ERC Advanced Grant project "Seamless Human-Robot Interaction in Dynamically Changing Environments" (SHRINE).
The SHRINE project is located in the field of robotics research with an emphasis on the interdisciplinary field of human-robot interaction. The project aims at providing future robots with interactive capabilities similar to those of humans. Joint action of humans and robots sharing one world is to be facilitated.
SHRINE concentrates on the following three research areas:
Focus on understanding, modeling and recognizing non-verbal cues that express human intention and emotion. The gained information will be used in 2 and 3.
Investigation and implementation of joint manipulation scenarios with special focus on closed kinematic chains and dynamic manipulation.
Realization of navigation in dynamic environments, e.g. human crowds, while considering human behavior when avoiding or approaching humans.
extracted from the midterm-activity report (Oct 2013)
The goal of SHRINE is to alleviate barriers that prevents today’s robots from smoothly acting in joint workspaces with humans in an efficient and socially compatible manner.
Based on observations in human-human interactions, social-psychological human-robot interaction mechanisms conveyed by speech and facial expressions are developed. Evaluation with an expressive robot head show increased user acceptance and the willingness to cooperate and interact with robots. Most important in this context is the introduction of a behavior control model that enables robots to successfully trigger empathy and prosocial behavior in human users towards a robot. Further improvement of user acceptance is achieved in the field of navigation in human populated indoor and outdoor environments with a special focus on human approach. A motion planning algorithm based on optimal control, which respects social guidelines, is proposed.
In order to enhance robots with more advanced manipulation capabilities, optimal control methods are applied to dynamic manipulation of objects with complex geometry while being restricted to simple end effectors. Beyond that, the SHRINE framework brings together these advances in dynamic manipulation with the research field of human-robot cooperative object manipulation. As a first step towards the manipulation of elastic objects, leader and follower controllers for cooperative swinging of complex pendulums are proposed.
Another major goal of SHRINE is to achieve a great leap forward in terms of nonverbal human-robot coordination in dynamic environments. Whilst addressing this challenge in parts already in the above research, we propose a pioneering deterministic and feasible approach for the prediction of other traffic participants in a partially structured nvironment, such as a driving scenario. While former works predicted vehicles independently of each other, this work explicitly considers the current environment situation.