Lab 7: Subsumption architecture
Due April 16

Genghis: One of the first robots controlled by subsumption architecture.

Starting point code

1. First you need to run setup63 to create a git repository for the lab. If you want to work alone do:

   setup63 labs/07 none

   If you want to work with a partner, then one of you needs to run the following while the other one waits until it finishes.

   setup63 labs/07 partnerUsername

   Once the script finishes, the other partner should run it on their account.

2. For the next step only one partner should copy over the starting point code.

   cd ~/cs63/labs/07
   cp -r ~meeden/public/cs63/labs/07/* ./
   

   This will copy over the starting point files for this lab.

3. Whether you are working alone or with a partner, you should now add all of the files to your git repo, commit, and push them as shown below.

   git add *
   git commit -m "lab7 start"
   git push
   

4. If you are working with a partner, your partner can now pull the changes in.

   cd ~/cs63/labs/07
   git pull
   

Using pyrobot

Begin by familiarizing yourself with pyrobot, a python library for experimenting with both physical and simulated robots. Go through the pyrobot overview. Start by invoking pyrobot and experiment with it using the command line. (You can skip the sections on the camera, gripper, and simulation devices since you won't be using these for this lab.)

Once you have familiarized yourself with the interface via the command line, edit the basicBrain.py file. Experiment with using this brain to control the robot. Modify the file in various ways, reload the brain in the robot, and retest it.

Building a subsumption brain

You will write a single subsumption-style brain that can find a light in the series of increasingly more difficult worlds shown below. These worlds are named:

• LightBehindWall
• LightInCorral
• LightBehindCorral
• LightEnclosed
• LightInMaze

The file findLight.py contains an example of how to implement a form of the subsumption architecture. It creates a set of layered behaviors which are then added to a brain in priority order (from lowest to highest). For each time step, the highest priority behavior that is triggered will control the robot. To test this file on the LightBehindWall world do:

pyrobot -s PyrobotSimulator -w LightBehindWall.py -r PyrobotRobot60000.py -b findLight.py

You will create your own subsumption-style controller in the file subsumption.py. You are welcome to use/modify the behaviors provided in the findLight.py file. In order to solve this problem you will need to add behaviors that push the robot to explore the world more effectively. Some options you may want to consider are:

• Create a gap seeking behavior that looks for openings and forces the robot to explore the environment and search for the light.
• Create a wall following behavior to help you find the light when it is not immediately visible.
• Create a map making behavior that tracks where the robot has been and pushes it to go to new locations. This approach will be much more involved. I suggest trying one of the other approaches first before attempting this. Note that although the robot stores its (x, y) location relative to where it began, this will quickly become incorrect due to noise. You may not use any global information that the simulator maintains about the robot's true (x, y) location.
• Use a combination of these, or invent your own behaviors.

Be sure to comment each of your behaviors. Your subsumption program should be able to find the light in all of the worlds depicted above, even if it may take awhile.

More challenging worlds

There are four additional worlds called Challenge1-Challenge4, on which you can test your subsumption controller. If your controller can consistently solve these, well done! I will demo some of the best student controllers in class.

Submitting your code