Centje erbij centje eraf.
Alain had het over het platform ( transationel system ) en dat Here het niet heeft.
Lees het onderstaande en je weet waarom Microsoft Here heeft ingeruild voor Tom Tom.
Transactional map making system Slide #7
Let me get back to the subject, I promised to explain a little bit more what is transactional mapmaking system and why is it unique. Most map makers in the past and most map makers today produce maps in a batch system. What that means is you take a map, you store it in a map database, you make all kinds of edits in that map database and then once every month or once every week or once every quarter you basically make a release of that map database. So you have stored all those edits and in the release process, you quality check all the edits you have made. And that is done in a batch process. There is fallout. You correct the fallout and you go back to the release process and at the end of the day, you release your map once a week or once a month or once every quarter. That's the traditional way of doing things in mapmaking. The transactional mapmaking system works differently. What happens in the transactional mapmaking system, is that any change in reality can be taken through the whole process, including the quality checks and immediately be available as an update of the map to the application, which means that your database is actually continuously releasable and what that means is that with every change of your database, you create a new release. And it happens continuously, so you don't have to wait for a week or a month or quarter. It happens continuously and your applications can immediately use the freshest information that is in your database. That's quite a difficult system to make and we were very successful doing so. As I said, we introduced it in 2016 and I will show you what the results of that introduction are and what that did to our database and to our productivity. Automation delivers better maps faster, and at a lower cost Here's what we have today and it is again a staggering number, but we broke the record in August with 2.35 billion modifications in our database in one month. That's an enormous amount of modifications in our database. 85% of those modifications were automated. So 85% of those modifications went into the database in an automated way as a result of what we sometimes call fusion. Fusion is actually nothing more than combining all different sources, that can be probe data,
TomTom: MapMaking in the Era of Big Data 24 September 2019
5
that can be aerial pictures, that can be images and lidar data from our moma van, and fusing those to create transactions that in itself have multiple modifications that are applied to the database. Now in 2016, we were at the level of 250 million. Roughly. We are now at the level of 2.35 billion, which is almost tenfold. And what happened with the costs, because that's even more important. In the same period, we lowered the cost by a factor ten, which means that with the same amount of money compared to 2016, we can do ten times more upgrades of our database; ten times more modifications in our database, which is kind of a fantastic result and the foundation of that is our transactional system and a much higher level of automation, driven by the usage of machine learning.
Omzet automotive Here is nog veel groter dan bij TT, maar dasss bijna allemaal SD maps.
Maar met het platform en nu ADAS en dadelijk HD maps, gaat er veel veranderen.