Reproduce “Online Object-Oriented Semantic Mapping and Map Updating” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Nils Dengler, Tobias Zaenker, Francesco Verdoja, Maren Bennewitz
Researchers at the Humanoid Robots Lab, University of Bonn, developed an online object-oriented semantic mapping system that robustly creates multi-modal object representations (3D point clouds, 2D polygons, OBBs) and dynamically updates them in real-time. The system demonstrates improved accuracy in object shape and position, and effectively manages object additions and removals in dynamic environments.
Key insights
- Maintaining multiple object representations (3D and 2D) within a single map allows robots to select the most efficient and appropriate data for different tasks, enhancing versatility.
- Actively correcting data association errors through a refinement step is crucial for building long-term, consistent, and accurate object-oriented semantic maps in real-world settings.
- Tracking an object's existence likelihood dynamically enables robust map management in changing environments, distinguishing between moved objects, removed objects, and transient false detections.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.