Bosch and TomTom announce they have developed a new technology for creating high-resolution maps for autonomous vehicles. This uses a localisation layer compiled from radar signals. The maps enable autonomous vehicles to pinpoint their position to within a few centimetres in all conditions.
Contemporary self-driving vehicles use high-resolution maps to find their way and determine their exact position on the roadway. These maps were previously compiled using video data. However, this method has some limitations. For instance, both poor visibility and darkness can affect the quality of the video footage captured, which affects the quality of the map used by autonomous vehicles.
Radar signals
Bosch and TomTom have developed a new method to create such maps using radar signals. This has the significant advantage that the technology operates reliably and consistently under all conditions, even in darkness or poor weather conditions. In total, billions of individual reflection points are combined to create a detailed map that allows autonomous vehicles to determine their exact position with a margin of just a few centimetres.
The method has been in development since July 2015. The companies combine sensors from Bosch with high-resolution maps from TomTom. Bosch provides sensors with 77 gigahertz technology, which have a range of 250 metres. By comparison, video sensors have a maximum range of 150 metres under optimal conditions.
Registering static objects
The main challenge the companies faced was finding a way to use existing radar sensors for this purpose. Radar sensors have long been used in autonomous vehicles for vehicle support systems such as automatic braking systems and adaptive cruise control, among others. In these applications, the sensors detect moving objects. However, to create a detailed map of the environment and road, it is necessary to be able to register static objects as well. The existing radar sensors have been modified to make this possible.
The maps created using these sensors consist of three layers:
- A location layer - using this layer, an autonomous vehicle can determine its position on a roadway. This layer is composed based on signals captured by Bosch's radar sensors and data collected by image sensors. By comparing data collected with environmental sensors in a vehicle with matching data on the localisation layer, it is possible to determine the relative position of the vehicle to these objects.
- Planning layer - the planning layer is used to plan individual movements during autonomous driving. This layer includes information about the course of the roadway, traffic signs and speed limits, as well as curves and ramps. An autonomous vehicle uses the planning layer, among other things, to determine when the vehicle can change lanes.
- Dynamic layer - information about rapidly changing traffic situations are stored in the dynamic layer. These include traffic jams, roadworks, dangerous situations and available parking spaces.
Updated via the cloud
Incidentally, the maps created from these three layers are not created per vehicle, but shared by multiple vehicles and continuously updated via the cloud. Autonomous vehicles transmit data they collect via a communication module to the manufacturer's cloud environment. From this cloud environment, the data is forwarded to the Bosch IoT Cloud, where a 'radar signature' of the road is created based on the data. This signature is compatible with all conventional map formats. TomTom then integrates this signature into a map and distributes it.
Author: Wouter Hoeffnagel
Source: Bosch
Source photo: Bosch
