As a navigation provider, we are one of the leaders in the development of the cloud-based electronic horizon. Currently we have completed initial measurements to determine data consumption for our electronic horizon and to provide automobile manufacturers with indications of the data volumes and costs that can be expected for the future.
Autonomous driving means vast amounts of radar, ultrasound and laser sensor data as well as cameras that continuously monitor the area around the vehicle in order to steer it correctly. It also means exact information about the area several hundred metres in front of the vehicle, which cannot even be sampled by the sensors yet.
The electronic horizon from infoware is supplied with all information about the road ahead from the cloud: the radius of curves, incline/decline, road class, speed limits, traffic and weather data, road conditions and more. That means the vehicle knows in advance what is coming, before it can even see and measure the route. This requires a highly precise road geometry to serve as a basis for the continuous comparison of data currently measured by the vehicle. Lane edges, bicycle lanes, guard rails, centre lines and lane marker lines all have to be supplied by the electronic horizon, down to every detail. If an obstacle suddenly appears, the vehicle has to perform avoidance manoeuvres and should therefore know whether there is a concrete barrier to the right of it or whether a bridge pier is coming up ahead, even if it cannot detect them yet with its sensors. Here the greatest possible redundancy improves safety.
infoware has now conducted initial consumption tests in order to provide automobile manufacturers with an indication of the data volumes that can be expected.
Driving on the motorway and in urban areas was simulated. Statistical ADAS road data and dynamic data were transferred in the test:
The required data volume was between 30 kByte/100km (motorway) and 227 kByte/100km (urban area). Optimisation measures will make it possible to significantly reduce the transmitted data volume. At the same time however, we expect demand for more and higher-resolution data. Therefore these figures already serve as a good indication of the expected magnitudes.
This means between 9 and 68MB per year will be transmitted with an annual mileage of 30,000km. That is not a lot, so the question arises whether digital road maps even have to be permanently stored in the vehicle at all. This would be a fundamentally new approach. Currently however, all automobile manufacturers are still assuming that a complete map has to be permanently stored in the vehicle. Often these maps are obsolete by the time the vehicle first gets on the road. Issues surrounding incremental map updates and the NDS map format could theoretically be made irrelevant as well. Good reasons to keep developing and analysing this topic. The excitement continues!