Towards adjustable autonomy for the real world
Journal of Artificial Intelligence Research
| Hi-index | 0.00 |
Given the challenges faced by agent-based systems including dynamically changing environments, uncertainty, and failures, agent research must explore techniques to make these systems ever more flexible and adaptive. Adaptive systems are more robust and often perform better given changing situations. One area of research that offers promise for improved agent-based systems is adaptive autonomy. In general, adaptive autonomy gives agents the ability to seek help to solve problems or take initiative when otherwise they would be constrained by their design to follow some fixed procedures or rules for interacting with other agents. In order to gain run-time flexibility and any associated performance improvements, agents must be able to adapt their autonomy during system operation. This paper provides a mechanism through which agents can achieve this end. The examples presented in this paper focus on adapting one dimension of autonomy, specifically the decision-making framework under which agents solve problems in their system. The communication protocol presented here allows agents to dynamically modify their autonomy at run-time by forming agreements with other agents about autonomy-altering interactions. Conversation state machines and example event traces are provided. This communication protocol forms a solid foundation around which systems capable of dynamic adaptive autonomy can be implemented.