Applying Dynamic Planning Frameworks to Agent Goals

Agent-based systems require flexibility to perform effectively in complex and dynamic environments. When appropriate, agents should be able to take initiative for completing assigned tasks in the absence of supervisory input. Agents should also be able to determine if they can efficiently achieve a goal alone or if they need assistance from other agents in their system. In response to these decisions, agents should be able to change the planning-interaction frameworks in which they plan for their goals. Sensible Agents, using Dynamic Adaptive Autonomy (DAA), are designed to meet these challenges for multi-agent systems. DAA allows Sensible Agents to control their planning-interaction styles, called autonomy levels, along a defined spectrum (from command-driven to consensus to locally autonomous/master). This gives agents the capability to dynamically form, modify, or dissolve goaloriented problem-solving groups. This paper provides an introduction to the Sensible Agent architecture, describes DAA, and illustrates the application of dynamic planning frameworks to agent goals with a detailed example.