Artisense offers a real-time mapping engine for robots and automated vehicles, designed to help automakers, robotic-, drone- and map-companies and startups rapidly create and update geospatial localization maps.
- Precise, automotive-grade point clouds
- Low data volume
- Geo-coordinates enable global positioning
- Highly accurate trajectories
- < 0.3% RMSE (Root Mean Square Error)
- No compromise in GPS-denied environments
- Semantic Segmentation
- Video frames delivered optionally
Ego-motion Point Clouds
How it works
High definition street maps necessary for highly automated driving require up to date data to keep their detailed information valuable. To achieve this enhanced functionality and keep the map operational it has to be constantly provided with a continuous stream of updates. Artisense integrates its real-time mapping engine on automotive-, space- or industrial-grade embedded modules requiring a fraction of the power of other computing platforms.
The Artisense dynamic map is refined using cloud-based geospatial intelligence (GeoInt) algorithms such as loop closure detection, visual-inertial optimization and map merging of local maps collected from multiple vehicle and robot sessions.
Artisense Dynamic Mapping fuses dynamic global 3D maps from crowdsourced local map data generated by any available sensor, whether it is a low-cost automotive cameras, LiDAR, Dynamic Vision Sensors (DVS) or Synthetic Aperture Radar (SAR).
Artisense further decreases the amount of data, which has to be provided as map updates. There is no need to transmit raw video data or full LiDAR point clouds to the cloud. Therefore, we focus on the relevance of the sparse 3D point cloud data, which is needed for vehicle or robot 6-DoF localization and has to be transmitted back from the cloud to the edge.