How we tried to save electric scooter riders’ lives – and won the 3rd place

In December 2019, we got the chance to participate in EcoMotion’s “Mapathon” map-related hackathon, in Tel Aviv.

The goal Create an innovative and breakthrough application for autonomous vehicles and smart mobility technologies based on the future of maps in Israel.
The means a 90-km HD mapping section of 3 different cities in Israel (Hod Hasharon, Tel Aviv and Be’er Sheva).

In a nutshell

Supplied with centimeter-level HD maps road segments that contain geospatial information, such as road and traffic features (signs, lanes, sidewalks, etc.), we’ve created an application that increases the safety of electric scooter drives by aiming the planning and enforcement resources of municipalities and authorities at key risk geographic points.

Our project is based on an “intersection service” that analyzes the connections between the geo-entities that are considered potential risks and the driven routes. The potential risk areas of interest are presented in AnSyn and are designed to aim the relevant authorities at the places that need their attention.

Getting to know the data

The Fuel Choices & Smart Mobility Initiative in the Israeli government has the goal of making Israel a center of knowledge in the fields of smart mobility and fuel alternatives. As part of their efforts to support the local ecosystem, a national plan was created for smart mobility.
One of the most innovative and promising tools was the creation of an open-access data infrastructure that will support research and development activities in autonomous car technologies and smart transportation.

This pioneering set of data includes high precision at centimeter-level HD maps of 5 road segments in Israel. Each road segment is a loop of about 30 kilometers of mixed highway and urban roads with centimeter-level accuracy and rich descriptive information.

In our daily work at Webiks, we’re constantly dealing with geospatial information and geospatial technology, and we were excited about getting our hands dirty with this new kind of data. We started by taking our time exploring the data and getting to know the details that would help us reach our goal, whatever we’d choose it to be.

The data contained these four elements:

  • Accurate location data of the sensor (height, roll, pitch, etc.)
  • Images from a spherical 360° coverage camera
  • Vector layers that describe the road and traffic features, such as signs, bus stations, road lanes, etc. Each element holds the exact location and its azimuth
  • LiDAR Point Cloud, representing the 3D structure of the mapped scene

For more information about the HD mapping dataset, check out the website.

It all starts with an idea

It’s just that we weren’t quite sure what kind of an idea. Should we aim for solutions in the field of shared mobility? Autonomous vehicles? Driver assistance systems?
Many different concepts crossed our mind when we first started thinking — from a system that allows a driving student to debrief her driving lesson based on videos, maps and street photos to parking spot detecting apps, and even an application that allows mothers with a baby carriage to find routes best-suited to their destination.

And what about micro-mobility? As an electric scooter rider, I truly believe in the power of micro-mobility. I think that this is the future of transportation in cities — from bikes to scooters, privately owned or publicly shared — these services have the potential to solve some of the biggest problems confronting urban and suburban communities. A money-saving alternative that expands access to public transportation. And all of this without even talking about reducing our environmental footprint.

But with great power comes great responsibility. In the past years, there was a major rise in the number of casualties and accidents, including electric scooters in Israel. Laws and obligations were made by the government, some infrastructure planning work was done and executed, but it is clear that we’re not ready for the revolution.

So far, neither the rules and guidelines of the government nor the precautions taken by the riders helped to make the rides better, and many lives were harmed. We need something bigger that combines both the worlds of what’s going on in the field, in the drivers’ eyes, with a combination of the municipality and the government’s view.

And then we got it, our goal — an application that allows increasing safety in scooter rides by aiming the planning and enforcement resources of municipalities and authorities at key risk geographic points.

Making the dream come true

Firstly, we were assigned to our great mentor Ori Isenberg, GIS Technology Manager at Ofek Aerial Photography, who helped, advised, and pointed us to the best use in the data and for victory all along the way.

At this point, we started thinking about all the geographic points in the electric rivers’ routes that might be considered a risk or a good decision to make.
Looking through the user’s point of view, we arrived at these three elements:

  • Geo-entities of the roads — Markings that allows drivers to make an intelligent decision about their planned trip. For example, no entry signs, road arrow direction, elevated sidewalk edges, roads slope, the width of road lanes and sidewalks, the existence of a bike path, etc.
  • Actual images of the route —street images, drone images, high-resolution orthophotos. Every real piece of visual information that will let the users see the risks.
  • Additional informative layers – live traffic data, previous accidents, historic routes, recommended or dangerous routes as were created by the community and published publicly.

In order to understand the routes’ safety grades better, we decided to create an “intersection service” that would analyze the connections between the geo-entities presented and the route taken. The service would include “positive” and “negative” graded rules based on the entity that the route has crossed and its significance. For example, a positive rule can be riding in a bicycle lane, while a negative rule would be riding against the direction of traffic.

So what do we have so far?

  • A lot of informative data that would be accessible for the user to analyze and to draw conclusions from
  • A service that will allow the creation of optimal routes based on simple intersection rules

Let’s begin developing!

Building the application and developing the service

Looking at our needs, existing assets, and in perspective of the Mapathon’s time limitation, we found our open-source overhead image analysis application – AnSyn – useful to rely on as an app that contains most of the features we needed for the project.

AnSyn is a map-centered application that, after a bit of data preparation, allowed us to add editable layers to display. Furthermore, routes (defined as GeoJSON “LineString”) are presented as an annotation and are editable. AnSyn allows us to present heatmaps in it, and it’s already connected to Mapillary, a worldwide street-level imagery database.
That basically covered up most of the features we needed.

In our application, the intersection service is where the magic happens. The risk detection and indications can be considered the most critical part of this web app, as this is where we’re giving our added value.

To start with, we geographically indexed all the feature layers in a PostgreSQL database with the PostGIS plugin, and then we started to build our logics.
Each rule held information on the ride (the current location and the calculated direction of movement), parameters of a feature from the HD mapping layers (type, location, direction), and a grade (how bad or good is it, if the rule exists).

A small glance at some of the rules:

  • When crossing a no-entry sign that is facing the direction of the ride or when riding in a lane that is marked with arrows that point to the opposite direction of the ride, it means that the rider is going against traffic. This is marked as possibly one of the most dangerous things one can do on his ride.
  • When facing a bike trail sign, it is considered that the rider’s route is safe, so it’s marked as a good intersection.
  • Routes that make the driver step on or off a high curbstone are marked as a bad intersection.
  • Driving in a lane that has trees next to it by day or light poles by night means the driver had good visibility — which is marked as a good intersection.

And, of course, the way the service was built allows adding more layers and more intersection logics easily (such as taking the width of the lane as a parameter of danger).

As an output, all of the rules are being calculated, the route taken is being graded, and the intersections are marked on a map. Then, the user can further explore and investigate all of the information presented.

Two calculated routes and their identified intersections presented in AnSyn

All good things must come to an end

Looking at our final product and the outcome while preparing the pitch got us thinking about the relevance of the product. We think that any app related to micro-mobility users won’t be in great use and that the focus should be on the power of collaboration between the riders’ experience and the authorities. We trusted our product to represent the best solution to the problem.

Well, in the end, it all came down to the presentation on the final day. All the work that was done boils down to a 5-minute presentation in front of the judging panel. After much anticipation, the panel picked the top 3 teams that came up with an innovative product, and we were among them!

Posted in GIS