DynoSAM: Open-Source Smoothing and Mapping Framework for Dynamic SLAM
Jesse Morris, Yiduo Wang, Mikolaj Kliniewski, Viorela Ila
AI summary
Problem
Traditional SLAM systems discard dynamic objects as outliers, losing valuable environmental information, while the field lacks consensus on optimal formulations and standardized evaluation metrics. This gap hinders rigorous comparison and limits the practical deployment of dynamic object-aware navigation systems.
Approach
The authors built DynoSAM, an open-source factor graph framework that jointly optimizes camera poses, static maps, and dynamic object trajectories. It introduces a novel formulation explicitly encoding rigid-body kinematics and a frame-agnostic error metric to standardize performance evaluation.
Key results
- Novel Dynamic SLAM formulation explicitly encoding rigid-body kinematics
- Open-source modular factor graph framework for unified static/dynamic optimization
- Frame-agnostic error metric for standardized object motion evaluation
- State-of-the-art motion and pose estimation accuracy across indoor and outdoor datasets
Why it matters
Provides roboticists and researchers with a standardized, open-source benchmarking tool to advance dynamic object-aware SLAM for navigation, reconstruction, and autonomous exploration.
Abstract
Traditional Visual Simultaneous Localization and Mapping systems focus solely on static scene structures, overlook- ing dynamic elements in the environment. Although effective for accurate visual odometry in complex scenarios, these methods discard crucial information about moving objects. By incor- porating this information into a Dynamic SLAM framework, the motion of dynamic entities can be estimated, enhancing navigation whilst ensuring accurate localization. However, the fundamental formulation of Dynamic SLAM remains an open challenge, with no consensus on the optimal approach for accurate motion estimation within a SLAM pipeline. Therefore, we developed DynoSAM, an open-source frame- work for Dynamic Objects SLAM that enables the efficient implementation, testing, and comparison of various Dynamic SLAM optimization formulations. We further propose a novel formulation that encodes rigid-body motion model in object pose estimation as well as an error metric agnostic to object frame definition. DynoSAM integrates static and dynamic measurements into a unified optimization problem solved using factor graphs, simultaneously estimating camera poses, static scene, object motion or poses, and object structures. We evaluate DynoSAM across diverse simulated and real-world datasets, achieving state- of-the-art motion estimation in indoor and outdoor environments, with substantial improvements over existing systems. Addition- ally, we demonstrate DynoSAM’s contributions to downstream applications, including 3D reconstruction of dynamic scenes and trajectory prediction, thereby showcasing potential for advancing dynamic object-aware SLAM systems. Code is open-sourced at https://github.com/ACFR-RPG/DynOSAM