English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Zendy Plus

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Zendy Tools

Zendy Open

Synthetic data and novel rendering techniques have greatly influencedcomputer vision research in tasks like target tracking and human poseestimation. However, robotics research has lagged behind in leveraging it dueto the limitations of most simulation frameworks, including the lack oflow-level software control and flexibility, Robot Operating System integration,realistic physics, or photorealism. This hindered progress in(visual-)perception research, e.g. in autonomous robotics, especially indynamic environments. Visual Simultaneous Localization and Mapping (V-SLAM),for instance, has been mostly developed passively, in static environments, andevaluated on few pre-recorded dynamic datasets due to the difficulties ofrealistically simulating dynamic worlds and the huge sim-to-real gap. Toaddress these challenges, we present GRADE (Generating Realistic and DynamicEnvironments), a highly customizable framework built upon NVIDIA Isaac Sim. Weleverage Isaac's rendering capabilities and low-level APIs to populate andcontrol the simulation, collect ground-truth data, and test online and offlineapproaches. Importantly, we introduce a new way to precisely repeat a recordedexperiment within a physically enabled simulation while allowing environmentaland simulation changes. Next, we collect a synthetic dataset of richlyannotated videos in dynamic environments with a flying drone. Using that, wetrain detection and segmentation models for humans, closing the syn-to-realgap. Finally, we benchmark state-of-the-art dynamic V-SLAM algorithms,revealing their short tracking times and low generalization capabilities. Wealso show for the first time that the top-performing deep learning models donot achieve the best SLAM performance. Code and data are provided asopen-source at https://grade.is.tue.mpg.de.

GRADE: Generating Realistic And Dynamic Environments for Robotics Research with Isaac Sim