Evaluation of a tactile interface to steer electric wheelchairs for people with neuromuscular diseases
Online Ressources:
- Online Video: https://www.youtube.com/watch?v=l1jAwkIqg50
Publication:
Youssef Guedira, Delphine Dervin, Pierre-Eric Brohm, René Farcy, and Yacine Bellik. 2019. Evaluation d’une Interface Tactile pour le Pilotage de Fauteuils Roulants Electriques pour des Personnes Atteintes de Maladies Neuromusculaires. In Journal d’Interaction Personne-Système, Volume 8, Issue 1, Special issue : the best of IHM’2018, December 2019. https://doi.org/10.46298/jips.5961
Context
Power wheelchairs remain an efficient means of regaining mobility for many people around the world. Unfortunately, some are not able to use power wheelchairs because of difficulties using a standard joystick. People with neuromuscular diseases who experience a loss in muscular strength find it difficult to use a joystick.
Contribution
To help people with neuromuscular diseases having access to steer a power wheelchair, we have designed and developed a tactile interface that can be used on a smartphone or tablet.
We explored steering power wheelchairs using this tactile interface with eleven users with neuromuscular diseases in free learning sessions. Four among them were able to take part in an experiment where we evaluated their kinematic performance between the use of the tactile steering interface and the joystick. The paper presents data and observations from both sessions and tries to detect tendencies and draw hypotheses that can guide further and in-depth clinical testing of the tactile steering for wheelchair users suffering from neuromuscular diseases.
Overall, the user performance with the tablet was close or the same as their performance with the joystick. In addition, the users reported a lesser level of physical demand of the tactile steering over the joystick.
Impact
This interface was presented to the general public several times at the Fête de la science and was tested with many disabled people thanks to the help of different healthcare centers such as MAS-CORDIA (Maison d’Accueil Spécialisée) Saint Jean de Malte Paris, IES (Institut d’Éducation Spécialisée) in Champigny-sur-Marne, the Functional Rehabilitation Center Le Brasset, and the Ellen Poidatz Foundation.
A framework of evidence-based practice for digital support, co-developed with and for the autism community
Context
As the outcome of an international project involving researchers from the USA, UK, Spain and France, a paper was published in the main autism journal:
V. Zervogianni, S. Fletcher-Watson, G. Herrera, M. Goodwin, P. Pérez-Fuster, M. Brosnan, and O. Grynszpan. A framework of evidence-based practice for digital support, co-developed with and for the autism community. Autism, 24(6):1411–1422, Aug. 2020. [DOI: 10.1177/1362361319898331] [hal-03296593]
This research was funded by the International Foundation for Applied Research on Disability. There has been an exponential proliferation of digital technologies dedicated to individuals with autism, with no specific mechanism by which they or their caregivers can easily gauge whether such technologies are beneficial or not, and whether they can potentially be harmful.
Contribution
This paper presented a methodological framework to empower the autism community (professional caregivers, family members, autistic individual and also designers) in assessing digital technology based on the various sources of existing evidence. It was the result of a study based on the Delphy methodology with a panel of experts from four groups of stakeholders: individuals with autism, family members, professionals and researchers.
Impact
The outcomes of the study were presented on a website designed to reach out to the autism community. In addition, colloquiums were organized as part of this project in Spain, Canada, France and the UK.
Marcelle: A Toolkit for Composing Interactive Machine Learning Workflows and Interfaces
Online Ressources:
- Online Documentation: https://marcelle.dev/
- Online Demos: https://demos/.marcelle.dev/
- Source Code: https://github.com/marcellejs/marcelle
Publication:
Jules Françoise, Baptiste Caramiaux, and Téo Sanchez. 2021. Marcelle: Composing Interactive Machine Learning Workflows and Interfaces. In The 34th Annual ACM Symposium on User Interface Software and Technology (UIST '21). Association for Computing Machinery, New York, NY, USA, 39–53. https://doi.org/10.1145/3472749.3474734
Context
The AMI team has been involved in interactive machine learning research for several years. With the growth of artificially intelligent systems, studying human interactions with machine learning has become essential. Yet, there were few tools to prototype, program and deploy interactive applications that engage users with ML models. To address this need, we created Marcelle, a software toolkit dedicated to the development of web applications fostering human-AI interactions.
The toolkit with developed in Collaboration with Baptiste Caramiaux, who was a member of Ex-Situ until 20XX, and then joined ISIR, Sorbonne Université. The first draft of the toolkit was developed to support a class of the HCI Master's program called Interactive Machine Learning, and the toolkit was then completely redesigned to support research and development, as well as pedagogy.
Contribution
With Marcelle, we proposed an architectural model for toolkits dedicated to human-AI interactive systems where end-users are tightly involved with the learning process, which is detailed in the publication above. The architecture is built upon a modular collection of interactive components that can be composed to build interactive machine learning workflows, using reactive pipelines and composable user interfaces.
We proposed and published a concrete implementation of this architecture as Marcelle, an open-source toolkit written in Javascript. Marcelle was declared with CNRS to Agence de Protection des Programmes (APP), and is available under the MIT licence. A number of online demos are available, and programming guides are available in the online documentation.
Marcelle has been mainly developed by permanent researchers, with contributions from Master's students and PhD candidates. It was selected as part of the PNRIA program, during which two part-time engineers contributed to improving its interoperability with Python ML Frameworks.
Impact
Marcelle has been well received at the UIST conference and is part of a growing endeavor to provide interactive tools that allow various users to manipulate ML Models.
Marcelle is used every year as part of the Interactive Machine Learning class of U. Paris-Saclay's HCI Master's program. It allows us to create applications that introduce machine learning concepts through hands-on experience. Students create original projects every year using Marcelle for implementation.
Marcelle is also used to support research projects, in particular to build research prototypes for user studies. Marcelle has been used in the PhD thesis of Téo Sanchez and Behnoosh Mohammadzadeh, and is used by PhD students in the AMI and Ex-Situ Teams as well as at ISIR.
We are hoping to keep developing Marcelle and grow its user base. We are in discussion with CNRS valorization service at DR4 and SATT regarding valorization opportunities.
Art-Science Project: Vera Icona
Online Ressources
- Video of the performance Vera Icona at festival Exils de Mahmoud Darwich at Institut du Monde Arabe in Paris, from 20th to 23rd of September 2018
- Capsule MemoreKall
- Interview of Véronique Caye and Michèle Gouiffès, organized by La Diagonale, Université Paris-Saclay
- Participation to a panel organized for the inauguration of the Scene de Recherche, ENS Paris-Saclay at Centre Pompidou the 19th of April 2019.
Publication
Michèle Gouiffès, Véronique Caye. The Vera Icona Installation and Performance: A Reflection on Face Surveillance in Contemporary Society. Leonardo, 2022, 55 (5), pp.439-444 https://doi.org/10.1162/leon_a_02245. (reference hal-04402090)
Context
The AMI team is frequently involved in art-science collaborations including scientists and artists. These reflections provide a new perspective on their scientific activities and question their societal impact. They offer an original way to spread scientific knowledge to a diverse audience.
Contribution
The project Vera Icona is illustrative of AMI art-science activities. This comes from a collaboration between Michèle Gouiffès and the artist Véronique Caye from the LVV (Laboratoire Victor Vérité). Vera Icona is an interactive installation/performance about the surveillance society. Inspired by the legend of the Vera Icona, which marks the beginning of the representation of faces as a political instrument, this project offers a metaphor of this control tool by staging a video surveillance system that erases the image of spectators' faces in real-time, thereby granting freedom of action to those under surveillance. The AMI team contributed to the design and provided scientific expertise in video analysis and face swapping (Michèle Gouiffès) as well as the coding for face swapping (intern Toky Rakotomalala). The performance/installation have been shown at festival Exils de Mahmoud Darwich at Institut du Monde Arabe in Paris, from 20th to 23rd of September 2018. The installation has also been shown at festival CURIOSITas in 2019. The work has been published in the international art-science journal Leonardo.
Impact
Véronique Caye and Michèle Gouiffès were invited to participate in a panel at Centre Pompidou to discuss their project. Their reflection has continued and an artist residency has been organized in February 2020 at Scene de Recherche (ENS Paris-Saclay) for their new collaboration La Tempête. The resulting show was presented at the Scene de Recherche in February 2020.