Bosch, TomTom tap car radar to create maps for robo cars

Bosch, TomTom tap car radar to create maps for robo cars

Technology News |
By Christoph Hammerschmidt

A vehicle driving on autopilot (level 3 or higher) along a road, relies for accurate localization on two sources of information: its own sensors (radar, video, lidar) and a highly exact street map it receives from a service provider in the cloud. So far, these maps are composed of video signals from the cameras in the vehicles. In the cloud, these signals are processed to create a virtual map consisting of three data layers: The localization layer, the planning layer and the dynamic layer.


In a collaboration with TomTom, Bosch has developed all it needs to complement the localization layer with radar data. Towards this end, the radar echoes from the cars (all cars, not only autonomous vehicles) on the road (“Radar Road Signature”) are captured and sent to Bosch’s own cloud where the billions of radar echoes are combined to a radar map which in turn is passed on to the map provider’s cloud and, in most cases, to the cloud of the respective carmaker. From there, the current map data are transmitted to the automated vehicles on the road. These vehicles then use the virtual map to compare it with the data from their own local radar sensors to generate a trajectory plan. It helps them to navigate at an accuracy of a few centimeters, Bosch says. The main advantage of the Radar Road Signature is its robustness: unlike with maps that are based solely on video data for the vehicle location, the localization with the radar road signing works reliably even at night and in poor visibility. In addition, with the Radar Road Signature, only about 5 kilobytes of data per kilometer are transferred to the cloud. For video-based maps, the amount of data is at least twice as large. By 2020 at latest, the Radar Road Signature will be commercially available in Europe and the USA, Bosch announced in this context.

This is how the radar road signature is created and distributed

To compose the radar layer, Bosch is using its 77GHz radar sensors which are in widespread use already in today’s cars (as long as they are equipped with driver assistance systems such as automatic braking systems or distance and speed control systems (Adaptive Cruise Control, ACC). These sensors have a range of up to 250 meters, significantly more than video cameras which can only identify objects at distances not exceeding 150 meters.

To provide the data for the maps, the radar sensors needed to modified: For the abovementioned driver assistance systems, it is essential that the sensor can identify moving objects (for instance, other cars); static objects like traffic signs must be masked out. To acquire both types of radar targets, Bosch has to modify its sensors. “The next generation of our radar sensors will be able to provide the data required to create the radar road signature”, the company says.  

High-resolution cards are essential for automated driving and provide information that goes beyond the sensing range of the sensors. Unlike maps for today’s navigational devices, they consist of several superimposed layers:

  • Localization layer: Based on the localization layer, consisting of the Radar Road Signature plus additional video localization card, an automated car determines its position in a lane. For this purpose, for example, it compares information about objects that it has received via the environmental sensors with the corresponding information in the localization layer. Thus the vehicle determines its relative position to these objects.


  • Planning layer: The calculation of individual driving maneuvers (trajectory planning) takes place over the planning layer during automated driving. For this purpose, the planning layer includes information on road profiles, traffic signs and speed limits as well as curve radii and gradients. With the help of the planning layer, an automated vehicle decides, for example, when the track changes.
  • Dynamic layer: Information on all rapidly changing traffic situations such as congestion, construction and hazardous areas or free parking spaces are stored in the dynamic layer.

The individual layers of a high-resolution map for automated driving must be updated regularly – the dynamic layer even in real time. Experts assume that fleet of about one million vehicles each for motorways in Europe, North America and Asia-Pacific is required to keep a high-resolution map up-to-date. Based on on-board sensors of the vehicles, current data are generated for the individual layers during the journey. Via communication boxes such as the Connectivity Control Unit, for example, the data from the radar sensors from the vehicles enter the Bosch IoT Cloud via the manufacturer’s cloud. Bosch creates the “Radar Road Signature”, which is compatible with all common map formats. The integration of the radar road signatures into the overall map as well as their provision is made by third-prty service providers – in the example at hand, by TomTom.


Related articles:

Bosch rounds up components for automated driving

Mitsubishi Electric develops technologies for real-time 3D maps

Maps for autonomous vehicles

Will HERE maps reside in Amazon’s and Microsoft’s cloud?

BMW, Mobileye feed HD map data to service cloud

Volkswagen plans to create navigation cloud – with Mobileye’s help


If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :    eeNews on Google News


Linked Articles