Using snarpy to Connect a KR System to Pyro

This work involves the development of a layered, heterogeneous architecture for cognitive robotics that can be used to connect a knowledge representation (KR) system to Pyro (Python robotics). We have focused on connecting a Semantic Network Processing System (SNePS) (Shapiro & Rapaport 1992) agent to a Pyro robot, however, the framework we have developed, snarpy (an architecture linking SNePS with Pyro), is extremely flexible and allows for the integration of any formal knowledge representation, reasoning, and acting system with Pyro. Cognitive Robotics is focused on higher-level actions and perceptions rather than lower-level details of robot interaction with a world. Pyro is a Python-based environment for robotics (Blank et al. 2006; Blank, Meeden, & Kumar 2003; http://www.pyrorobotics.org ), and indeed, for cognitive robotics, that provides underlying support for a variety of actual and simulated robots by making available a large set of high-level primitives for interacting with sensors and effectors. We have built additional layers on top of Pyro so that we can use a powerful KR system, capable of reasoning and communicating about its actions as it performs them, to enhance the capabilities of Pyro agents. While our interest and investigations have to date involve using the KR system to control the robot’s actions and plans, the KR component of the system could just as easily serve as an auxiliary, powerful reasoner for a Pyro agent to call upon.