Our research topic: the cognitive and interactive robot learner able to achieve long term collaboration with humans and/or assistance to humans in an acceptable, verifiable and transparent manner by its human partners.
The integrative challenge of AI is fully realized in human-robot interaction and more precisely in human-robot collaboration when humans and robots decide and act together and share the space and the task.
One key concept appears to be essential, the concept of joint action. Some key abilities that successful joint action requires are: joint attention, mutual manifestness, joint intention as well as the ability to share a task and/or a plan. They are essential to help the human partner to understand the current state and the robot’s capabilities. They are also important to predict upcoming actions.
The scientific challenge here is to devise and build the cognitive and interactive abilities to allow pertinent, transparent, verifiable, and legible behaviors for a robot that is able to perform collaborative tasks with a human partner. Such an architecture should integrate an incremental learning system that will allow the robot to acquire new abilities for human-robot collaboration while ensuring transparency, explainability and verifiability of the overall decisional abilities and their evolution over time. Situated multi-modal dialogues should be used as a means to inform the human and ensure coherence. The system should then be evaluated in contexts where it is used to conduct joint action and/or learn or refine abilities with non-specialist users.
Other essential aspects that we have already started and which we continue to develop within ANITI are:
- a principled and long-term multi-disciplinary collaborative research with philosophers, development psychologists, ergonomists
- and a project-based deployment of AI-enabled robotic systems with potential users and mainly therapists as well as manufacturing and service industry.
This research is conducted in full interaction with the well-known robotics research teams and platforms already available at Toulouse.
We also seek to extend our current collaboration with development and ergonomic psychology colleagues and with the growing international community around human-robot joint action.
We also maintain and develop an international multi-disciplinary network around the key concepts of Human-Robot Joint Action and their implementation. Nine international multidisciplinary workshops (chaired by R. Alami and A. Clodic) have been organized regularly since 2014 (https://fja.sciencesconf.org/).
This chair activity is strongly linked to one of the ANITI core tracks and to the challenges and objectives defined in the ANITI “Collaboratibe AI” Integrative Program. We aim, in strong interaction within ANITI researchers and at the international level, at building, deploying, and evaluating our research program in three set-ups
- the autonomous teammate robot that will work in collaboration with human workers (the future cobots in industry)
- the cognitive and interactive assistant for frail people (with potential application to assistance and coaching in healthcare applications)
- the highly adaptive service robot in public spaces (with potential application to future transport systems).
Decisional issues for robotics
Decisional issues or Human-Robot Joint Action and Interaction
Human-Aware Task and Motion Planning
Theory of Mind for Cognitive and Interactive robots
Validation and Verification for Autonomous Systems
Evaluation of Assistive and Collaborative robots abilities
Programme : IA collaborative
Thèmes : #Langage #Neurosciences et IA # Robotique et IA
Porteur : Rachid Alami, DR CNRS, Laas
Contact : email@example.com
Site : http://homepages.laas.fr/rachid/
François Félix Ingrand (CNRS Senior Scientist) https://homepages.laas.fr/felix
Thierry Siméon (CNRS Senior Scientist) https://homepages.laas.fr/nic/
Aurélie Clodic (CNRS Research engineer)
Arthur Bit-Monot (INSA teacher assistant) https://arthur-bit-monnot.github.io/
Hendry Chame (Post doctorant)