Ing. Vítězslav Beran, Ph.D.

Improvements of Change Detection in Point Cloud Data and Visualization of the Changes by Incorporating Semantic Information

Czech title:Vylepšení detekce změn v mračnech bodů a vizualizace těchto změn s využitím sémantické informace
Research leader:Španěl Michal
Team leaders:Beran Vítězslav, Veľas Martin
Agency:Ministry of Education, Youth and Sports Czech Republic
Keywords:Robotic systems, environment perception, point cloud data, change detection, semantic information, human-robot interaction
The project deals with improvements of the change detection in point cloud data and its visualization, as a part of robotic systems, by using semantic environmental maps of the surrounding environment, and with a user friendly representation and visualization of these changes.

Typical approaches for the detection of changes in point cloud data obtained by sensors like MS Kinect or Velodyne LiDAR use only the measured 3D information about the object geometry (points, surfels). The changes are typically visualized by highlighting changed parts of the scene. The newly proposed approach will benefit from the semantic description of the environment what means description of particular objects, information on their class (floor, table), connections (a cup on the table), etc.

The prepared software solution will be able to identify changing (moving) objects in the scene and present them to a remote operator of the robotic system in more clear and precise way: "The cup is no more present on the table." The semantic information might also enable multi-modal communication with the operator by communicating the changes in the environment using synthesized voice messages, etc.

Project outputs:
  • Software solution for the change detection and its representation using semantic information.
  • Publication in proceedings of an international conference
  • Technical report describing the proposed solution and results of the cooperation project

Your IPv4 address: