Unmanned Systems Technology 006 | ECA Inspector Mk2 USV | Antenna systems | Northwest UAV NW-44 | Unmanned ground vehicles | Navigation systems | Lunar X challenge

6 Mission-critical info for UST professionals Platform one Three companies have recently launched technology for high-resolution mapping in the cloud in a bid to provide driverless cars with more accurate locations. An understanding of road layouts and traffic rules (including speed limits and various road signs) is essential for the successful implementation of automated driving technologies, and precise measurement of positional information requires the collection of information on road markings, curbs and other features such as traffic calming measures. To that end the Toyota Central R&D Labs in Tokyo has developed a high-precision map generation system that uses data from onboard cameras and GPS devices installed in current production vehicles as well as driverless cars. This gathers road images and information on vehicle positions, and sends them back to data centres where they are automatically pieced together, corrected and updated to generate road maps covering a wide area. Using cameras and GPS gives a lower probability of error than existing mapping systems that use 3D laser scanners, as any positional errors can be mitigated by using image matching technologies that integrate and correct road image data collected from multiple vehicles. This is providing a minimum accuracy of 5 cm on straight roads, and has the added value of being updated in real time and scalable across millions of cars. Toyota plans to include this system in autonomous production vehicles in 2020, initially for use on motorways. It will also seek to collaborate with map makers to use the map data in services offered by the public sector and private companies. Chip designer Mobileye has taken a similar approach by developing a ‘crowd- sourced’ mapping system, tapping into its own image sensors in cars to provide a highly optimised communications channel. The Road Experience Management (REM) software runs on Mobileye’s EyeQ processing platforms that extracts landmarks and roadway information at extremely low bandwidths, about 10 kbytes per kilometre of driving. Software running in the cloud integrates the segments of data sent by all vehicles with the onboard software into a global map. Prof Amnon Shashua, co-founder, chairman and chief technology officer of Mobileye, said, “The low bandwidth of the model, and the fact that it requires only a camera, enables the creation and updates to be managed by a cooperative car crowdsourcing mechanism.” Mobileye is working with General Motors to integrate REM into vehicles it is about to launch, as well autonomous vehicles in the next few years. Mobileye is also working with Volkswagen to integrate REM into Volkswagen’s fleet. HERE, the joint venture between BMW, Audi and Daimler that acquired technology from Nokia, has also launched a crowdsourced mapping system composed of tiles that contain dynamic content layers. Each layer provides different details: lane- level information, dynamic road network and situation changes, and speed profile data. These combine static map content, temporary information from updates and analytics data. The system, called HD Live Map, uses data from multiple sensors on cars and by the roadside for real-time updates that can be layered onto the map in small files without having to update the whole image. Ten car makers are already using the technology in tests of autonomous vehicles, including Google and Ford in the US. Unmanned ground vehicles insight, page 56 HERE’s HD maps are designed for the level of detail needed by autonomous vehicles Driverless vehicles Firms unveil map techs February/March 2016 | Unmanned Systems Technology