This project
aims to investigate a bio-inspired approach to developing computer systems.
By studying social insect based models we hope to gain a new understanding
of complex systems and how the seemingly intelligent behaviors of the
assemblage arise from the behavior of the individual agents, leading
to innovative solutions to complex tasks. Here we present the 3D simulation
of the emergent behaviour of Macrotermitinae termites. Similar processes
can be seen in many other complex systems. For example: a termite colony
can be contrasted with cells in an organism, communicating via various
quantitative communication media: morphogens, electric signals and hormone
gradients. This use of reaction-diffusion, stochasticity, dynamic self-organisation
and adaptive behaviour are all recurring aspects of complex systems
that self-organise to form global patterns in a given environment.
The top-down
approach models the global behavior directly without considering the
local interaction that contributes to this global behavior. For example,
some use Partial Differential Equations (PDEs) to model termites. Equations
are easy to simulate and evaluate. However, they suffer from many disadvantages.
They are difficult to describe as the model become more complex. They
are difficult to modify to incorporate more complex interactions. These
models are also often deterministic - the specified parameters are used
each time the simulation is run, so the same results will occur. A better
approach is to model a topology within which interaction may occur and
at each location in that topology how the contents may change based
on its current state, neighboring states and possibly global effects.
Agent-based
simulation model individuals within their topology and their behavior,
e.g., how they interact with their model. The key aspect of the simulation
is that the larger simulation has been broken into a number of smaller
components. This often makes it easy to modify the simulation in order
to examine different scenarios.
A representative Macrotermes mound can
be 3 to 4 meters high and is essentially made from selected, graded
and bonded soil pellets. Inside, the mound is dominated by passages
and ducts, which surround the nest and perform air-conditioning functions.
Within these fully enclosed and protective architectures, survival of
such large populations of termites is dominated by the requirement to
stabilise the mound’s temperature, moisture levels and respiratory gas
balance. Solutions to the problem are impressive and are analogous to
the function of the animal lung and thermo-regulatory systems which,
in the case of Macrotermes mounds, is driven by the activity
of the termites within. In fact, in physiological terms, the termites
have evolved to outsource many of the homeostatic functions (including
the digestive function which is performed within fungus gardens surrounding
the nest) into the mound structure itself. The ‘outsourcing’ of biological
function results in a net energy gain within the colony, which drives
the activity of the termites to maintain and modify the dynamic adaptations
exhibited by the mound structure which is referred to as social homeostasis.
Of key importance to this research is how Macrotermes produce
such complex architectures, from their combined actions, without any
obvious global control mechanism. The system operates on a remarkable
‘indirect’ communication mechanism where workers respond and adapt to
constantly changing internal conditions and external influences to maintain
equilibrium within the mound. The study of emergent behaviours, displayed
by social insect, is termed Swarm Intelligence.
|