Session 3: Humanoid Movement

Morphological and movement similarity in HRI dyadic situations

Benoît Bardy

EuroMov, University of Montpellier, Montpellier, France

Schizophrenia, autism, or social phobia are typically accompanied by social interaction deficits. The objective of the AlterEgo european project (www.euromov.eu/alterego) is the creation of a flexible interactive cognitive architecture, implementable in various artificial agents — avatars and humanoid robots - allowing a continuous interaction with patients suffering from social disorders by virtue of changes in behavioral (robot-based) as well as morphological (avatar-based) properties of that agent. In this presentation, I will present the scientific foundations of the AlterEgo project and its main predictions, grounded in the Similarity concept, originating from Neurosciences, Robotics, and Social Sciences. I will present the main results of the project, which show that patients functionally adapt their social motor interaction when they interact with agents (real or artificial) morphologically and behaviorally similar to them, as a route toward more natural interactions. These results have consequences for the implementation of digital cognitive architectures in the clinical context, and for the rehabilitation of socially deficient patients.

Parsimonious kinematic control for highly redundant robots. Application to generate antropomorphic movement.

Philippe Fraisse

LIRMM, Montpellier, France

We propose in this talk to introduce a new paradigm for kinematic control of highly redundant robots that provides sparse control vector in the joint space (joint velocity). Indeed, when a robot is highly redundant in comparison to the task to be performed, current control techniques are not “economic” in the sense that they demand, most of the time unnecessarily, all the joints to move. Such a behavior can be undesirable for some applications. In this direction, this work proposes a new control paradigm based on linear programming (LP) that intrinsically provides a parsimonious control strategy, that is, one in which few joints move. We will present the method as well as simulation and experimental results on the HOAP-3 humanoid robot. Finally, a comparison with the pseudoinverse will be detailed as well as a comparison with the postural control on human being showing the same strategy (ankle strategy) for small movement. 

Is human movement optimal? Application to the walk-to-run transition and sit-to-stand transition in cycling 

Bruno Watier

LAAS, Toulouse, France

Despite the apparent simplicity of a skilled movement, the organization of the underlying neuro-musculo-skeletal system remains unclear. A reason is the redundancy of the motor system: a given movement can be realized by different muscle and joint activity patterns. However, despite this conundrum, the central nervous system always finds robust and stable trajectories. Nowadays, no computational model can generate movements in terms of adaptability, robustness, accuracy and swiftness as fast as the central nervous system. One possible explanation is that the proper control criteria have not yet been identified. 

Our presentation will focus on biomechanics of the musculo-skeletal model, which allows us to determine the inner dynamic of the body. 

Then we will demonstrate how minimization of some cost functions may explain transitions in locomotion or cycling.

How highly dynamic motion is performed: an example case of parkour

Galo Maldonado

LAAS, Toulouse, France

Parkour is relatively a new discipline invented in the suburbs of Paris by David Belle and Sébastien Foucan at the ends of the 1980s in which practitioners have to overcome obstacles in the most efficient manner by exploiting human physical capabilities. This presentation aims to present the physical characteristics of parkour practitioners "traceurs" and to show how they succeed at performing highly dynamic motion.

From human locomotion to humanoid locomotion

Justin Carpentier

LAAS, Toulouse, France

In this talk, we first focus on the establishment of a centre of locomotion among human beings. This centre of locomotion is related to the notion of invariance inside the whole process of locomotion. We then show how to exploit this relative new concept in order to compute locomotive trajectories in the context of humanoid robots. Several real experiments on the HRP-2 robot conclude the talk.