Society for the Study of Artificial Intelligence and Simulation of Behaviour
AISB Quarterly - Spring 2007 - http://www.aisb.org.uk/


                             Austin Tate

                       Planning and Doing Things

           Artificial Intelligence Applications Institute
                      University of Edinburgh

I was interested in computers by the age of 15 and gave talks on them
at school. I attended evening classes a couple of years later while
still at school travelling on the bus for an hour in the evening to a
college in Leeds to learn programming (in COBOL!). Computers at that
time filled a room, you submitted your exercises on punched card and
got the results the following day. I built my first AI planner over 35
years ago. I'd already been on an early AI course at Lancaster
University where the language of choice for teaching a range of topics
was POP-2 and wanted to do a Summer project to create a problem
solver. With support from Donald Michie and his team at Edinburgh I
tried to create a Graph Traverser along the lines they were working
on. Boy, am I glad I got involved with Computers, AI and planning
technology!

Planning is a key area for creating intelligent behaviour and a long
term aim of the AI community. I have been doing my bit to advance the
concepts, technology and applications we have in his area. And, in
doing so, I have been able to bring in a number of my other interests
in search and rescue teams (from a childhood TV programme - Supercar),
space travel and future habitats. This has involved collaboration with
scientists, systems developers and creative people worldwide. I love
collaborative projects and the joint demonstrations of the results.

I joined Donald Michie's Department of Machine Intelligence in
Edinburgh in the early 1970s during which time my AI planners
Interplan and Nonlin were created. I shared a flat with my fellow PhD
student Dave Warren, who worked on the early Prolog compilers, using
remarkably similar technology to that I was interested in for
planning. I shared an office with Aaron Sloman who was visiting and
thinking deep thoughts about the philosophical aspects of AI. Earl
Sacerdoti, who did work on multi-level plan representation and
hierarchical planning was a visitor and we worked together on concepts
for flexible hierarchical, partial ordered task network planners with
internal goal structure recorded. Edinburgh is a fantastic environment
with many visitors and collaborators. Nonlin was created to be used on
real applications with the UK electricity generation utility for
turbine overhaul procedure project planning. It was also used to drive
the Freddy II robot in a project led by Robin Popplestone. Its core
algorithms are still at the heart of most Hierarchical Task Network
(HTN) and partial-ordered (POP) planners.

I had a period in the second half of the 1970s working on IBM and ICL
mainframes leading a team which developed commercial data base
software for engineering applications, and the team supported that for
business critical applications. Just afterwards I headed up Edinburgh
University's microcomputer support and office systems team, when these
systems were novel and distributed computing was just starting to take
off. Most computing before that was on large centralised mainframe
systems. This work on software engineering and being involved in
systems and user help desk support was something that has stayed with
me through my career.

Edinburgh University used some of the funds generated by its
commercial activities, in which I was engaged, to give a number of
academic entrepreneurs a fellowship to allow them to explore
collaborations with industry and with international organisations. The
three of us that got these fellowships explored AI (me), Computer
Science (Malcolm Atkinson, now Director of the National e-Science
Centre) and Electronic Engineering (Peter Denyer, who formed Vision
group, a company successfully floated on the Stock Exchange). I joined
Jim Howe (Head of the School of AI at Edinburgh) as he was creating
the Artificial Intelligence Applications Institute (AIAI) and we
obtained commercial and other funding to start this organisation. I
took over as its Director after the first year of operations. AIAI has
been a pioneer in using AI technology for a wide range of
applications. It has concentrated on knowledge systems, planning
systems, and adaptive systems. Its been involved in some pioneering
and deployed influential systems in everyday use - for yellow pages
layout, spacecraft assembly, integration and test, industrial plant
diagnosis, logistics support and so on.

My own Planning and Activity Management Group within AIAI is exploring
representations and reasoning mechanisms for inter-agent activity
support. The agents may be people or computer systems working in a
coordinated fashion. The group explores and develops generic
approaches by engaging in specific applied studies. Applications
include crisis action planning, command and control, space systems,
manufacturing, logistics, construction, procedural assistance, help
desks, emergency response, etc. Our long term aim is the creation and
use of task-centric virtual organisations involving people, government
and non-governmental organisations, automated systems, grid and web
services working alongside intelligent robotic, vehicle, building and
environmental systems to respond to very dynamic events on scales from
local to global.

Over the last decade, I have concentrated on applying planning and
collaboration technology to emergency response tasks, and a number of
my projects contribute to this. In the UK, the Advanced Knowledge
Technologies (AKT) project with 5 university groups and a number of
industrial and government has pushed the research agenda for the
semantic web and knowledge systems on the web and we are applying
these in a challenging accident scenario situated in central
London. In the US, we have had DARPA (Defense Advanced Research
Projects Agency) support for our work since the late 1980s) and this
has been directed towards multi-national coalition operations for
peace-keeping, collaborative operations for disaster relief, search
and rescue coordination, etc. in projects such as CoAX, CoSAR-TS and
Co-OPR. In Europe, we are engaged in projects such as OpenKnowledge
which has emergency response interests in dealing with the aftermath
of floods in Italy. The FireGrid project, in the UK, involves a large
team from academia, industry and government agencies who are exploring
concepts for emergency response with advance simulation in intelligent
buildings.

Our aim is to create helpful agents and collaboration between
organisations and people which can use knowledge of other agents,
tasks, procedures, services and the environment so as create a
"helpful environment" which improves safety and the lives of everyone.

References

AIAI Planning and Activity Management Group - http://www.aiai.ed.ac.uk/project/plan/

Tate, A. (2006) The Helpful Environment: Geographically Dispersed
Intelligent Agents That Collaborate, Special Issue on "The Future of
AI", IEEE Intelligent Systems, May-June 2006, Vol. 27, No. 3, pp
57-61. IEEE Computer Society.

See Austin Tate being interviewed about the use of artificial
intelligence techniques and their use in emergency response centres in
March 2007 - http://www.ukfuturetv.com

(C) 2007, Austin Tate, AIAI, University of Edinburgh