WP4-14
Map Enhancement
ID | WP4-14 |
Contributor | ALMENDE |
Levels | Platform, Function |
Require | |
Provide | 2d/3d Map data with added information: quality, heatmaps, semantic information |
Input | SLAM output and pre-existing maps |
Output | ROS TF/map data, shared among multiple robots |
C4D building block | |
TRL | 4 |
Parent Building block | [] |
Contact | ludo at almende.org |
Description
There are various sources for the creation of information-rich maps/grids, which can be used for SLAM and navigation. This component explicitly focuses on a generic way of providing query-like access to this information, providing temporal, shared situational awareness. This is achieved by leveraging the current map meta-information (timestamps and frame ids), augmented by additional meta-information per map-layer, e.g. trustworthiness, resolution, capture-time-range, etc. Especially additional information on whether the information is based on predictors or real sensors is relevant. Using this extended meta-info, a query service is provided through which map-products can be obtained for usage in navigation and as input for SLAM algorithms. Input map information can come from a variety of sources, including multiple robots, existing 3d & BIM models, hardware-in-the-loop simulation, semantic maps, AI & ML trained models.
Improvements
Currently, there are some ROS libraries for map merging, however, there are only a few fully-functional solutions that aim at shared situational awareness at a map level. Furthermore, there are some experiments for an "environment descriptor", an approach for semantic maps. This component is aiming for a more generic, complete map enhancement service for seamless merging, enhancing, predicting maps, including more semantically rich logic.