Introduction

Sometimes, we set out to solve one problem and arrive at a bunch of even greater discoveries along the way. This story starts with my curiosity about whether you can get a “GPS” track log underground - like in a tunnel or underground car park. GPS is our go-to tool for mapping most things that we can’t see on aerial imagery, but what can we do in places where GPS signals cannot be received? In the course of my investigation, I uncovered a few even more interesting insights:

• Even if you can code, it’s impressive what an off-the-cuff LLM prompt can build for you
• The openstreetmap.org site UI would work very differently had it been built in the smartphone era
• Capturing rich mapping data from stock vehicles with no extra hardware is feasible
• With relatively little effort, we can improve the effectiveness of GPS track log capture for OSM mapping

Oh, and I did manage to get that underground track log, but more on that anon…

Motivation: the desire to improve tunnel mapping

Mapping underground features in OSM can be challenging. Sometimes we are lucky - a tunnel or covered roadway may be a straight line between two known points on the surface. Perhaps the tunnel was built using cut-and-cover and we were able to establish the geometry during construction. But sometimes, we just have an underground linear feature with bends in it. We know where each end dives underground, but GPS signals cannot be received underground, so our traditional mapping approaches won’t help us.

Road tunnels, of course, are designed for vehicles, and many modern vehicles have moving map displays as part of a navigation system. When in a tunnel, many even show a plausible vehicle position that updates. Without GPS. How do they do that? They could simply infer movement along the mapped path of the road based on distance travelled. But they may also use more sophisticated dead-reckoning inferring direction from sensors. I have such a car. I wanted to find out.

… See full entry