Master 2015 2016
Stages de la spécialité SAR
EMBRACe Expressive Movement Body fRAmework and Computing


Lieu : University Paris Sud, LRI, Team Exsitu, Orsay, France IRCAM - team, Paris, France
Encadrant : Sarah Fdili Alaoui, LRI, Exsitu http://insitu.lri.fr/ , saralaoui@lri.fr Frédéric Bevilacqua, Ircam, Laboratoire de sciences et technologies de la musique et du son (STMS), ISMM http://ismm.ircam.fr/ frederic.bevilacqua@ircam.fr
Dates :01/03/2016 au 31/08/2016
Rémunération :The monthly allowance is 3,60 € per hour, 7 hour a day (the number of days for each month will be confirmed by HR).
Mots-clés : Parcours ATIAM : Informatique musicale, Parcours ATIAM : Traitement du signal

Cliquer ici pour vous authentifier


Description

>>Context : This work will take place in the context of a collaboration between LRI (Exsitu) http://insitu.lri.fr/ in Orsay France, IRCAM (ISMM) http://ismm.ircam.fr/ in Paris, France, and SIAT/SFU (MovingStories project) http://movingstories.ca/ Vancouver, Canada.

>>Introduction : Human Computer Interaction (HCI) have been inspired by recent developments within neuroscience validating the primacy of movement in cognitive development. However, one of the remaining challenges is the computational modeling of expressive movement. Movement models have emerged in domains such as non-verbal communication, sign language, motor control and dance. One of the most notoriously used and investigated system that clearly formalizes Movement expressivity is that developed from the work of Rudolf Laban, called Laban Movement Analysis (LMA) [Laban 1974]. It is both a somatic and embodied practice as well as an observational and analytical system. LMA historically emerged from dance but has been applied to various other contexts including HCI. Recently, several HCI researchers have explored modeling movement through LMA categories [Fdili Alaoui et al 2012]. While a number of LMA based computation models of movement are emerging, each adopts a unique approach based on a their own particular interpretation of the theory. Moreover, these models are constrained by the limited capacities of the technology. In these different approaches taken to approximate LMA categories, we notice a lack of coherence in defining salient features that can be mapped to LMA definitions of movement expressive qualities.

>>Goal of the internship : In this project, we aim to propose a methodology by which movement models as well as the technology can be remapped in order to diminish the gap between the experiential theoretical understanding of movement characteristics in LMA and movement data. This requires re-thinking both the LMA analytical system and the computational model.

From the computational point of view, the task is challenging. First, the continuous aspect of the movement must be taken into account with high temporal precision. Second, the movement expressive qualities must be reliably quantified.

The goal of the proposed internship is to adapt the LMA framework and to explore movement features that best fit the system. Probabilistic models, such as Hidden Markov Models, that have been used extensively for recognizing movement, will also be tested with high level features of movement qualities in order to characterize movement expressivity. A selection of movement databases will be chosen to evaluate these models. In particular, the possibility to use these models to create interaction techniques for the whole body expression will be explored. The outcome of the project is to define a framework and a computational model for the analysis of movement in HCI. We aim at integrating a more richly articulated movement modality within embodied and expressive interactions.

>>Program : 1) State of the Art and implementation of a library of movement features across modalities (motion capture, accelerometers, EMGs, MMGs…etc.)

2) Data acquisition : Contributing into an existing platform for movement visualization, annotation and a movement database.

3) Designing new theoretical model of movement inspired by LMA in correspondence with the computational features of movement.

>> Pre-requisite : − Previous knowledge on either : − HCI, interaction design, full body interaction, or − motion capture systems, motions sensors, or − signal processing, or − Machine learning. − Software : Max-MSP − Language : Python or C++ or Java.

Bibliographie

Sarah Fdili Alaoui, Baptiste Caramiaux, M. Serrano, and Frédéric Bevilacqua. 2012. Movement Qualities as Interaction Modality. In Proc of ACM DIS . Newcastle, UK.

Rudolf Laban and F. C Lawrence. 1974. Effort Economy of Human Movement . Princeton Book.