To implement the telexistence of humans onto the humanoid robot iCub, we pursue research in different directions, and we exploit several technologies that combine the robot's fundamental locomotion and manipulation capacities with virtual reality systems.
to exist elsewhere via
humanoid robot avatars”
iCub 3
It is 125 cm tall (50"), and weighs 52 kg (115 lbs). It possesses in total 54 degrees of freedom including those in the fully articulated hands and in the eyes.
iCub 3 features six-axis force/torque sensors, tactile sensors that act as an artificial skin on the upper arm and the hands. It possesses two cameras, a microphone on both ears and a speaker behind the face cover. A set of LEDs define the robot’s facial expressions.
Humanoid robots can emulate many of the human capacities. Manipulation, walking, vision, and hearing are the common capacities and perception senses that humanoid robots possess. Furthermore, the human-like shape of humanoid robots is best suited for the human environment, thus rendering them good candidates for achieving tasks in place of humans.
Humanoid robots can then represent a valid solution for all those scenarios where remote maintenance and emergency response require physical interaction between knowledge operators and the surrounding environment.
Thus, we aim at providing human beings with the capacity of telexistence through the humanoid robot iCub by implementing advanced teleoperation systems. These teleoperation systems allow us to transfer human motions in terms of locomotion and manipulation onto the humanoid robot iCub. We research and deploy effective teleoperation systems for industrial setting and emergency response scenarios.
How
Research on humanoid robot motor control
To transfer, or more technically to re-target, the human motions onto the humanoid robot, fundamental motor control skills must be implemented on iCub. We work on the so-called whole-body torque control, aiming to give the robot balancing, locomotion, and manipulation skills that can be teleoperated via human beings. To achieve this, we apply the fundamental tools of control theory.
Whole-body balancing
The use of feedback linearization control techniques framed in the context of quadratic programming optimization leads to the emergence of balancing behaviors for the humanoid robot. In these cases, the robot uses its entire body, i.e. whole-body, to balance while being compliant with human-robot interaction.
Whole-body walking
The use of predictive control techniques still framed in the context of quadratic programming optimization leads to the emergence of walking behaviors for the humanoid robot. Again, the robot uses its entire body, i.e., whole-body, to achieve the walking task.
Whole-body jumping
Although iCub hardware is not optimized for jumping, we applied feedback linearization control techniques that are still framed in the context of quadratic programming optimization to make the humanoid robot jump. In emergency response scenarios, jumping capacity may be fundamental to skipping obstacles and recovering and balancing quickly.
Integration of virtual reality devices
To allow a human being to exist in another place, we have to transfer some of the human capacities onto the humanoid robot (e.g., walking, manipulation) and give some of the robot senses to the human being (e.g., hearing, vision).
For these reasons, we are working on integrating several virtual reality systems to achieve this transfer.