AASS Seminar - Collaborative Adaptive Autonomous Agents

The research centre AASS arranges a seminar with Mirgita Frasheri, Mälardalen University.

Abstract

Intelligent agents are usually defined as autonomous software entities capable of perceiving and acting on their environment, with various degrees of learning and adaptation skills, which may or not be embodied, i.e.  having a physical a body through which to act. The element of autonomy can refer to degrees of self- sufficiency (ability to do a task without help) and self-directedness (ability to decide on ones own goals).  Moreover, autonomy can be defined in terms of dependence theory. It states that in case an agent A lacks a means to achieve a goal and depends on another agent B for it, then A is not autonomous from B with respect to that means. Also, dependencies between agents may change during run-time, thus their autonomy also changes. Adaptive autonomy (AA) allows agents to decide on their own autonomy levels based on specific circumstances. This research tackles the problem of determining when an agent should change its autonomy and enter into collaboration with other agents. To this end, firstly, a high- level agent framework is proposed which models the agent's internal operation. Secondly, the adaptive autonomous behavior is achieved by introducing the parameter of willingness to interact, composed of the willingness to ask for and give assistance. In the seminar, these ideas will be discussed, along with the results obtained so far in the course of the research.

Speaker's bio

Mirgita Frasheri is a PhD student in the Robotics group at Mälardalen University, Västerås, Sweden.  Mirgita received her B.Sc. and M.Sc. degrees in Computer Engineering from the Polytechnic University of Tirana, Albania in 2012 and 2014 respectively. After graduation, in 2015 she started her PhD studies at Mälardalen University. Mirgita's research topic covers adaptive autonomy, agent frameworks, and collaboration between agents.