Title: Self-Management Framework for Unmanned Autonomous Vehicles
Author: Eskindir Asmare
Partner: Imperial College London
Supervisor: Morris Sloman, Naranker Dulay
Year of start: 2005
Year of end: 2009
Unmanned Autonomous Vehicles* (UXVs) need to adapt their behaviour to current context - location, activity, available resources such as battery power and available services such as quality of the communications link. They should be self-managing in that they have to recover or adapt to component failures and optimise performance to best utilise available resources. Additionally, a team of UXVs should cooperate to achieve a particular mission such as surveillance of a specific area or search for specific targets.
Our objective is to develop a framework which enables UXVs to manage themselves as an individual and as a team. Management of UXVs involves resource, task, behaviour, communication and team management. We make use of policy-based techniques to support adaptive management as the management strategy can be easily modified by changing policies without reloading the basic software within a UXV.
In brief, our UXV self-management framework addresses mission management, capability description, and communication management. A UXV has a capability description that describes its resources and the services it can perform. A commander, which could be a human or another UXV, receives a mission specification from its command base. Based on the mission specification and its capability description, a UXV is assigned to perform specific roles within the team which is assigned to a particular mission. This role assignment process results in a team of UXVs cooperating to achieve a certain goal. The team adapts to intermittent communication link disconnection as well as permanent UXV failures.