Search the Mindcreators site: 

Update Alert!

I have now finished work on a much more advanced version of the insect simulator named AnimatLab. AnimatLab is a software tool that combines biomechanical simulation and biologically realistic neural networks. You can build the body of an animal, robot, or other machine and place it in a virtual 3-D world where the physics of its interaction with the environment are accurate and realistic. You can then design a nervous system that controls the behavior of the body in the environment. The software currently has support for simple firing rate neuron models and leaky integrate and fire spiking neural models. In addition, there a number of different synapse model types that can be used to connect the various neural models to produce your nervous system. On the biomechanics side there is support for a variety of different rigid body types, including custom meshes that can be made to match skeletal structures exactly. The biomechanics system also has hill-based muscle and muscle spindle models. These muscle models allow the nervous system to produce movements around joints. In addition, there are also motorized joints for those interested in controlling robots or other biomimetic machines. This allows the user to generate incredibly complicated artificial lifeforms that are based on real biological systems. Best of all AnimatLab is completely free and it includes free C++ source code!

4.5.4 Food Vulture

1. Introduction

Once I had reproduced all of the systems outlined in Dr. Beer's book, along with a few I had added on myself like the obstacle avoidance system, I began running long term tests. I wanted to see how long the insect could remain alive. I was surprised because the insect was not lasting as long as it should have. It had a problem if a food pile was to close to one of the walls it could get trapped in a loop around the food and would just end up circling it like a vulture until it finally died. This situation did not happen often, but it was frequent enough that it was usually the cause for the insects eventual death. If the insect was going to be able to survive for any length of time then some solution needed to be found to break the insect from this behavior. It should also be noted that I did find a brief reference to a similar behavior in Dr. Beer's book, but it did not go into any details of what he had found that caused it or what he did to fix it.

2. What Causes an Insect To Circle?

Caught In Circle
Video 1. This video shows the getting caught in a circle.

Video Size: 2.2 Mb

What causes the insect to get stuck in this death loop? The reason it happens is that the insect only has a certain turning radius. If the food is a smaller distance from the food than this radius then in certain situations the wall can change the trajectory of the insect so that it will always just miss the food on its next attempt and causes it to hit the wall in the same way forming a loop. Video 1 demonstrates this happening. The insect tries to turn to get to the food but just misses it. This would normally not be a problem because it would just keep going and get it on the next pass. However, the wall keeps it from having the necessary room to correctly line up with the food.

3. Food Vulture Neural System

Food Vulture Breakout Controller Neural Layout
Figure 1. This is the neural layout that is responsible for breaking the insect out of the vulture circle.

So how can the insect get itself out of this infinite loop? The neural system shown in Figure 1 is what does it. When the insect gets caught in one of these loops then it will be turning to one side for a long period of time. In video 1 the insect gets caught in a right handed loop. So ROS fires continuously trying to get it lined up with the food and LOS is silent. Since it can never turn sharply enough to line up with the food this never changes. Under normal conditions the insect would turn enough that it would be facing the food and neither LOS or ROS would be firing. So the key is to check to see if LOS or ROS is firing for an extended period of time while the other neuron is silent, and to make sure that it this happens while the insect is actively following an odor. If LOS is firing and ROS is silent then the Following Circle Left (FCl) neuron is stimulated. This neuron has a large capacitance and a high threshold voltage. This ensures that LOS must be firing for a long period of time while the Following Odor Motor Command (FOMC) neuron is also firing. When FCl does fire it only produces a brief spike that strongly stimulates the Following Circle Memory (FCM) neuron. When this neuron is firing it inhibits the FOMC neuron and makes the insect quit trying to follow the scent to food. FCM also stimulates the Following Circle Reset Memory (FCRM) neuron. This neuron is responsible for resetting the FCl and FCr neuron so that it will again take a prolonged firing period from either LOS or ROS in order to trigger them to fire. The basic effect of this is that when the insect gets into a circle it will go around the food three or four times and then one of the FC neurons will fire and cause it to briefly stop trying to get to the food. It will either begin to wander off or start following the wall or something. After a short period though it will again start to try and get to the food. But since it is now in a different position and orientation the odds of it getting stuck in a loop are drastically lower. Basically all this system does is to knock it out of the loop and then let it try again.

4. Food Vulture System Analysis

Escape From Circle
Video 2. This video shows the insect escaping from a circle.

Video Size: 2.2 Mb

Escape From Circle Analysis
Video 3. This video shows the analysis of the insect escaping from a circle.

Video Size: 2.2 Mb

The analysis in video 3 demonstrates how the insect gets out of its little predicament. Once it gets in the loop ROS goes high and stays there. During that time the membrane voltage of FCr is steadily increasing. (This is not shown in the graph.) When the voltage finally crosses the threshold it fires, and is then damped down again because of the inhibition from the FCRM neuron. This causes the output of FCr to be a brief pulse. However, that pulse is enough to stimulate the FCM neuron. As it discharges it inhibits the odor following ability of the insect allowing it to escape the loop, and it stimulates FCRM.

5. Overview

This new system eliminates the last major way in which the insect can kill itself in the current version. It is still possible for the insect to get stuck and die, but all of these systems working together make it much more unlikely now. The next section shows the insect over a number of long term runs using different starting parameters.


<< Previous Contents Home Next >>

MindCreators.Com is edited and maintained by David Cofer. If you have any questions, comments, or just want to discuss the contents of this website, then email me at: dcofer@MindCreators.com.

Copyright © 2002 by David Cofer. All rights reserved.