Kimera + Hydra Baseline (from MIT Spark)

This is a baseline implementation for our upcoming work in Metric Semantic SLAM and 3D Dynamic Scene Graph Representations, orginally created at @mit. Description: State-of-the-art techniques for spatial SLAM have led to the development of metric-semantic SLAM, which combines geometric mapping with semantic understanding to create accurate and robust maps in dynamic environments. This seamless integration of metrics with semantic information is enabled by the open-source library- Kimera. Unlike traditional SLAM approaches that focus solely on geometric primitives (such as points, lines, and planes), Kimera introduces a novel representation called the 3D Dynamic Scene Graph (DSG). Our work revolves around the implementation of the visual-inertial SLAM using Kimera which combines state-of-the-art techniques for visual-inertial SLAM, 3D reconstruction, object localization, human pose estimation, and scene parsing. In this report, we present experimental results and insights into the challenges and oppo