Automatic decluttering of 3D maps

Motivation and scope

When a robot constructs a map of a new environment, it is often the case that some “clutter”, either in the form left-over traces of moving people or objects, or other objects that can be expected to move soon. Recently, a method was published that attempts to separate structure from clutter in 2D maps, and can be used to automatically remove the clutter (ROSE). This project aims to connect the output of this method to 3D maps.

Specific tasks

Given a 2D and a 3D map of the same place, use the ROSE method to “declutter” the 3D map. Compare to other recent methods for cleaning 3D maps.

Necessary skills

Good programming skills, preferrably in Python.

Benefits

Develop experience with 3D perception and interact with ongoing research projects on industrial mobile robots. Connections to companies such as Electrolux and Kollmorgen Automation.

Contact

Martin Magnusson, martin.magnusson@oru.se