Windy Grid World

This assignment is to use Reinforcement Learning to solve the following "Windy Grid World" problem illustrated in the above picture. Each cell in the image is a state. There are four actions: move up, down, left, and right. This is a deterministic domain -- each action deterministically moves the agent one cell in the direction indicated. If the agent is on the boundary of the world and executes an action that would move it "off" of the world, it remains on the grid in the same cell from which it executed the action.
Notice that there are arrows drawn in some states in the diagram. These are the "windy" states. In these states, the agent experiences an extra "push" upward. For example, if the agent is in a windy state and executes an action to the left or right, the result of the action is to move left or right (respectively) but also to move one cell upward. As a result, the agent moves diagonally upward to the left or right.
This is an episodic task where each episode lasts no more than 30 time steps. At the beginning of each episode, the agent is placed in the "Start" state. Reward in this domain is zero everywhere except when the agent is in the goal state (labeled "goal" in the diagram). The agent receives a reward of positive ten when it executes any action {\it from} the goal state. The episode ends after 30 time steps or when the agent takes any action after having landed in the goal state.
You should solve the problem using Q-learning. Use e-greedy exploration with epsilon=0.1 (the agent takes a random action 10 percent of the time in order to explore.) Use a learning rate of 0.1 and a discount rate of 0.9.
The programming should be done in MATLAB. Students may get access to MATLAB here. Alternatively, students may code in Python (using Numpy). If the student would rather code in a different language, please see Dr Platt or the TA.
Students should submit their homework via email to the TA (suchismi@buffalo.edu) in the form of a ZIP file that includes the following:
1. A PDF of a plot of gridworld that illustrates the policy and a path found by Q-learning after it has approximately converged. The policy plot should identify the action taken by the policy in each state. The path should begin in the start state and follow the policy to the goal state.
2. A PDF of a plot of reward per episode. It should look like the diagram in Figure 6.13 in SB.
3. A text file showing output from a sample run of your code.
4. A directory containing all source code for your project.
5. A short readme file enumerating the important files in your submission.
Updates
You can initialize the Q function randomly or you can initialize it to a uniform value of 10. That is, you can initialize Q such that each value in the table is equal to 10.
There have been questions about how to know when the algorithm has converged. The algorithm has converged when the value function has stopped changing signficantly and the policy has stopped changing completely. Since we are using q-learning, the algorithm should converge to a single optimal policy.
Please also submit a short readme file with your homework that enumerates the important files in your submission.

Posted Date: 2/9/2013 2:30:26 PM | Location : United States

Ask an Expert

Related Discussions:- mathlab , Assignment Help, Ask Question on mathlab , Get Answer, Expert's Help, mathlab Discussions

Write discussion on mathlab
Your posts are moderated

Write your message here..

Related Questions

Design a 8 to 1 multiplexer using the fourvariable function, Design a 8 to ... Design a 8 to 1 multiplexer by using the fourvariable function given by F(A, B, C, D) = ∑ m(0,1,3,4,8,9,15). Ans. Design of 8 to 1 Multiplexer: It is a four-variable function a

What are the differences between one hot and binary encoding, What are the ... What are the differences between one hot and binary encoding? Common classifications used to explain the state encoding of an FSM is Binary or highly encoded and one hot. A bin

And-elimination rule, And-Elimination rule: In generally English says ... And-Elimination rule: In generally English says that "if you know that lots of things are all true, so you know like any one of them is also true". Because you can specify a c

Exit command, Your shell must accept the exit command. This command will ... Your shell must accept the exit command. This command will be in the form: e x i t When you encounter this command, your shell should terminate. Please note that when you

C LANGUAGE PROGRAMME, find a c program to find the area under the curve y=f... find a c program to find the area under the curve y=f(x) between x=a and x=b,intregrate y=f(x)between the limits of a and b.the area under a curve between two points can be found b

What are the characteristics of the e-cash, What are the characteristics of... What are the characteristics of the E-Cash? These kinds of payments, turning the Internet within a transaction oriented forum, need mediums which are easy and cheap (through a

How to clear computer motherboard cmos password, As CMOS is a special chip ... As CMOS is a special chip with its own battery, the best way to clear out a CMOS chip is to cut off it from its power supply. To clear the CMOS password you just take away the

Define e-commerce, Define e-commerce? The term 'electronic commerce' ha... Define e-commerce? The term 'electronic commerce' has develop from electronic shopping, to imply all aspects of business and market processes enabled by the Internet and World

Explain the typical functions of operating systems, List the typical functi... List the typical functions of operating systems. Classical functions of operating system are given as: (1) Process management: A process is a program throughout execution.

Chains of inference, Chains of Inference: Now we have to look at how t... Chains of Inference: Now we have to look at how to get an agent to prove a given theorem using various search strategies? Thus we have noted in previous lectures that, there i