Case study

Pipeline for Compiling HD Automotive Maps

We rebuilt a map compilation to produce more NDS maps and compile them faster

Key features

  • Build feature-rich automotive-grade NDS maps

    Build feature-rich automotive-grade NDS maps

  • Validate compliance with the Navigation Data Standard

    Validate compliance with the Navigation Data Standard

  • Offer wireless map updates from the cloud

    Offer wireless map updates from the cloud

Industry:
Automotive, LBS
Headquarters:
Amsterdam, the Netherlands
Market:
Global
Team size:
70+ members
Cooperation:
2015 – present
Technologies:

Java / Python / Oracle / AWS

About the client

Our client is a global provider of location-based services and mapping solutions for automotive giants, technology companies, municipalities, and consumers. Being a trusted and long-time member of the Navigation Data Standard (NDS) association, our client delivers best-in-class NDS location services and software development tools for building embedded in-vehicle navigation and advanced driver assistance systems.

The world’s leading automakers and original equipment manufacturers rely on our client’s products and services to deliver the most comprehensive navigation experiences. These products and services include timely offline and over-the-air (OTA) map updates, live traffic feeds, routing, guidance, and point-of-interest search.

Pipeline for Compiling HD Automotive Maps

Our client came with the challenge

Back in 2015, our client maintained a map production infrastructure consisting of a vast array of Oracle Exadata Database Machines. These servers accepted multi-source geospatial data and compiled it into NDS 2.2.x-compatible maps. However, after winning a major contract with Germany’s biggest automotive company, our client envisioned the need for a change. Their map compilation environment turned out not to comply with the challenging requirements put forward by the new customer.

First, their maps had to correspond to a newer NDS specification, version 2.4.x, which offered a richer feature set. Second, apart from pure map data, the compiler was to process and output specific data for the navigation SDK that our client was developing. The third requirement was a radical increase in map compilation performance through the introduction of pre-production map data validation and full verification of NDS compatibility.

Intellias developed the solution

Intellias joined this project with a team of 16 software and quality control engineers. Our initial task was to migrate the map compiler from NDS v2.2.x to v2.4.x. The compiler transformed the source map data to the target navigation data in the strictly structured NDS format containing product databases, update regions, and building blocks. The migration was carried out in several milestones, including moving to a newer database architecture, correcting and verifying new and broken dependencies, and calibrating the map packaging and validation mechanisms.

To work seamlessly with in-car navigation solutions built with our client’s SDK, the maps, apart from pure geospatial data, were to contain specific data used by certain navigation functions. Our team upgraded the entire map compilation process to enable the correct operation of routing, guidance, point-of-interest search, and OTA map updates.

Another project milestone called for migrating the map compilation pipeline and the map testing framework from Oracle servers to Amazon Web Services. The new solution, the Hadoop MapReduce-based framework was a proprietary standard for running automated map tests, including validating the integrity and uniformity of map content. Adapting the testing framework for NDS 2.4.x and moving it to the cloud resulted in cost savings on each map testing session. This proved a significant contribution since each map was recompiled several times per release and the number of releases was constantly growing.

Improving the map compiler’s performance was another challenge our team succeeded in overcoming. Our client was acquiring new automotive map customers and constantly expanding their map coverage to new countries and regions, so they required faster map compilation speeds. After having quadrupled in size in the course of the year, our Intellias team reworked the map compiler. This, coupled with ample computing capacity in the cloud, allowed us to reduce the time for world map compilation from days to hours.

We’ve achieved great results together

The continuous success of our collaboration has resulted in Intellias gradually taking ownership of all product development operations including solution management, product ownership, software development, and finally testing and release management.

Our team is now migrating all map compilation and testing environments to the most recent NDS 2.5.x that’s destined to go into next-generation premium car models. The overall team size is 70+ at the moment, but there are plans to grow up to 100 members by the end of 2018 to maximize productivity and satisfy increasing development loads.

Intellias’ expertise helped our client

  • Extend collaboration on next-generation navigation solutions supporting the latest NDS 2.4.x and 2.5.x maps with existing automotive customers
  • Gain new markets and opportunities by increasing the coverage of NDS maps to 120 countries
  • Enable faster and more stable map releases by reducing world map compilation time from 2–3 days to 4–6 hours
  • Enrich maps with specific information that allows OEMs to build modern end-to-end navigation solutions
  • Move map compilation and testing environments to AWS for enhanced scalability, performance, cost efficiency, and stability
  • Increase test automation coverage for maps from 70% to 97%

The partnership resulted in increased efficiency and a 40% reduction in costs, while allowing in-house staff to focus on strategic projects. Intellias impressed with their solid talent acquisition process and quality standards. The agency’s proactive management kept projects moving forward.

Andrew, R&D Director at Global LBS company

Thank you for your message.
We will get back to you shortly.