FAQ Hot Research Topics -
Personal Interests
MSc or 3rd year U/G Projects
Here a strongly recommended strategy is to take some published piece of
work, a paper from a journal or conference in an area that interests
you, and recreate that same work in your own words and your own
programming code. This always has the
potential to be interesting, challenging, and lead to original
extensions and variations of the work.
I can supervise such projects in any areas of Artificial Life,
including Evolutionary Robotics, Genetic Algorithms, Cellular Automata,
Game Theoretic modelling, GAs + ANNs for data-mining, analysis of
CTRNNs. I will be especially interested in projects close to my own Hot
Topics.
My Hot Topics -- Doctoral Research
areas of interest
I am personally interested in brand new research areas where there are
only around 10 people in the world currently interested, yet it looks
like having a big future impact. If there is only 1 person interested,
you are a lone madman; if there are 100+, it is already a bandwagon and
I am less interested. These are areas that I promote in my Artificial
Life course, I try and focus my own research on, and that I want to
have doctoral students working in. I will list them roughly in personal
historical order.
Evolutionary Robotics
ER had only a handful of people when I started in the area in the late
1980s (I claim to be the first person employed as an Evolutionary
Roboticist), but has now become mainstream and a bandwagon. So I am
only interested now in radical new angles on ER. Eg ER approaches to
studying developmental issues (cf work by Rachel Wood). Or ER
approaches to Passive Dynamics Plus (see below, and work by Eric
Vaughan).
Artificial Evolution
My doctoral thesis was on this. For current interests see Neutral
Networks, Microbial GA, and the applications listed here. SAGA (Species
Adaptation Genetic Algorithms) will be covered on another FAQ; Neutral Networks are the hot topic
derived from this. cf also the Microbial Genetic Algorithm.
Neutral Networks
A significant part of my thesis was on this area -- eg SAGA focussed on the importance of neutral
ridges for evolution of a genetically converged population (but the
term "Neutral Networks" only came later). Lionel Barnett in his
doctoral work with me made very important advances. Adrian Thompson
demonstrated the existence of these in Hardware evolution, and we
collaborated on a paper. There is still an enormous amount
of important research to be done, and very few people in the area. We
are currently putting together a grant proposal for a research effort
here.
Minimal Cognition
The more philosophical /cognitive science end of ER involves studies in
minimal cognition; Braitenberg vehicles fall into this paradigm, and we
have rather similar motivations here to those of Randy Beer (Case
Western, now moving to Indiana). What is the simplest model of an
artificial agent that can learn, that can discriminate between two
examples, or over a continuum of examples? (cf work by Eduardo
Izquierdo-Torres, previous work by Elio Tuci with me). Lots of room for
more cutting-edge research. One project I am interested in is getting a
minimal cognition model of an agent that uses representations
(...'external' representations, of course, since 'internal
representations' are just the product of confused people). CTRNNs (as
with Beer) are a common tool used here. Another project is promoting
the understanding (at a fundamental, philosophical level) of perception through
minimal cognition models.
Philosophical Issues
I call some of what we do "Philosophy of Mind with a screwdriver".
Interests in a Dynamical Systems understanding of Cognition, in
Autopoesis.
Passive Dynamics Plus
Passive Dynamic Walking allows bipedal (potentially) multipedal)
walking down a gradient with no motors, no control system, a graceful
gait and near-maximum efficiency. Humanoid bipedal walking should start
here, adding control and power. Eric Vaughan has made major advances in
this here, and we are shortly applying for a grant for a major research
effort in this area.
Homeostasis
There is lots of potential hot research to do so as to understand
homeostasis further. Ezequiel di Paolo is also much interested in this.
One route through minimal cognition experiments, another through
Daisyworld theory.
Daisyworld
A potential mechanism for explaining Gaia theory, I see this as
an important model for homeostasis in any organism. I finally
understood this properly through producing a much simpler model (Alife
9 paper) analysing it in terms of Rein Control. Scope for much
more work here, including the relationship with the MEP (Maximum
Entropy Production) Principle, very recently put on a firmer footing by
Roderick Dewar. James Dyke is working with me in this area.
Complexity and Stability of Ecosystems
The assumptions built into Robert May's models must be wrong, or
inadequate, or missing out further constraints. I have various
proposals in this area.
Autonomous Glider
As a practical robotics project, using minimal perception, it seems
feasible to put a minimal control system into a model glider for 3-D
Braitenberg-like behaviour; for ridge soaring, tracking up and down the
windward side of a ridge of hills gaining lift. I have the basics
worked out, we have a powered glider (radio control plus power for
take-off, then hand over to autonomous control), we need the time and
effort put into this. As well as a good technical challenge, this has
the tremendous appeal of a form of artificial life that could genuinely
extract the energy it requires to maintain itself from the
environment -- from the wind.
Adaptive Text Entry
Mobile phones, ipods, PDAs, need a better method of one-handed text
entry. Dasher (David Mackay) is one amongst many proposals that
incorporates prediction of likely sequences, adapting to the user, but
they all have usability problems. I have some proposals in this area,
and have supervised an MSc project by Matt Evans. Lots of scope for
more work, potential enormous commercial demand.
[More ....]
I really want to hear other peoples' original proposals in similar
areas, in the hope that I will find them interesting or important.
Return to Inman Harvey
FAQs
Return to Inman Harvey Home Page