Abstract
By a News Reporter-Staff News Editor at Robotics & Machine Learning Daily News Daily News-Investigators discuss new findings in Robotics - Robotics and Automation.According to news reporting out of Zurich,S witzerland,by NewsRx editors,research stated,"Real-time detection of moving o bjects is an essential capability for robots acting autonomously in dynamic envi ronments.We thus propose Dynablox,a novel online mapping-based approach for ro bust moving object detection in complex unstructured environments." Funders for this research include Microsoft,Swiss National Science Foundation ( SNSF),Wallenberg Foundation,WASP Postdoctoral Scholarship,Swiss National Scie nce Foundation (SNSF).Our news journalists obtained a quote from the research from the Swiss Federal I nstitute of Technology,"The central idea of our approach is to incrementally es timate high confidence free-space areas by modeling and accounting for sensing,state estimation,and mapping limitations during online robot operation.The spa tio-temporally conservative free space estimate enables robust detection of movi ng objects without making any assumptions on the appearance of objects or enviro nments.This allows deployment in complex scenes such as multi-storied buildings or staircases,and for diverse moving objects such as people carrying various i tems,doors swinging or even balls rolling around.We thoroughly evaluate our ap proach on real-world data sets,achieving 86% IoU at 17 FPS in typ ical robotic settings.The method outperforms a recent appearance-based classifi er and approaches the performance of offline methods.We demonstrate its general ity on a novel data set with rare moving objects in complex environments."
NETL
NSTL