A behavior monitoring architecture for a mobile robot

This dissertation presents a new agent-, behavior-based architecture for a mobile robot. Two issues of the new system are addressed: Behavior Adaptation that enables a robot to adapt its actions to cope with unproductive actions and Speed Adaptation that enables a robot to match its maximum speed with its internal condition. The system was implemented based on the multi-agent robot control architecture called the Intelligent Machine Architecture (IMA).; This new behavior-based approach coordinates behaviors based on a robot's internal condition and environment information. The architecture combines the subsumption and schema-based approaches together with a unique feature called Request or Redundant Mode. By monitoring a behavior's expected outcome, an unproductive behavior can request its redundant behavior to take over a task. The results show that a system with the new architecture yields better performance than a system without it. Moreover, the new architecture can add new behaviors, delete unused behaviors, and create new behaviors or tasks from existing behaviors at run-time, without reprogramming or restarting the entire system. With low-level monitoring, a mobile robot can adapt its speed corresponding to its internal conditions using a Heartbeat Histogram.