# Engineering

We use collisions to detect whether an obstacle lies in front of the generalized great wonderer (GGW) but in a practical scenario that method is less than feasible depending on the physical properties of the colliding objects which may damage the great wanderer.

However, we abstract away from collisions and replace it by the notion of detection, which can be done in several ways: for example by using heat sensors with a certain sensitivity (perhaps calibrated to the design parameters) on all the faces of the GGW. Another technical problem is propulsion. In the Second Life environment, propulsion is achieved by simply requesting it from the simulator. In a practical scenario, some sort of engine is required in order to accelerate the GGW to some velocity. The question is, how could one achieve that because all the platings / faces have sensors which must detect collisions.

One way to do achieve propulsion would be to design the platings so that they contain holes, similar to a sieve which would allow particles to flow from the GGW and propulse the object in the opposite direction.

Perhaps something like this:

``` +---------------------------+ <---- sensor
|                           |
|   o o o o o o o o o o o   |
|                           |
| o o o o o o o o o o o o o |
|                           |
| o o o o o o o o o o o o o |
|                           |
| o o o o o o o o o o o o o-------
|                           |    | l
|   o o o o o o o o o o o---------
|                       | | |
+-----------------------|-|-+ <----- sensor
^                       |-|
| sensor                 l```

It is important that the distribution of the holes is symmetrical relative to the axes of the plating in order to make sure that there is no stray from the direction of movement. Another possible problem is that the sensors would have to be placed on the plate as well which, in case they are heat-sensors, they would be easily impressed by the heat dispersed through the sieve. One could perhaps use a photo-electric set of sensors that are calibrated to only be impressed over some threshold that exceeds the output of the engines.

One could perhaps add another layer of detection by using the surface points given by the plating and place the sensors in the vertices, with a ruleset that would only consider that an object is in the vicinity of the plating. The most trivial could be a an inclusion-condition that says that all 4 sensors have to be impressed over some threshold in order for a detection could be sensed.

This could also provide a new spray of the probability set mentioned earlier and would be great for detecting (and correcting) minor aberrations from the flight path practically adding a feedback element to the system.

We would wager that the efficiency of the GGW is surprisingly good, given that the geometric body is a reduction of a sphere which provides a great volume compared to the surface of the plates responsible for the ejection of fuel and would allow instruments to be placed inside the GGW.

Again, if possible, regardless of the instrumentation placed inside the GGW, the center of gravity should be as close as possible to the centre of the sphere in order to maintain the overall symmetry. The symmetry is implicitly responsible for the efficiency of the system since by placing the centre of gravity in the centre of the sphere, the same amount of effort will be required in order to propulse the GGW for all directions.

It is perhaps interesting to note that given a large amount of plates, when the GGW converges to a sphere, the amount of effort required per plate in order to perform a course correction is uniformly distributed.

Not only that, but less fuel could be used by performing adaptive course-corrections in order to avoid an object.

For a cube (a 6-GGW), one can only avoid an object performing sets of two movements along perpendicular axes: (up, forward), (left, backward), etc... However, by having a large automaton, the movements could be reduced by following arcs instead of rectilinear pathways. Here is an example:

```
y ^                   Movement
|                     for
|               2-axes, 4 directions
-x     |
- - - G------->             +---+
|      x           G  |   |
|  +---+
|-y                +----------->

Movement
for
y ^                2-axes, 8 directions
b(x,y)+\   |  /b(x,y)+
\  | /                    _
-x   \ |/                      /\
- - - G------->              /
/ | \    x              / +---+
/     \                 G  |   |
b(x,y)-/   |-y \b(x,-y)-           +---+
```

The figure above shows that on a Cartesian system with 2-axes and 4 possible directions, the only way to avoid the object (except for returning back on the same path), is to perform two course corrections: once going downward and then the other to the right. In case of 2-axes and 8 possible travel directions, given by the bisectrices of the system plane, only one course correction is needed: one burst to push the object on a path that would avoid a collision with the object.

Interestingly, by increasing the automaton to span many states for different directions, you can use less fuel to prevent collisions with objects. In some way (meh), you are converting the effort that the engines have to make to a computational effort in order to process a larger automaton: you are converting physical work to computational work (given a generalized notion of energy).