Active sensing, interpretation of sensory information and manipulation in service robots

Aim of project
This project focuses on control requirements for service robots, especially on sensing and manipulative capabilities. Active sensing involves purposeful motion of the robot to obtain relevant information from the environment. Once the measurements are obtained they need to be transformed into symbolic form in the interpretation process. The other aspect of the research is two-handed manipulation and multi-fingered grasping. A multi-fi ngered gripper is developed for that purpose. Force sensing and visual servoing are used to perform service tasks. Moreover, Human-Machine Interface is under investigation. Both speech understanding and recognition of gestures are studied. The experiments are conducted on a two-handed robot system equipped with cameras and force sensors. The control software is based on the MRROC++ robot programming framework.

Expected results
In particular, the project will focus on:
Fast extraction of elementary features from images of objects, on the basis of which using fuzzy switching objects will be recognized and localized;
Control of concentration of attention and stereovision using independent eye movement;
Integration of visual servoing schemes using stand alone cameras and eye-in-hand configuration;
Utilization of force sensing in the process of recognizing objects;
Fusion of information from sensors with diff ering credibility,
Integration of senso-motoric (reactive) control with deliberative reasoning;
Control of motions of a multi-fi ngered gripper that will be developed within this project;
Two-handed manipulation of objects;
A list of elementary behaviors enabling the formulation of diverse service tasks will be defi ned (the basis for the defi nition of those behaviors will be transition functions used by the MRROC++ robot programming framework);
Multi-modal communication (three scenarios of human-robot interaction will be investigated: recognition of a silhouette of a human, detection head, mouth and eye motions, and recognition of hand gestures);
Speech recognition will concentrate on the symbolic description of voice signal.