3D high definition urban map for smart city

The main challenge of generating a consistent and accurate map of the urban environment is building a reliable baseline vehicle trajectory. This is particularly difficult for complex urban environments such as metropolitan areas due to multi-lane roads, GPS blocking from high-rise buildings, and abundant dynamic objects and pedestrians. The core algorithm relies on Simultaneous Localization and Mapping (SLAM) technique in order to garner sensor data from a multi-modal platform and optimize the final trajectory.

The overall dataset includes a fully multi-modal phase for the robotics researchers. The environment covers mid-side downtown areas to the highly complex Gangnam district of Seoul. A typical residential form, a large group of a building complexes, is also introduced in the dataset. All of the abovementioned areas address the sporadic GPS reception issue, which was alleviated by Prof Kim’s research.