CS 495  Artificial Intelligence   Fall 1999,  Instructor:  Jeffrey Horn

HOMEWORK 3:  Stimulus-Response Brains (Digital Braitenberg Vehicles
 

Handed out/Assigned:	Fri. Nov. 19, 1999
Due:	Wednesday, Dec. 1, 1999

FROM THE TEXT:  Chapter 2, early parts, for background.

(1)  DIGITAL Braitenberg Vehicle 1:
 
 

  Assume two inputs and two outputs.  Input is from an IR proximity detector, similar to ours but different.  Assume the following detector geometry:

 
So a "01" input vector would mean something detected to the right, but nothing center or left, while "11" would mean something in the center (or something left and right!).

We have two binary outputs, one for each motor.  LM is left motor, RM is right.  A "0" output means no power to the motor, while a "1" means go forward.  (so an output of 01, which is 0 to the left and 1 to the right motor, would result in an arcing turn to the left, while 10 would be a right turn)
 

Here is an example truth table (ie., a "brain") for a strategy to move toward obstacles (hopefully!):

 

INPUTS			OUTPUTS
LS	RS		LM	RM
0	0		1	0
0	1		1	0
1	0		0	1
1	1		1	1


(1A)  Fill in the following blank truth table for a strategy to move AWAY from obstacles:

 

INPUTS			OUTPUTS
LS	RS		LM	RM
0	0
0	1
1	0
1	1

(1B)  Fill in the following blank truth table for the "Wall-hugger" strategy.  This strategy should cause the robot to "look" for an obstacle, then try to keep the obstacle on its right (or left, your choice, but stick with it and be consistent).  If this obstacle is a wall, the robot should follow it.  If it is a "point" obstacle, or otherwise small, the robot should go into "orbit" around it.  Think about it!
 

INPUTS			OUTPUTS
LS	RS		LM	RM
0	0
0	1
1	0
1	1


VEHICLE 2:

Now let's add some resolution to our sensors and effectors.  Let there be THREE detectors, arrayed as follows:




Also, let there be two binary outputs to the two motors, allowing us four different speeds;  reverse (00), stop (01), slow forward (10) and fast forward (11).    Now redo the above strategies for Vehicle 2:
 

(2A)  Fill in a blank truth table for a strategy to move AWAY from obstacles

INPUTS			OUTPUTS
LS	CS	RS	LM  1st bit	LM 2nd bit	RM 1st bit	RM 2nd bit
0	0	0
0	0	1
0	1	0
0	1	1
1	0	0
1	0	1
1	1	0
1	1	1

(2B)  Fill in a blank truth table for the "Wall-hugger" strategy.

VEHICLE 3:

Design your own!  Take vehicle 1 or vehicle 2 (v1 is twice as simple as v2, but also has only half the sensory information!), and add AT LEAST ONE NEW sensor, and/or AT LEAST ONE new effector.  These can be just one bit each (note that for every bit of additional sensor data you double the number of rows in the lookup table!  this is NOT true for the output (effector) bits however).  Be inventive.  And explain fully what the sensor gives you, and/or what your effector does, and what the strategy is (yes, you have to give an example lookup table using your new capablities).   ALSO:  DRAW an example of your robot's behavior using the sensors and effectors and your strategy.   This should be a 2-D sketch of the robot, obstacles, walls, light sources, other robots, WHATEVER is appropriate, and trace a path of how you would expect it to move, turn, fire, or whatever, over time in that world.  This will help me understand what you have in mind!

Acceptable examples include:

1. A one-bit sensor that tells you if the object is friend or foe (or "predator or prey"), plus a weapon that fires forward (one bit:  "1" means "fire"), and then a strategy to pursue prey, flee from foes, or pursue and fire on foes, etc.
2. A one (or more) bit state variable that functions as memory.  This could be both input and output:  as output, one sets the state variable (to "0" or "1") depending on the current state, and then the state variable acts as input in the next state (i.e., the current state of the state variable is in the input columns in the truth table.   (How could this help you get out of "infinitely looping behavior"?)
3. A two-bit sensor telling you which of the four cardinal directions (N,E,S,W) your are most closely facing, and then a strategy to always (try) to go NE!
4. And so on!