Plug-n-Play Reasoning Components Enabling Team Formation Among Autonomous Agents

In this SBIR Phase I project, we are proposing a capability for the human operators to control the level of decision-making autonomy assigned to the robotic agents performing space or planet exploration, while encouraging team behavior among them. The level of autonomy is governed by a set of reasoning components or objects that interpret the incoming messages (e.g., task directive, request for information etc.) of a robotic agent. By replacing these components called role objects, the agent's planning autonomy can be altered to a level that requires human intervention to certain degree. This approach is efficient than the human operators adding and replacing the generated tasks after the planning in order to change the agent's plan. Reasoning components are formalized based on peer-to-peer relationships (e.g., trust, dependability and authority). These relationship parameters can be updated with time encouraging team behavior as orthogonal to controlling agents' autonomy behavior. Having plug-n-play reasoning components in the agent's planning layer or task sequencer/executor are also appropriate within the context of NASA's 3T architecture for autonomous control of robotic agents.