HD Vector Tiles power data efficiency and accuracy for autonomous vehicles
By: Eric Gundersen
Mobileye, an Intel Company’s, RoadBook™ provides lane level maps for semi- and fully-autonomous vehicles. RoadBook™ is now available for data-efficient distribution into vehicles through our HD Vector Tiles.
Precision maps provide a critical source of value across the vehicle automation spectrum, enhancing adaptive cruise control and lane-keeping support systems (L2 → L2+) to fully autonomous vehicles (L5).
“Through the cooperation with Mapbox, we enabled a large-scale, low data-rate, cloud-to-car distribution of the RoadBook. This was achieved with the VectorTiles format and by implementing ‘pre fetching’ logic on the agent vehicle” — Erez Dagan, SVP Advanced Development and Strategy at Mobileye
RoadBook™ provides a highly-accurate, rapidly-refreshed representation of the static driving environment. This includes road geometry (lanes, drivable path, paths through complex intersections), static scene semantics (traffic signs, the relevance of traffic lights to particular lanes, on-road markings), and speed information (i.e. how should average vehicle speed adjust for curves, highway ramps, etc).
With this data, an automated vehicle is able to localize itself within the map, have a critical source of redundancy to the physical sensors, and anticipate conditions beyond the range of physical sensors.
“In order for the information contained in the map to be reliable for supporting partial/full autonomy, it must be updated with an ultra-high refresh rate to secure its low Time to Reflect Reality (TTRR) qualities. To address this challenge, Mobileye is harnessing the power of the crowd: exploiting the proliferation of camera-based ADAS systems. These intelligent camera systems are used as harvesting agents to build and maintain in the cloud a near-real-time accurate map of the environment.” — Erez Dagan, SVP Advanced Development and Strategy at Mobileye
RoadBook™ is based on Mobileye’s Road Experience Management (REM), a solution comprised of three layers: harvesting agents (any camera-equipped vehicle), map aggregating server (cloud), and map-consuming agents (semi and autonomous vehicles). The harvesting agents collect and transmit data about the driving path’s geometry and stationary landmarks around it. Mobileye’s real-time geometrical and semantic analysis, implemented in the harvesting agent, allows it to compress the map-relevant information. This facilitates very small communication bandwidth (10KB/km on average).
Mobileye cameras crowdsource live data to the cloud through current wireless communication networks, updating the map in real time. As the data is ingested, it’s fed directly into Mobileye’s private vector maps and then distributed to the vehicles.
This data is tiled in small pieces, distributed with low latency around the globe, and then partially updated in real-time as the road network gets smarter. Basically, this brings Snapchat-level scale for HD automotive maps, giving your fleet the latest up-to-date maps, in the fastest way possible, anywhere on the globe.
HD Vector Maps support encoding arbitrary metadata from Mobileye, also in a bandwidth efficient, encrypted, high precision format. In the vehicle, embedded Mapbox software decodes the high definition map and dynamically loads map data for 200 meters ahead. This gives the car effectively a look ahead of 200m with the data for localization and redundant perception.
Coordinates traditionally are represented in memory as double triplet (Latitude, Longitude, Altitude). Our HD Vector Map format turns real-world coordinates into a local tile grid. This means a long sequence of coordinates is now represented as sequence instructions. This radically saves bandwidth for streaming data to a car and reduces storage size for larger area coverage.
To take advantage of Roadbook data through Mapbox HD Vector Tiles, get in touch with our team.
Mobileye RoadBook: High precision HD Maps distributed at scale was originally published in Points of interest on Medium, where people are continuing the conversation by highlighting and responding to this story.