interference/reflection pulling measurements in that direction.
Anyways, I am at a loss. I apologize for such a long list of questions, but I would really like to get my robot up and running asap. Would PointPerfect help in my case?