Dynamical system approach for navigation using behavior coordination in indoor environments

摘要

Behavior based navigation has been an active research problem in the field of mobile robot navigation for more than a decade. The problem is to study each individual behavior and coordinate individual system behaviors to achieve the goal of the system. Competition or collaboration among behaviors to achieve the goals of a system requires arbitration and/or fusion of control signals. Arbitration in most of the systems is specified in terms of dasiaeventspsila. Discrete switching between groups of behaviors is achieved by detection of these dasiaeventspsila. Several individual behaviors are modeled using dynamical system approach. A simple rule-based system is applied as the coordination system for behaviors. In addition ldquoDesperationrdquo / FED UP behavior as in human behavior is modeled so as to instill a bio-inspired behavior to the existing navigation system. FED UP factor is intended to mimic common ldquodesperationrdquo in human behavior to find a solution to avoid working on the similar tasks continuously at a time period. The FED UP factor as in Human desperation can lead to Malefic or Benefic effects. However as seen in the results, the effects of desperation can be controlled, moderated and made beneficial. The behavior coordination framework illustrates the arbitration and fusion of behaviors which facilitates smooth transition between behaviors.