Homework 5 - Problem 1

1. The goal of this exercise is to become familiar with the BraitenBug simulator (bbsim) that will be used in this homework set. To do this exercise, you will need to place the following three m-files in your working directory: bbsim.m, data_scope.m, brain_avoid.m. (a) After downloading the files into your local directory, start the BraitenBug simulation by typing "bbsim" at the matlab command prompt. The functions of the various buttons and dialog boxes will be explained in lab.  The default controller is a two neuron implementation of Braitenberg's vehicle 2a: After getting familiar with the interface, reset the simulation, then run the brain_avoid controller for approximately 30 seconds (simulated time) and hit pause. Answer the following questions: What was the elapsed time? __________ How many times did the bug run into the light? __________ What was the mean firing rate of  neuron 1 over the last 3 seconds of the simulation? ________ (Neuron 1 is channel 2 on the data_scope display) What was the mean firing rate of  neuron 2 over the last 3 seconds of the simulation? ________ (Neuron 2 is channel 4 on the data_scope display) (b) Create a copy of the brain_avoid.m file and call the new file "brain_approach.m" Edit brain_approach.m and change the connections so that neuron 1 now drives the right motor and neuron 2 drives the left motor. This will be an implementation of Braitenberg's vehicle 2b: Save the file. On the bbsim interface, enter "brain_approach" in the "Brain:" dialog box.  Test the new controller to make sure it works. Now, reset the simulation, then run the brain_approach controller for approximately 30 seconds (simulated time) and hit pause. Answer the following questions: What was the elapsed time? __________ How many times did the bug run into the light? __________ What was the mean firing rate of  neuron 1 over the last 3 seconds of the simulation? ________ What was the mean firing rate of  neuron 2 over the last 3 seconds of the simulation? ________ What is the average time between "captures" (collisions) ? ________

 

EXTRA CREDIT: Modify the parameters of the neural controller to increase the capture rate to over 20 captures in a 30 second trial. Do not alter the basic network architecture. You may change gains, thresholds, noise levels, steady-state current injection levels, etc. Once you have developed an improved controller, run the simulation for 30 +- 1 sec (simulated time) and check that the number of captures is greater than 20.  Turn in the following: a) a hardcopy of the simulation screen at the end of the 30 second trial (use the print button). b) a hardcopy of the data_scope display (from the command line: >>figure(2); print; ) c) a hardcopy of your neural controller code (e.g., brain_approach2.m) d) a brief explanation of why the changes that you implemented improved the performance.

 

HOMEWORK home page