Abstract
This paper presents a planning system based on semantic reasoning for a general-purpose service robot, which is aimed at behaving more intelligently in domains that contain incomplete information, under-specified goals, and dynamic changes. First, Two kinds of data are generated by Natural Language Processing module from the speech: (i) action frames and their relationships; (ii) the modifier used to indicate some property or characteristic of a variable in the action frame. Next, the task’s goals are generated from these action frames and modifiers. These goals are represented as AI symbols, combining world state and domain knowledge, which are used to generate plans by an Answer Set Programming solver. Finally, the plan’s actions are executed one by one, and continuous sensing grounds useful information, which makes the robot use contingent knowledge to adapt to dynamic changes and faults. For each action in the plan, the planner gets its preconditions and effects from domain knowledge, so during the execution of the task, the environmental changes, especially those conflict with the actions, not only the action being performed but also the subsequent actions, can be detected and handled as early as possible. A series of case studies are used to evaluate the system and verify its ability to acquire knowledge through dialogue with users, solve problems with the acquired causal knowledge, and plan for complex tasks autonomously in the open world.
Highlights
Research on service robots has received increasing attention in recent years [1,2,3]
Speech is the primary communication method in General-Purpose Service Robot (GPSR) test defined by RoboCup@Home https://athome
This paper addresses these issues by developing a semantic task planning system, which combines natural language understanding, task-oriented knowledge acquisition, and semantic-based automated task planning
Summary
Research on service robots has received increasing attention in recent years [1,2,3]. There are two main challenges for task planning: (1) the robot’s perception of the world is often incomplete, a command may refer to an object that is not in its knowledge base, lack of information will fail to generate a plan; (2) changes in the dynamic environment may not be expected by the robot, which will cause the planned action to fail. Our method features: (1) a method of confirming task type, extracting the roles of the task and the roles’ constrained information; (2) assumption and grounding methodology to “close” the open-world, it is only introduced when there are not enough instances, which can benefit from the use of existing knowledge; (3) continuous sensing and conflict detection mechanism for all actions not performed that captures dynamical changes in the environments and triggers special processing as soon as possible.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
