Applying Dynamic Conditions to an Auction Behavior-Based Robotic Architecture

摘要

Robotic systems in the real world will work in dynamic environments and have to choose between multiple, and occasionally conflicting goals. In order to facilitate these two requirements an Auction Behavior-Based Robotic Architecture (ABBRA) was developed that allowed different behaviors to compete by bidding for control of a robot. Each behavior would bid with an activation level calculated from metrics based on the environment. This paper introduces two new features to the Auction Behavior-Based Robotic Architecture, which are the dynamic addition of a goal at run time and the ability to set time constraints on more than one behavior. Dynamic situations may require these new features in order to ensure a strict critical timing and adaptability to the environment. Coupling these two features together with the existing ABBRA system increased the goal selection performance across six test scenarios.