maplab 2.0 – A Modular and Multi-Modal Mapping Framework

Integration of multiple sensor modalities and deep learning into Simultaneous Localization And Mapping (SLAM) systems are areas significant interest in current research. Multi-modality is a stepping stone towards achieving robustness challenging environments interoperability heterogeneous multi-robot with varying setups. With maplab 2.0 , we provide versatile open-source platform that facilitates developing, testing, integrating new modules features fully-fledged SLAM system. Through extensive experiments, show ’s accuracy comparable to the state-of-the-art on HILTI 2021 benchmark. Additionally, showcase flexibility our system three use cases: i) large-scale ( $\sim$ 10 notation="LaTeX">$\text{km}$ ) multi-session (23 missions) mapping, ii) integration non-visual landmarks, iii) incorporating semantic object-based loop closure module mapping framework.