Search papers, labs, and topics across Lattice.
This paper introduces a hybrid pipeline for dynamic semantic mapping that integrates geometric mapping and localization with ontology-driven updates. By utilizing an external calibrated camera for homography projection alongside object detection and persistent tracking, the system constructs a dynamic semantic world model that adapts in real-time to sensory data. The incorporation of linear regression for correcting real-world coordinate estimates enhances the accuracy of spatial properties and semantic relations, significantly improving the robot's contextual understanding of its environment.
Real-time updates to semantic mappings using ontologies can revolutionize how robots interact with complex environments.
Semantic mapping plays a crucial role in the ability of a robot to interact with objects, operate and navigate a complex environment. The most common pipeline for semantic mapping consists of geometric mapping and localization (SLAM), perception, semantic fusion and semantic representation. However, more recent works also integrate a form of prior knowledge in their application, most notably knowledge graphs or semantic scene graphs, to improve contextual understanding of the environment. In this paper, we present a hybrid pipeline for semantic mapping. Our system incorporates an external calibrated camera using homography projection for geometric mapping and localization, combined with object detection, persistent object tracking and ontology driven semantic updates to build a dynamic semantic world model. Linear regression models are also used for correction of the estimated values of real world coordinates. The system continuously updates object instances, spatial properties and semantic relations based on real time sensory data. Ontologies are selected as form of knowledge representation due to their hierarchical structure, semantic expressiveness and support for dynamic world modelling.
