Overview

Learning maps of the environment with mobile robots has been studied intensively in the past. Most of the approaches proposed in the past, however, are purely based on the perception of the three-dimensional structure. In this phase of the SFB/TR8, we study the problem of learning 3D representations from a new perspective. The key idea is to learn such representations by the interpretation of human activity. We observe the individual activities of a human using an XSens Moven data suit. This device consists of various sensors attached to a suit that return their positions in space and allow to reconstruct a human's pose and movement. The goal is to develop a system that is able to generate a virtual environment of typical indoor environments with multiple floors and containing several relevant objects including doors, tables, and chairs.

To achieve this, we investigate the problem of recognizing activities from high-dimensional data suit information to associate activities or gestures with objects to be reconstructed. In a first step, we will rely on the recognized gestures and a known mapping from activities to objects and other reconstruction operations such as constraint insertion for walls or tables. Later, we plan to drop this assumption and we will implicitly reconstruct features of the environment from natural human activities.

To estimate the virtual environment, we investigate appropriate data structures to represent the reconstructed objects. These structures will provide effective insertion and modification operations. Moreover, as we will include the user into the virtual environment, they will provide appropriate techniques for the interaction including collision detection, for example. To be efficient in terms of memory, we will investigate approaches to detect symmetries and similarities in reconstructed objects. We will furthermore develop techniques to detect similar features of partially reconstructed objects and the objects learned thus far to allow for an effective completion. Additionally, we will include the human into the reconstructed environment to be able to check whether the current activity of the human is consistent with the reconstruction and to perform update operations in the case of inconsistencies. We will furthermore investigate approaches allowing the human to directly modify the scene by manually grabbing, replacing, and reshaping objects.