Our team is going to start a mapping initiative - “Road attributes”.
Road attributes improvements Initiative focuses on conducting editing activities in OSM for improving quality of such road attributes as speed limits, turn restrictions, turn lanes, oneways that will improve OSM data. We make edits such as missed or wrongly mapped on OSM speed limits, turn restrictions, turn lanes, oneways, which is identified automatically by the internal tools built by Mapbox developers based on GPS traces.
Here is list of features we commonly focus on:
Speed limits
Turn restrictions
Turn lanes
Oneways
Road geometry (if necessary)
We will use a unique hashtag - #mapbox_roads_attributes
For edits we will use all available sources on OpenStreetMap site itself, sources with ODbL license, Mapbox datasets and tools**.
**For cases where there is not enough ground truth data from the OSM (Backgrounds, Overlays, etc) we will use collected Mapbox telemetry data, GPS traces, knowledge of the on the ground test drivers, etc. to produce the edit. We are ready to share all sources that we will use for mapping in case and provide as much context as needed.
This project is ongoing and will be continuously active without any time constraints. Data improvements will be made worldwide. The results of our review made by our team in OpenStreetMap will be displayed quarterly in the mapping repository (separately for each category)
We will be able to answer all the questions that arise right under this announcement.
For cases where there is not enough ground truth data from the OSM (Backgrounds, Overlays, etc) we will use collected Mapbox telemetry data, GPS traces, knowledge of the on the ground test drivers, etc. to produce the edit.
Thank you for sharing. I wonder will it be possible to see from the changeset comment which of these sources were used for a particular edit? Can you share the details about the Telemetry (how frequently do you record positions/speed/direction in a trip? How many users/trips have to be aggregated in order to be considered “reliable”?)
Hi, @dieterdreist! Thank you for interest to the topic! For sure, we’re going to add used sources in the comment to changesets.
For the used sources, we are in a position to mainly use OSM sources for mapping and use telemetry data as additional source. Let me explain little bit wider:
We have models, that can highlight possible attributes based on the traces and sign detections. So we are going to confirm this detections with the sources in the ID Editor and make a correction. If there is no such information in the OSM connected sources we’re going to add edits only in case of hundreds+ detected cases for specific area. It will be hard to say exact numbers for all attributes that will be “reliable”, as it will depend on the low/high traffic, attribute and road situation.
We know, that traces will not guarantee 100% of accuracy, because some people still can drive through the oneway road accidentally or intentionally even if it is forbidden. But for example if there is a 100x difference in the traffic depending on the direction (5 cases in one direction vs 500 in another) - it is a case for us to look deeply to such difference.
This is reasonable. The only concern I’ve had in the past is that sometimes the other direction is just unpopular, not prohibited. This is particularly challenging to discern for alleys and other low-traffic roads. But you should be OK as long as you corroborate with other indicators, such as signs in street-level imagery or parked cars on the left side of the street (in a right-hand-driving country).
Make sure to involve the local communities in case you use telemetry so you can get pointers to other sources you can use.
E.g., in Oslo, Norway, there are streets with “motor vehicle forbidden, except buses” that are just routinely ignored by motorists. Google Maps have clear photography on their Streetview that it’s not allowed, but is not accepting corrections because their “telemetry” shows otherwise…
Hi, @SomeoneElse! Appreciate your time and interest! We’re mapping as a company and our mapping interests are spreed worldwide, we’re working with all types of edits, and we use our main Wiki page to describe our intentions and focus.
Sometimes we’re going to have more focused mapping with the growing amount of edits inside some geographies or specific layers and want to notify the community to get an advice or feedback. So, we used tags to have an ability to monitor the quality and also have faster turnaround about community feedback for specific types of edits. Also, we use GitHub open repo for the all details and updates that could appear.
If this posting option is incorrect or uncomfortable, it will be highly valuable to have your advice on how it could be done in the more convenient way! Thank you for your question!
Hi @Minh_Nguyen! That is exactly the reason why we do not fully trust telemetry only and are going to confirm the detections using additional information and other indicators! Thank you!
As part of the expansion of our project, we are starting to apply additional road attributes, such as accesses and road names, using the hashtag #mapbox_roads_attributes
Previously, when adding road names, we used the hashtag from another project—#mapbox_editing , as the number of edits was low. However, now that the volume of these edits has significantly increased, we believe it is more appropriate to include the application of this attribute in the current project.
For adding road names, we will utilize information from sources such as Overlay and TIGER Roads. To incorporate access attributes, we will use all available resources from the OpenStreetMap website, sources with the ODbL license, as well as Mapbox datasets and tools. This will allow us to further enrich the map with relevant and useful information.