Notes

PROFESSOR: In this video, we will
look at the computational approach to system thinking.
Now remember that system thinking is problem
solving in complex settings.
A system is an entity with interrelated and interdependent
parts.
The system is defined by its modules,
by the boundaries between the modules,
and by their interaction.
And the system is more than the sum
of its parts or its subsystems.
Changing one part of the system affects other parts,
and generally, the whole system.
The global behavior of the system
depends on local functions and interactions.
And for this reason, we can think of systems
as being hierarchically composed of subsystems,
with increasingly simple function
as we go towards the bottom of the hierarchy.
Let's start with an example.
Imagine a world where anybody can employ systems thinking
to reason conceptually about a complex system,
a complex device that they want to create,
and they get automatic support to realize that idea.
So here's our user.
Let's call her Alice.
And let's say Alice would like to invent
a device that will entertain her cat while she's at work.
Now to do so, Alice can head to a new kind
of store called 24 Hour Robot Manufacturing that
embodies the principle of system thinking.
And in this store, Alice can explore the design space
using a very intuitive interface.
And when she settles on a design,
the store will print the device immediately for not much cost,
and the cat will have a playmate.
So how crazy is this idea?

Can we imagine creating a computational infrastructure that allows us to think about complex concepts and employ computation to do the co-design of all the components of those complex systems?
We're going to look at exactly how to do this.

Suppose I want a robot to play chess with me.
Is it possible to input this desire, a robot to play chess, and from this very high level specification
to extract the exact device, the exact construction that meets my specification?
How would this computational process look like?

The actual system specification, I want a robot to play chess with me, is quite conceptual.
But we can employ computational tools to parse these specifications to determine
the behaviors required by a chess playing robot.

For example, picking up a piece, dropping the piece,
moving a piece from here to there
without bumping into other pieces.

Once we know these behaviors, we can identify actual mechanisms
that realize those behaviors.

Then we can assemble them all together,
and we have this chess playing robot designed
by a non-expert roboticist in about two hours
using this intuitive approach of designing machines
from high level specifications using systems thinking, using
modularity, hierarchy, and compositions.

In order to implement this approach to computational system thinking,
we need several modules that address different aspects of the process.
First of all, we need a database of existing designs.
And so we have to address what this database would look like,
what should the representations in the database be.
Then we need to identify a segmentation process and a search process that allows us
to choose the correct modules from the database.
And so this addresses the question, what modules should be composed, and how should they
be composed?

Once we have identified the modules, we need to know exactly how do we assemble them together.
Once we have an assembled system, what do we have to do in order to simulate and verify
that the system meets the specification?

Finally, after the overall system is shown to meet the specification,
it can be fabricated using rapid prototyping.
We are going to look at how exactly we can employ these concepts to create custom designs that
achieve complex functions.