HKBU  |  SCI  |  BUniPort  |  Library  |  Alumni  |  Job Vacancies  |  Intranet  |  Sitemap        
Undergraduate Admissions
Taught Postgraduate Admissions
Research Postgraduate Admissions
Job Vacancies
News & Achievements
Research Highlights
Contact & Direction
International Exchange and Internship Programmes

Department of Computer Science Seminar
2008 Series

AgentMT(TR) – a Multi-threaded Architecture Using Teleo-Reactive Plans

Prof. Keith Clark
Department of Computing
Imperial College

Date: October 10, 2008 (Friday)
Time: 1:30 - 2:30 pm
Venue: OEE601, Oen Hall (East Wing), Ho Sin Hang Campus


In this talk we will argue that a multi-threaded control architecture with a library of partial plans that are a generalization of Nilsson's Teleo-Reactive (TR) rule procedures allows for the smooth integration of three key levels of robot control:


1:   Speedy but goal directed response to changing sensor readings

2:   Switching between level 1 control procedures when higher level inferred beliefs about the state of the environment change

3:   Reacting to events and goals by selecting appropriate level 2 control plans

We start by introducing and demonstrating the use of TR rule procedures with a simulation and visualisation of purely reactive can collecting robot.


We will then show how a mutually recursive set of TR procedures, with rules that test beliefs inferred from sensor reading using concept defining rules, can be used to program a block stacking robot arm that can systematically build a tower of named blocks and ‘knows’ when the tower is completed.  A simulation will be demonstrated.


A key feature of TR procedure control is that the robot can be helped or hindered in its task and the TR procedure will immediately respond by skipping actions, if helped, or by redoing actions if hindered.  This operational semantics leads naturally to a multi-threaded implementation, with one thread for each active procedure.


A multi-tasking robot can respond to events: new goal events or just significant belief updates triggered by sensor readings.  It then selects an appropriate plan of action for each event using   
event/plan selection rules.


We conclude by describing our top level control architecture which borrows from classic BDI architectures, particularly AgentSpeak(L). It has a separate event handling thread that selects and invokes appropriate plans for events, sometimes suspending currently executing but incompatible plans for lower priority events.  The plans are a generalization of TR procedures.

Keith Clark joined the Department of Computing, Imperial College, in 1979 and was appointed Professor of Computational Logic in 1987. From 1983-88 he was an EPSRC Senior Research Fellow and was active in the Japanese Fifth Generation and the related UK Alvey, developing new declarative parallel programming paradigms with colleagues and students at Imperial. Since 1990 his interests have moved to distributed symbolic programming, still with a declarative emphasis, and its use for multi-agent systems and cognitive robotics applications. The output of this twenty-year research programme has been to establish the principles and practical implementations of a series of high-level symbolic programming languages, including Parlog, April and Go!, and a multi-threaded Prolog system, Qu-Prolog. April and Go! were developed in collaboration with Fujitsu's research laboratory in California, Qu-Prolog in collaboration with the University of Queensland, where he is a visiting Professor. Selected papers can be down-loaded from the web page:

********* ALL INTERESTED ARE WELCOME ***********
(For enquiry, please contact Computer Science Department at 3411 2385)
Copyright © 2017. All rights reserved.Privacy Policy
Department of Computer Science, Hong Kong Baptist University
Hong Kong Baptist University