Fuzzy control of a cartpole
Computational intelligence for Control and Embedded Systems
Construct an action mapping, based on fuzzy logic, for a simulation of a cartpole balancing and centring task (see attached CTM example). Define one or more performance criteria to evaluate how well your action mapping produces the desired behaviour of the cartpole system. Try to optimise the performance by experimenting with several alternatives, like increasing the number of rules or using different fuzzyfication and defuzzification methods.
Hints and suggestions
Attached is the matlab code for the dinamic equations of the system. Also attached is an example assignment which uses fuzzy logic control of the pid controller (code attached), which is a recommended way to attack the problem.
Keep the simulator as simple as possible and do not bother in writing a graphical interface to your simulator.
A concise report in the format of a technical paper consisting of the following sections:
1. A reasoned overview of how you construct your action mapping.
2. Problems encountered and your solutions to these problems.
3. Discussion of the behaviour obtained with regards to your performance criteria. Include graphs.
4. Analysis of the dependence of the behaviour on the number of rules and other design factors that affect the fuzzy action mapping.
5. Commented source code listing of your Matlab code.
Conciseness, clarity and completeness of the report 25%
Soundness of the action mapping design 25%
Correct implementation of the fuzzy action mapping 25%
Discussion of results 25%