An Abstract Behavior Representation for Robust, Dynamic Sequencing in a Hybrid Architecture

Hybrid robot control architectures try to provide reactive and deliberative functionality by decomposing a complex system into layers of increasing abstraction and temporal complexity. In building complex systems that react quickly to dynamic environments, the division of components are generally characterized as a Deliberator, a Sequencer, and a Controller. This paper presents a descriptive representation of reactive robot behaviors and the environment that the behaviors anticipate and affect. This paper also presents a control algorithm that uses this representation to create a robust link between the Sequencer and Controller in a hybrid reactive control architecture. The behavior representation promotes robustness and modularity as being a semantic suggestion rather than a syntactical burden like that of the tasklevel control languages currently in use. Thus, it allows for reduced development overhead and duplication of work for system modifications.