Using an Auction Behavior-Based Robotic Architecture in Order to Fulfill Necessary Functionality for Service Robotics.

摘要

Service robots have great potential for improving the quality of life and assisting with people's daily activities. Such robots must be capable of performing multiple tasks and schedule them appropriately while interacting with people over long periods of time. In this context, service robots will have to deal with tasks that require making intelligent decisions based on environmental data as well as dealing with different users. In order to achieve this, the robot must meet certain requirements. With a large number of tasks that the robot could be asked to perform it must be capable of dealing with multiple tasks whose goals may be in conflict with each other. This forces the robot to overcome the problem of dynamically determining the importance of each tasks in order to select one to pursue.;In addition, since robots may need to handle multiple tasks there is the possibility that robots may be faced with multiple time constraints for different tasks forcing the robot to handle a wide diversity of time constraints when determining the relevance of each task. A human operator should also expect the capability of adding a new task while the robot is running without having a detrimental effect on its performance. A problem arises because many robotic systems design the tasks directly into the architecture itself, therefore, making an introduction of a new tasks difficult. This must all be achieved while safely interacting with different users. This interaction can interrupt a process and may affect tasks that have a critical time constraint. Therefore, it is necessary that the robotic architecture be capable of asking the user to wait.;In order to achieve these requirements and overcome these problems this dissertation presents the Auction Behavior Based Robotic Architecture (ABBRA). This architecture uses an auction mechanism in a novel approach to determine the relevance of a task to run at that particular moment. ABBRA also allows the robot to handle multiple time constraints along with incomplete information found in dynamic environments. ABBRA is also capable of dynamically accepting new tasks while running without sacrificing performance of the robot. ABBRA can also handle scenarios with multiple tasks that have conflicting goals. This is possible because ABBRA uses environmental data to determine the relevance of various tasks therefore, reducing task switching and allowing the goal to complete. ABBRA has long-term autonomy and interaction with known and unknown users. ABBRA handles multiple user requests while dealing with potentially critical time constraints. ABBRA can track and choose between two different faces for interaction and allows safe interaction with people. ABBRA supports flexible interactive capabilities such as requesting that the user wait in order to complete a time sensitive task.;The architecture was validated on physical robotic platforms (Segway RMP, pioneer 3D), demonstrating that ABBRA has the capability of being versatile in a dynamic environment and provides the basis for allowing a service robot to interact with users while making the best decision for which tasks to platform.