4.6 2D Simulator Results
All of the different systems for this version of the insect simulator have now been discussed and it is time to put them to work and see what happens. Throughout these pages simple videos have been given to demonstrate specific items relevant to the current system under discussion. This section will display complete and long term runs of the simulator system. Each video will show the insect over an extended period of time as it interacts with its environment and tries to stay alive. Each of the videos changes something. Most of the runs change the properties of the food, and the metabolism of the insect. I played with the values to try and get an insect that could survive until it ate all of the food and starved to death. These videos are much larger than the previous ones. They are speeded up a lot also, and show the insect over a period of several hours of real time.
All of the blue lines in the figures trace where the insect went from the beginning of the simulation until it died. The table gives some of the important values that were used for this simulation.
So if the bite size is 10, and EU Per Piece is 10, then the insect will get 100 energy units every time it takes bite of food. And if there are 300 pieces of food in a pile, and each piece has an EU of 10, then the total pile of food has 3000 energy units. The metabolism of the insect is controlled by the energy usage variables. The higher these values, the more energy the insect burns and the higher the metabolism. An insect with a higher metabolism must eat more often if it is to stay alive. Most of the examples here have these values changed a little to see how it affects how long the insect can remain alive. The goal was to have the insect stay alive until all of the food had been eaten, and then die of starvation.
This one decreased the total amount of food energy available to the insect so that the video would not be so long and large.
This example decreased the metabolism so that it could go further between feedings and decreased the amount of food per pile. One thing to notice is that the food piles are so small that if the insect is really hungry then just eating one pile does not fill it up.
This example decreased the metabolism much more than the last example, and it bumped the amount of food in a pile back to the previous level. This insect finally reaches the goal of surviving until it eats all of the food. It only dies because it has run out of food.
This example uses an environment with a different set of obstacles and food distribution.
2. Future Experiments
There are a number of other experiments that would be very interesting to run using this system. The following is a list of three of the ones I think would be most interesting. At some point I hope to do some work on this.
These examples show a simple system that demonstrates complex and remarkable behavior. When scientists first started attempting to make robots that could interact with the world they built massive computational systems that sometimes took hours to move a few inches. Not necessarily because of any slowness problems with hardware, but just because the software that constituted their brains was so cumbersome. They tried to implement a top down, functionalist approach that attempted to form frames of the environment so the robot could navigate its way around the obstacles. But that is NOT how things work in the real world. Simple little insects just take the world as it comes, adjust to the unexpected, and move on. They have systems which at first glance appear very simple, but that can interact between themselves and the environment in highly dynamic ways that allow it to quickly adapt to changing situations and unpredictable landscapes. Understanding how insects can perform these miraculous feats with so little brain power is the first step towards being able to build flexible robotic systems that can be used for a wide variety of goals.
I really appreciate your interest!