Real-world implementation of an Auction Behavior-Based Robotic Architecture (ABBRA)

摘要

Robots for real world applications face multiple challenges due to the dynamic nature of the environment. These challenges require that robotic architectures can handle multiple goals, some of which may be conflicting, and can also handle time-constraints for completion of those goals. In addition, the architectures need to be flexible enough to change their priorities should a change in the environment require it. The Auction Behavior-Based Robotic (ABBRA) architecture proposed in this paper meets these requirements by allowing the robot to handle dynamic changes in the environment while making the most suited decision for the robot at that moment. This paper details two new features of ABBRA and proves that the architecture can function in a real environment. ABBRA supports mapping and the use of clock (real) time for enforcement of time-constraints of goals while handling multiple conflicting goals simultaneously, as well as dynamically adding a new goal during run time. The paper presents the results of five real world experiments using the Segway Robotic Mobile Platform (RMP) in which the proposed auction behavior-based system is deployed in a real and dynamic environment, with dynamically changing constraints.