A Hybrid Deliberative Layer for Robotic Agents

Fusing DL Reasoning with HTN Planning in Autonomous Robots

verfasst von: Ronny Hartanto

Verlag: Springer Berlin Heidelberg

Buchreihe : Lecture Notes in Computer Science

Enthalten in: Springer Professional "Wirtschaft+Technik" , Springer Professional "Technik" , Springer Professional "Wirtschaft"

Einloggen, um Zugang zu erhalten

Über dieses Buch

The Hybrid Deliberative Layer (HDL) solves the problem that an intelligent agent faces in dealing with a large amount of information which may or may not be useful in generating a plan to achieve a goal. The information, that an agent may need, is acquired and stored in the DL model. Thus, the HDL is used as the main knowledge base system for the agent. In this work, a novel approach which amalgamates Description Logic (DL) reasoning with Hierarchical Task Network (HTN) planning is introduced. An analysis of the performance of the approach has been conducted and the results show that this approach yields significantly smaller planning problem descriptions than those generated by current representations in HTN planning.

Inhaltsverzeichnis

Frontmatter

Introduction

Abstract

In recent years, the robotics field has seen rapid advancements in both software and hardware areas. Good algorithms for perception, learning, planning, navigation and inferencing have been researched and used in autonomous robots. They enable the robot to be more useful in daily life, for example as a courier, or a museum guide, such as Rhino [1,2], Minerva [3], and Robox [4,5]. Several new robot platforms have been developed which support major research in robotics, from advanced mobile robots, such as legged robots, to lightweight manipulators.

Ronny Hartanto

The Hybrid Deliberative Layer

Abstract

In this chapter, a brief introduction to robot control architectures is presented. It highlights the need for a deliberative layer in robotics and its role as a planning system of sorts. In large domains, computation time is in danger of exploding, as the size of the domain grows. A novel approach that amalgamates Hierarchical Task Network (HTN) planning with Description Logic (DL) reasoning is presented to keep the planning domain size within limits. With this approach the planning domain is modelled using the DL representation instead of being modelled directly in the planning representation. Finally, algorithms to automatically generate planning problems from the DL system are shown.

Ronny Hartanto

HDL Systems in the Robotics Domain

Abstract

In the previous chapter, the terminological concept of HTN planning was defined. In order to generate a valid HTN planning problem for SHOP2, a terminological concept that models the environment needs to be defined. In addition, the HTN planning operators, actors and objects must be instantiated and inserted into the corresponding ABox. In this chapter, a generic method for modelling the environment in HDL and filling the HDL system is introduced. Two implementations from the robotics domain are also presented, namely the robot navigation domain and the pick-and-place domain.

Ronny Hartanto

Case Study: “Johnny Jackanapes”

Abstract

In the previous chapter, two robotics domains were implemented using the HDL system. This chapter presents a case study that uses the HDL system in a mobile robot. This example demonstrates how the HDL system may be integrated within an existing robotic system.

Ronny Hartanto

HDL Systems in the AI Domain

Abstract

The HDL system has been developed for solving planning problems in robotics. Chapter 3 presented its application in solving problems in two robotics domains, namely those of navigation and pick-and-place. As the HDL extends the HTN planner, it can be used for solving any planning problem that can be solved by an HTN planner. In this chapter, HDL is used for solving the well-known Blocks-world problem in the AI domain.

Ronny Hartanto

Results and Evaluation

Abstract

In the previous chapters, a number of planning problems in robotics and AI are solved using the hdl system. They show the benefits of using the system. In this chapter, the system is benchmarked in comparison to the htn planner. This is done for both the robotics domain and the AI domain. The main question is that of the hdl system’s complexity hdl system’s complexity hdl system’s complexity. Firstly, a look into how the hdl system processes the user requests and what the activities are that produce the output. Secondly, the experiment setup is introduced. Finally, the empirical experiments are discussed.

Ronny Hartanto

Discussion

Abstract

In this chapter, the HDL system is discussed. A number of open questions and possible improvements of some methods that have been presented in the previous chapters will be addressed.

Ronny Hartanto

Conclusions

Abstract

In this work, a novel approach for integrating dl reasoning and htn planning is presented. The Hybrid Deliberative Layer (hdl) solves the problem that an intelligent agent faces in dealing with a large amount of information which may or may not be useful in generating a plan to achieve a goal. The information, that an agent may need, is acquired and stored in the dl model. Thus, the hdl is used as the main knowledge base system for the agent.

Ronny Hartanto

Backmatter

Titel: A Hybrid Deliberative Layer for Robotic Agents
verfasst von: Ronny Hartanto
Verlag: Springer Berlin Heidelberg
Electronic ISBN: 978-3-642-22580-2
Print ISBN: 978-3-642-22579-6
DOI: https://doi.org/10.1007/978-3-642-22580-2