How HERE HD Live Map perpetually heals itself
This was originally published as a LinkedIn article.
I have been asked, “Why do autonomous cars need a map to drive when humans have no need for a map?” But, in reality, you do. Between the first time you drive a road, and the twentieth time you drive a road, you develop a map – in your mind. That map is a combination of all the other times you’ve been down that road, and what you see happening around you during the current trip. That’s exactly what we’re doing with the HD Live Map.
More than four years ago, we started our research in Highly Automated Driving (HAD). We partnered with Daimler and asked ourselves some difficult questions. What type of data would be needed to drive automatically? What kind of next generation of map will be required to enable an autonomous vehicle?
The result, a car that drove itself over a 100km overland trip, was displayed at the 2013 Frankfurt Motor Show. That was happening at a time before the hype for autonomous cars became what it is today. But, this made us think seriously about our product vision.
So, we asked ourselves a question – What is the solution that we as a company want to build to enable automated driving? We answered it by creating the HD Live Map.
This name is very intentional, and it carries two important elements. HD, meaning that the map is super precise, far more accurate and detailed than what you have normally. The second element is Live. It is a live system with updates in near-real time. These two pieces come together to provide accuracy richness and freshness.
With these elements, the HD map and the Live updates, we can assemble a complete digitally driven picture of the world around us. Of course, the world changes, frequently. When reality changes, the map needs a way to correct itself. This is why we refer to the HD Live Map as self-healing.
Imagine a guard rail at the side of the road. In our HD map, that guard rail is part of the picture of the world. That picture is reflected in our cloud system, the HD Live Map. The cloud knows about the rail, so the car knows about the rail.
Now, imagine that the rail has been damaged – perhaps there was a collision, or a tree fell on it, or the rail simply decayed. In any case, the reality has changed, and the map needs to be able to sync reality without human intervention.
As a car drives by the damaged rail, it compares its sensor information to what the cloud has provided. When the data does not match, the car might make changes in how it drives, while also sending updated data to the HERE cloud in real-time. As more cars report the same information, the cloud intelligently determines that the map should be updated. This method of gathering, analyzing and redistributing information is highly reliable because the updates come from a crowd-sourced picture of reality.
In this way, an autonomous car uses the HD Live Map the way a driver uses their experience. How the road is navigated is based on all the data that has been collected beforehand combined with what the car is experiencing at that precise moment. What happens in that moment, in turn, affects how the driver will react the next time they travel down that road. Autonomous cars will behave the same way.
All of this is about trust. In autonomous driving, you have to trust your vehicle. With that, you have to trust the data your vehicle has. That data must always be fresh. A map for an autonomous vehicle can no longer be updated every year or every quarter, so we built a map that updates in minutes.
And these technologies are advancing as we speak. More and more cars carry advanced sensors, which means more data is collected. As the data increases, the accuracy of the map increases, as does the speed of the updates, which increases the level of trust that enables autonomous driving.
Dietmar Rabel is Director of Product Management in the HERE Highly Automated Driving group.
Sign up for our newsletter
Why sign up:
- Latest offers and discounts
- Tailored content delivered weekly
- Exclusive events
- One click to unsubscribe