Grab about to work on enhancing OSM maps for Bangkok

Hello Mappers,

We are a team from Grab, who would start working on OSM enhancing the map quality for Bangkok city.
Our scope of work will limit to the areas served by Grab and not the entire city.
We only concentrate on creating / correcting the road network.

While we continue working on pre-analysis of the city, identifying best imagery to be used, posting the update here for help from local team in guiding us with relevant information on imagery and/or country specific mapping guidelines.

Will revert in the same thread after we complete our pre-analysis report

Thanks in advance

Grab Team

Did you participate in the “#GrabOSM mapathon” (see thread )?

Seems to be a different team, but we’re glad they’ve learned, eh? :smiley:

Hello Everyone,

In further continuation to our earlier post on mapping efforts to enhance OSM map for Bangkok, we have done pre-analysis on sample areas and found that there is more scope for adding missing roads, while existing data corrections are minimal.

We identified that Bing Imagery is well aligned with existing data and strava gps, however, in some areas, when zoomed in, imagery is found cloudy.
Hence in such instances we will refer to digi standard imagery to identify missing roads, but will come back to Bing imagery for creation and/or alignment.

We have created an issue in our project page and request for any feedback and/or suggestions
Here is the link for the same -

This issue also has the snapshots of the example missing roads instances we found.

We would be happy to clarify questions if any

Grab Team

A quick note -
We also checked with IODB (Imagery Offset Database) to check for any existing offsets recorded and we did not find any. We will keep a track and record relevant offsets while we continue working on the city
Grab Team

@Grab OSM

Hi, I went through the issue page. Some comments:

  1. I think that the ESRI and DigiGlobe imagery have been consistently better than Bing. Granted I haven’t looked at Bing since DigiGlobe came out. Either way, your call.

  2. Do you have an internal DB of GPS traces that you can use to verify if a road is public or private or even accessible? I’ve seen roads that have been barricaded by locals because they consider it part of their private property (or to deter thiefs and misfits from entering). One useful thing to do is to export these tracks similar to what strava has done and made it available as an OSM baselayer so that the community knows which roads are accessible for sure.

  3. Probably good to tag the change set so that we can keep track of your changes and review/comment as necessary.

  4. Although it’s useful to map from Satellite, you have people on the ground who are an incredible resource, I suggest that it’s essential that you incorporate them into your process.

Best regards

Hello Mishari,

Apologize for the delay in responding

  1. Though we state that we would use Bing as the default imagery, we will also refer to other images for any new developments and map accordingly
    In addition we also load strava heat maps to check for any alignment errors and correct them accordingly - Record the new offsets used(if any) into IoDB

  2. We have GPS traces but that would not decide if a road is public or private. We however follow other hints like, existence of gates, and/or large communities from satellite imagery to decide if a new road mapped is public or private. We would not modify the existing road type unless there is a strong evidence from Imagery. In such cases we usually contact respective mapper through their individual changeset comments and act accordingly

  3. While uploading a changeset, we usually describe the type of edits made and imagery source used. Please suggest if we can add more appropriate comments so that we can receive feedback and correct our processes

  4. We also take help from people on ground, at the instances where we are unable to judge if a road exists or not from satellite imagery but we have traces supporting a road existence.
    We also refer to the team for cases where we are not sure of the road type.

Grab Team