Improving the Efficiency of Robot Task Planning by Automatically Integrating Its Planner and Common-Sense Knowledge Base

Abstract

This chapter presents a newly developed approach for intelligently generating symbolic plans for mobile robots acting in domestic environments, such as offices and houses. The significance of this approach lies in its novel framework which consists of new modelling of high-level robot actions and their integration with common-sense knowledge in order to support robotic task planner. This framework will enable direct interactions between the task planner and the semantic knowledge base. By using common-sense domain knowledge, the task planner will take into consideration the properties and relations of objects and places in its environment, before creating semantically related actions that will represent a plan. A new module has been appended to the framework which is called Semantic Realization and Refreshment Module (SRRM). This module has the ability to discover and select entities in the robot’s world (entities related to robot plan) which are semantically equivalent or have a degree of similarity (where they don’t exceed a predefined threshold) by using techniques and standards (metrics) for similarities. SRRM supports robotic task planning to generate approximate plans to solve its tasks when there is no exact plan can be generated according to initial and goal state by extending initial state and action details with similar or equivalent objects. The extended framework enables direct interactions between task planner, Semantic Action Models (SAMs) and knowledge-base through creating planning domain (or extended planning domain) with predicates (or semantically equivalent or similar predicates) which specify domain features. The proposed framework and approach are tested on some scenarios that cover most aspects of robot planning system.