Project 3: Reinforcement Learning

Pacman seeks reward.
Should he eat or should he run?
When in doubt, q-learn.

Introduction

In this project, you will implement value iteration and q-learning. You will test your agents first on Gridworld (from class), then apply them to a simulated robot controller (Crawler) and Pacman.

The code for this project contains the following files, which are available in a zip archive:

Files you will edit

Files you should read but NOT edit

Files you can ignore

What to submit: You will fill in portions of valueIterationAgents.py, qlearningAgents.py, and analysis.py during the assignment. You should submit only these files. Please don't change any others.

This assignment should be submitted with the assignment name cs343-3-reinforcement using these submission instructions.

Evaluation: Your code will be autograded for technical correctness. Please do not change the names of any provided functions or classes within the code, or you will wreak havoc on the autograder. However, the correctness of your implementation -- not the autograder's judgements -- will be the final judge of your score. If necessary, we will review and grade assignments individually to ensure that you receive due credit for your work.

Academic Dishonesty: We will be checking your code against other submissions in the class for logical redundancy. If you copy someone else's code and submit it with minor changes, we will know. These cheat detectors are quite hard to fool, so please don't try. We trust you all to submit your own work only; please don't let us down. If you do, we will pursue the strongest consequences available to us.

Getting Help: You are not alone! If you find yourself stuck on something, contact the course staff for help. Office hours, discussion board, and the mailing list are there for your support; please use them. If you can't make our office hours, let us know and we will schedule more. We want these projects to be rewarding and instructional, not frustrating and demoralizing. But, we don't know when or how to help unless you ask. One more piece of advice: if you don't know what a variable does or what kind of values it takes, print it out.

MDPs

To get started, run Gridworld in manual control mode, which uses the arrow keys:

python gridworld.py -m

You will see the two-exit layout from class. The blue dot is the agent. Note that when you press up, the agent only actually moves north 80% of the time. Such is the life of a Gridworld agent! 

You can control many aspects of the simulation.  A full list of options is available by running:

 python gridworld.py -h

The default agent moves randomly

 python gridworld.py -g MazeGrid

You should see the random agent bounce around the grid until it happens upon an exit.  Not the finest hour for an AI agent.

Note: The Gridworld MDP is such that you first must enter a pre-terminal state (the double boxes shown in the GUI) and then take the special 'exit' action before the episode actually ends (in the true terminal state called TERMINAL_STATE, which is not shown in the GUI).  If you run an episode manually, your total return may be less than you expected, due to the discount rate (-d to change; 0.9 by default).

Look at the console output that accompanies the graphical output (or use -t for all text). You will be told about each transition the agent experiences (to turn this off, use -q). 

As in Pacman, positions are represented by (x,y) Cartesian coordinates and any arrays are indexed by [x][y], with 'north' being the direction of increasing y, etc.  By default, most transitions will receive a reward of zero, though you can change this with the living reward option (-r). 

Question 1 (6 points)  Write a value iteration agent in ValueIterationAgent, which has been partially specified for you in valueIterationAgents.py.  Your value iteration agent is an offline planner, not a reinforcement agent, and so the relevant training option is the number of iterations of value iteration it should run (option -i) in its initial planning phase.  ValueIterationAgent takes an MDP on construction and runs value iteration for the specified number of iterations before the constructor returns.

Value iteration computes k-step estimates of the optimal values, V_k. In addition to running value iteration, implement the following methods for ValueIterationAgent using V_k.

• getValue(state) returns the value of a state.
• getPolicy(state) returns the best action according to computed values.
• getQValue(state, action) returns the q-value of the (state, action) pair. 

These quantities are all displayed in the GUI: values are numbers in squares, q-values are numbers in square quarters, and policies are arrows out from each square.

Important: Use the "batch" version of value iteration where each vector V_k is computed from a fixed vector V_k-1 (like in lecture), not the "online" version where one single weight vector is updated in place. The difference is discussed in Sutton & Barto in the 6th paragraph of chapter 4.1.

Note: A policy synthesized from values of depth k (which reflect the next k rewards) will actually reflect the next k+1 rewards (i.e. you return π_k+1). Similarly, the q-values will also reflect one more reward than the values (i.e. you return Q_k+1). You may assume that 100 iterations is enough for convergence in the questions below.

The following command loads your ValueIterationAgent, which will compute a policy and execute it 10 times. Press a key to cycle through values, q-values, and the simulation. You should find that the value of the start state (V(start)) and the empirical resulting average reward are quite close.

python gridworld.py -a value -i 100 -k 10

Hint: On the default BookGrid, running value iteration for 5 iterations should give you this output:

python gridworld.py -a value -i 5

Your value iteration agent will be graded on a new grid. We will check your values, q-values, and policies after fixed numbers of iterations and at convergence (e.g. after 100 iterations).

Question 2 (1 point) On BridgeGrid with the default discount of 0.9 and the default noise of 0.2, the optimal policy does not cross the bridge. Change only ONE of the discount and noise parameters so that the optimal policy causes the agent to attempt to cross the bridge. Put your answer in question2() of analysis.py. (Noise refers to how often an agent ends up in an unintended successor state when they perform an action.)  The default corresponds to:

python gridworld.py -a value -i 100 -g BridgeGrid --discount 0.9 --noise 0.2

Question 3 (5 points) On the DiscountGrid, give an assignment of parameter values for discount, noise, and livingReward which produce the following optimal policy types or state that the policy is impossible by returning the string 'NOT POSSIBLE'. The default corresponds to:

python gridworld.py -a value -i 100 -g DiscountGrid --discount 0.9 --noise 0.2 --livingReward 0.0

1. Prefer the close exit (+1), risking the cliff (-10)
2. Prefer the close exit (+1), but avoiding the cliff (-10)
3. Prefer the distant exit (+10), risking the cliff (-10)
4. Prefer the distant exit (+10), avoiding the cliff (-10)
5. Avoid both exits (also avoiding the cliff)

Note: You can check your policies in the GUI. For example, using a correct answer to 3(a), the arrow in (0,1) should point east, the arrow in (1,1) should also point east, and the arrow in (2,1) should point north.

Q-learning

Note that your value iteration agent does not actually learn from experience.  Rather, it ponders its MDP model to arrive at a complete policy before ever interacting with a real environment.  When it does interact with the environment, it simply follows the precomputed policy (e.g. it becomes a reflex agent). This distinction may be subtle in a simulated environment like a Gridword, but it's very important in the real world, where the real MDP is not available. 

Question 4 (5 points) You will now write a q-learning agent, which does very little on construction, but instead learns by trial and error from interactions with the environment through its update(state, action, nextState, reward) method.  A stub of a q-learner is specified in QLearningAgent in qlearningAgents.py, and you can select it with the option '-a q'. For this question, you must implement the update, getValue, getQValue, and getPolicy methods.

Note: For getValue and getPolicy, you should break ties randomly for better behavior. The random.choice() function will help. In a particular state, actions that your agent hasn't seen before still have a Q-value, specifically a Q-value of zero, and if all of the actions that your agent has seen before have a negative Q-value, an unseen action may be optimal.

With the q-learning update in place, you can watch your q-learner learn under manual control, using the keyboard:

python gridworld.py -a q -k 5 -m

Hint: Use the util.Counter class in util.py, which is a dictionary with a default value of zero. Methods such as totalCount should simplify your code. However, be careful with argMax: the actual argmax you want may be a key not in the counter!

Question 5 (2 points) Complete your q-learning agent by implementing epsilon-greedy action selection in getAction, meaning it chooses random actions epsilon of the time, and follows its current best q-values otherwise.

python gridworld.py -a q -k 100 

You can choose an element from a list uniformly at random by calling the random.choice function. You can simulate a binary variable with probability p of success by using util.flipCoin(p), which returns True with probability p and False with probability 1-p.

Question 6 (1 points) First, train a completely random q-learner with the default learning rate on the noiseless BridgeGrid for 50 episodes and observe whether it finds the optimal policy.

python gridworld.py -a q -k 50 -n 0 -g BridgeGrid -e 1

Question 7 (1 point) With no additional code, you should now be able to run a q-learning crawler robot:

 python crawler.py

This will invoke a crawling robot using your q-learner.  Play around with the various learning parameters to see how they affect the agent's policies and actions.   Note that the step delay is a parameter of the simulation, whereas the learning rate and epsilon are parameters of your learning algorithm, and the discount factor is a property of the environment.  

Approximate Q-learning and State Abstraction

Question 8 (1 points) Time to play some Pacman! Pacman will play games in two phases. In the first phase, training, Pacman will begin to learn about the values of positions and actions. Because it takes a very long time to learn accurate q-values even for tiny grids, Pacman's training games run in quiet mode by default, with no GUI (or console) display. Once Pacman's training is complete, he will enter testing mode. When testing, Pacman's self.epsilon and self.alpha will be set to 0.0, effectively stopping q-learning and disabling exploration, in order to allow Pacman to exploit his learned policy. Test games are shown in the GUI by default. Without any code changes you should be able to run q-learning Pacman for very tiny grids as follows:

 python pacman.py -p PacmanQAgent -x 2000 -n 2010 -l smallGrid 

Hint: If your QLearningAgent works for gridworld.py and crawler.py but does not seem to be learning a good policy for Pacman on smallGrid, it may be because your getAction and/or getPolicy methods do not in some cases properly consider unseen actions. In particular, because unseen actions have by definition a Q-value of zero, if all of the actions that have been seen have negative Q-values, an unseen action may be optimal.

Note: If you want to experiment with learning parameters, you can use the option -a, for example -a epsilon=0.1,alpha=0.3,gamma=0.7. These values will then be accessible as self.epsilon, self.gamma and self.alpha inside the agent.

Note: While a total of 2010 games will be played, the first 2000 games will not be displayed because of the option -x 2000, which designates the first 2000 games for training (no output). Thus, you will only see Pacman play the last 10 of these games. The number of training games is also passed to your agent as the option numTraining.

Note: If you want to watch 10 training games to see what's going on, use the command:

 python pacman.py -p PacmanQAgent -n 10 -l smallGrid -a numTraining=10

Make sure you understand what is happening here: the MDP state is the exact board configuration facing Pacman, with the now complex transitions describing an entire ply of change to that state. The intermediate game configurations in which Pacman has moved but the ghosts have not replied are not MDP states, but are bundled in to the transitions.

Once Pacman is done training, he should win very reliably in test games (at least 90% of the time), since now he is exploiting his learned policy.

However, you'll find that training the same agent on the seemingly simple mediumGrid may not work well. In our implementation, Pacman's average training rewards remain negative throughout training. At test time, he plays badly, probably losing all of his test games. Training will also take a long time, despite its ineffectiveness.

Pacman fails to win on larger layouts because each board configuration is a separate state with separate q-values. He has no way to generalize that running into a ghost is bad for all positions. Obviously, this approach will not scale.

Question 9 (3 points) Implement an approximate q-learning agent that learns weights for features of states, where many states might share the same features. Write your implementation in ApproximateQAgent class in qlearningAgents.py, which is a subclass of PacmanQAgent.

Note: Approximate q-learning assumes the existence of a feature function f(s,a) over state and action pairs, which yields a vector f₁(s,a) .. f_i(s,a) .. f_n(s,a) of feature values. We provide feature functions for you in featureExtractors.py. Feature vectors are util.Counter (like a dictionary) objects containing the non-zero pairs of features and values; all omitted features have value zero.

The approximate q-function takes the following form

By default, ApproximateQAgent uses the IdentityExtractor, which assigns a single feature to every (state,action) pair. With this feature extractor, your approximate q-learning agent should work identically to PacmanQAgent. You can test this with the following command:

 python pacman.py -p ApproximateQAgent -x 2000 -n 2010 -l smallGrid 

Important: ApproximateQAgent is a subclass of QLearningAgent, and it therefore shares several methods like getAction. Make sure that your methods in QLearningAgent call getQValue instead of accessing q-values directly, so that when you override getQValue in your approximate agent, the new approximate q-values are used to compute actions.

Once you're confident that your approximate learner works correctly with the identity features, run your approximate q-learning agent with our custom feature extractor, which can learn to win with ease:

 python pacman.py -p ApproximateQAgent -a extractor=SimpleExtractor -x 50 -n 60 -l mediumGrid 

 python pacman.py -p ApproximateQAgent -a extractor=SimpleExtractor -x 50 -n 60 -l mediumClassic 

If you have no errors, your approximate q-learning agent should win almost every time with these simple features, even with only 50 training games.

Congratulations! You have a learning Pacman agent!