Improve machine-usability of location_verify_method

[j-d-b]

Currently,location_verify_method, on the road_event_data_source in v3.0, is a text field with a very general description:

The method used to verify the accuracy of the location information

While there is an example included, there is too much flexibility in what ends up in the content of this field for it to be useful for machines. It seems, based on the example "Survey accurate GPS equipment accurate to 0.1 cm", that this is intended to be just a human-readable, text description of the method. However, there is valuable data (accuracy!) here which could be parsed by machines if structured differently.

I would like to hear from other data producers what they are using as a location_verify_method, as in the example there are two pieces of separate information:

1. Method: GPS
2. Accuracy: 0.1 cm

I think spatial accuracy especially would helpful to have as a number (units separate). This could instead be added alongside/instead of "estimated" or "verified" for start/end coords, as currently even "verified" doesn't given that much information, especially when location_verify_method is not provided (it is optional).

Curious other's thoughts as it seems there is more information that could be added to the road_event_data_source to benefit machines.

[davidcraig4300]

Jacob, I would be interested in know this as well. What you are asking for 0.1cm or 1mm accurate GNSS is extreme survey grade GNSS and it usually requires a static GNSS receiver for many minutes. Did you really mean 0.1cm? My belief is, this level of accuracy is not required at all. Honestly, I don’t believe 1mm accurate GNSS is usable for virtually anything. I would be thrilled if we could hold the WZDx event reports at 1 meter accurate GNSS. 1m accurate GNSS can be easily achieved with moving machinery and some type of dual-band differential GNSS (PPP, RTX, PP-RTX, etc.). We actually are at 1.8m accuracy in our GNSS for our Automated Driving vehicles. Using single band GNSS. I think we have to ask, what does increased accuracy buy us? From 1 meter accuracy to something larger like 10+ meters, we can accurately define individual lanes on the road where the larger accuracy cannot. The limit would be 1.825 meter accuracy, above this you cannot correctly identify individual lanes (DOT spec’d 3.65m wide lane as a reference), below this you can identify lanes. Going sub meter I don’t believe gives us any more usable information. Now, I think this would be fun to debate and get other’s inputs. Thanks, Dave

[j-d-b]

@davidcraig4300 the 0.1cm came from the example for a values of location_verify_method that has been in the spec since before v2.0.

I was not intending to discuss the level of accuracy required or desired, just the ability for the producer to specify the level of accuracy of their feed with a distance unit rather than as part of a longer, string text field.

[CraigMoore-Sea]

This should align with any recommendations about location verification coming out of the Workers Present group.

[sergebeaudry]

Adding those field would be beneficial for the data user to judge how they are going to handle the data.

• some equipment will provide high level precision as DavidGraig mentionned above. Usually a moving equipment will be able to provide a great accuracy with the caveat that it needs to be always powered. Not a problem on expensive devices.

• When we go with standing device: a device that is moved to a location and power up, and stay still. . The precision is not at the 1m level due to millions of factors. and it is going to take a while to get to a 1m accuracy. Just take your phone, turn it on, open a map software and see the precision, not great.. start walking.. it will become better. Are those equipment can sample 24x7 at 1Mhz, yes, however battery live and cost will become an issue. As equipment provider we have to balance between the precision and the overall cost and form factor of the solution.
  Do we need to know at 1m level where are the work zone? No for many applications, Yes for others.

So providing a source and an accuracy element would help.

[j-d-b]

Resolved in #199.