Subscribe to Magma RSS

Multiagent System Team - Laboratoire d'Informatique de Grenoble

 

MAGMA – Socially Relevant Multi-Agent systems

The term Multi-agent system was originally coined to mean a collection of software agents interacting within a computer environment. From this beginning, the community’s view of MAS was extended to include other types of entities such as virtual characters, robots, and humans, all interacting in a broader interpretation of a system.  We have now arrived at a situation where our extended view of a social-technical system is composed of a multitude of different types of agents, interacting in mutually beneficial ways.  In this new social system, the term utility takes on new meaning, referring to something that is of value for the mixed agent community as a whole, or a sub-set of it.

In this light, MAGMA focuses on the study, development and use of autonomous agents (AA) and MAS to solve problems that address a social concern or issue.  Its goal is to produce AA and MAS that are of value to citizens, solving real-world problems and enriching society. Concentrating on AA and MAS that have an individual and social utility, we are interested in developing new roles for AA and MAS in society. In order to study and develop socially relevant AA and MAS, we must address a number of challenges:

Scientific challenges:

  • Analysis, modelling, and replication of human reasoning, social behaviours and emotional factors,
  • Representation and formalisation of artificial agents,
  • Personalisation of artificial agents and agent groups,
  • Representing long term relationships between human and artificial companions (social robots or virtual characters),
  • Creation and validation of well-grounded agent-based models,
  • Understanding society as a complex system, including the emergence of collective phenomena and collective intelligence, and the design of robust and resilient socio-technical systems,
  • Developing adaptive/human-aware planning algorithms,
  • Engineering large scale and scalable MAS,
  • Ensuring agents’ privacy and data confidentiality,
  • Micro and macroscopic observation of natural and artificial systems,
  • Analysis and prediction of system dynamics and social networks.

Social challenges:

  • Ensuring acceptability and/or believability of AA and MAS
  • Engendering Social intelligence (self and social awareness, including sincerity, empathy, politeness, etc.)
  • Promoting natural/easy interaction with humans

Approach:

MAGMA is concerned with developing AA and MAS that are anchored in the social world. This implies that we adopt a multi-disciplinary approach, engaging directly with researchers from other disciplines. Our research practices are those that actively include end-users and stakeholders.

  • Agent Based Social Simulation (ABSS),
  • Artificial Intelligence and Simulation of Behaviour,
  • Distributed/Real-time/Continuous/Interactive Planning,
  • Participatory simulation,
  • Affective Computing, Intelligent Interaction,
  • Intelligent Virtual Agents, and Social Robotics,
  • Cognitive modelling and Belief/Desire/Intention (BDI) architectures,
  • Logical formalisation of psychological and sociological theories,
  • Spatio-temporal analysis of big data flows using agents,
  • Multi-scale analysis and information visualisation.

Application areas:

MAGMA is concerned with developing practical applications of AA and MAS that stand to impact positively the well-being of society:

  • Energy management (smart energy and eco-behaviours),
  • Crisis and emergency management,
  • Companion agents for everyday life,
  • Coupled robot and service dog support for human mobility assistance
  • Web intelligence, Digital Homes, Smart cities,
  • Communication media analysis (News, blogs, social media, etc.)