New technology, new mapping standards

From Wazeopedia

Revision as of 03:54, 29 August 2021 by LostInMyMaps (talk | contribs)

This page is now a final draft and has entered a review period for feedback. This page will go live on 13 September 2021. Please discuss this page with your peers and leaders, and share your feedback with the community. If you spot any typos or errors, please feed them back to LostInMyMaps.

New technology

Previously, positioning on the Waze map used a technology called “matcher”, which depended solely on GPS from the device to tell the application where you are on the map. With the varying quality of GPS receivers in mobile devices, strength of GPS signals to triangulate position, high levels of interference in built-up areas, and the increasing prevalence of underground tunnels - the user experience could often deteriorate unexpectedly in situations where users rely on their navigation app.

Matcher.png

In 2020, Waze introduced a new core technology into the Waze app that fundamentally changed how the user's position was represented and moved on the map. This technology, referred to as the "snapper", relies on a highly accurate Waze map, combined with GPS, sensors on the mobile device (gyroscope and accelerometer), and bluetooth in tunnels, with intelligent code to provide highly accurate and predictive positioning.

Because the Waze map is an essential data source behind snapper, we need to change our approach of mapping to new methods.

New mapping standards

This led to a massive change in the way we now edit the Waze map. Where previously the map was basic and simple, and manipulated with map hacks. Many old mapping guidelines are now out the door as we move towards a new standard of mapping.

Revised Waze Editing Philosophy

Before new standards could be written, the fundamental principles of our mapping were revised from a long chucky block of text down to six simple points. These now guide us whenever we make a change to the map:

Usability

The Waze map is intended for navigation purposes only; it is not a general-purpose mapping application. We only make changes to the map that will add value to a driver using the app.

Representative of Reality

The map needs to be as accurate as possible and representative of reality: the position of the segment on the road, and the direction / angle of intersecting segments. Segments that do not meet these new principles could make the snapper hiccup, switch the arrow to another segment/route, and cause a poor user experience.

Only a Map

We are moving away from using the map as a way to control the app. A map should be a map; it should be accurate and represent the world.

Retention

The map stores an average time / speed it takes a driver to traverse a segment, for every 30 min block in a day, for seven days a week. This data is used for route optimisation and selection. When a segment is deleted, this data is lost. Given a choice between deleting or reusing a segment, it is better to "recycle" it.

Simplicity

We should avoid unnecessary complexity and clutter because it increases the risk of introducing errors, maintenance, and the learning curve (making it harder for new editors to get involved).

Communication and Documentation

An edit made one day, may not be clear a year later through the eyes of another editor. All editors are strongly encouraged to add map comments to explain unique or complex scenarios. E.g. increasing the lock level of a road, adding a plus intersection (+) where the satellite and street view images show a roundabout.

Road Quality

Alignment and angles

Previously with “matcher”, Waze would firmly stick to the road and keep you there. Accuracy of the road was not important, as long as there was a road nearby. If the driver exited the highway early however, the matcher might not detect the change for a long time.

“Snapper” however, considers a larger dataset to determine a more precise current and predicted driver position. One of the datasets mentioned above is a highly accurate map with accurately placed roads and angles of intersections. Snapper detects the early exit of the highway and adapts fast to real world driving conditions giving the driver the experience they expect.

Review this page on Road alignment and angles.

Removing excess junction and geometry nodes

Similarly to the previous heading, this point is further addressing the accuracy of roads on the map. Unnecessary splits between segments forming junction nodes, and excess geometry nodes on an otherwise straight segment add further expense to the intelligence behind the snapper technology.

Review this page on Removing excess nodes.

Dividing roads

In line with our goals of mapping closer to reality, there has been a massive push towards the division of roads. Take a look at the latest guidelines on dividing roads.

Review this page on Dividing roads.

Road Data

Lane guidance

A new addition to turn instructions, lane guidance value adds to the driving experience by providing the driver with confidence on approach and while executing turns. Lane guidance can be added to any intersection, however our preference is to focus on higher road types.

Review this page on Lane guidance.

Road width (coming soon)

This is another layer of data that allows snapper to determine which lane a driver is in, but will also quickly determine if the driver leaves the road to take another route.

Review this page on Road width.

Road shields

This feature displays shields or route symbols over the road on the map to help orientate a driver exploring the map. Road shields added to roads support their display in turn guidance discussed below.

Review this page on Road shields.

Turn instructions

The 2021 launch of turn guidance led to a refresh of the turn instruction experience within the Waze app. Map editors can now optimise the turn experience for drivers to provide extra information to ensure correct turns.

Review this page on turn instructions.

Junction boxes

These polygons have struck fear in the souls of editors for years. They’ve been traditionally incredibly difficult to use, restrictive on lower editors and at times, just hard to understand. But many improvements have come to junction boxes (JB) over 2021. And with the move to more complex intersections as we map roads and intersections more accurately, JBs are having an ever greater and essential role on the map.

Review this page on Junction boxes.

Rethinking intersections

Exits, forks, splits, way finders, and slip/turn lane geometry

Will the driver turn left, exit there or continue straight? Managing the driver experience at each and every intersection is the heart of snapper. Many of our old mapping styles need to change to improve the quality of data being fed to snapper’s intelligence.

Review this page for changes to Mapping intersections.

Remove map hacks

The map is no longer a mechanism to control the app, with micro-doglegs, detour prevention hacks and more now being banned. This ensures accurate map data is supplied to snapper’s intelligence. An active effort is required to remove these old styles for old problems, by instead favouring newer and more powerful features.

Review this page on Banned map edits.