Professor: Marek A. Perkowski, Electrical Engineering.

MIDTERM.



You have more than one week to complete this assignment.
The midterm should be returned on Saturday, May 15.
You can use lecture notes, web pages, books, whatever you want. The grade will be based on:
  1. knowledge and understanding of the class material.
  2. creativity and innovation.
  3. good usage of resources.
  4. quality of writing, good usage of tools and edititing/graphic software. Microsoft Word and graphics, formula editor in it. Possible, Matlab and free software from the Net.
This is what should be done:.
  1. Write an essay about a new type of a robot. Use your class knowledge. This robot may be a walking robot, mobile robot, flying robot, fish-like robot, or humanoid robot. It does not have to be similar to the robots that you know from the class or laboratory.
  2. There are only two constraints:
    1. You have to use Fuzzy Logic for some part of robot's software.
    2. You have to evolutionary approach for some part of robot's software.
  3. What is this robot supposed to do. Why it is innovative?
  4. Describe the mechanics of the robot. Effectors, sensors, kinematics.
  5. Write a scenario describing the robot behavior, such as mapping a cave, mapping a Moon crater, communicating with children in a museum, communicating with elderly patients in a care center, etc.
  6. What kind of sensor integration is necessary for this task?
  7. How can you use fuzzy logic expert system for your application: to integrate sensors, to control the robot, for human-robot interaction?
  8. For evolutionary approach you can use Genetic Algorithm, Genetic Programming, Evolutionary Strategies, GA combined with search, whatever algorithm (rule based for instance) combined with GA.
  9. You can evolve the robot body, the robot behavior, the robot body and behavior together, the vision system, the movement system (for a walking robot, for instance), the way how the robot responds in interaction. You can take any example from the class but you must deeply modify it. For instance, you may evolve rules to the fuzzy robot controller that we disscussed in the class but you have to develop more rules and be more specific what is evolved in the rules (not only fuzzy literals but rules themselves).
  10. You can use decision trees, SOP formulas, Ashenhurst-Curtis decomposition. You can combine them with fuzzy approach, how? You can combine them evolutionary approach, how?
  11. You have to find arguments why your proposed system is better (or at least from some point of view better) than the existing approaches.
  12. You can use your knowledge from the previous class or other classes succh a Neural Nets of Evolutionary Programming, for instance on robot vision, but you have to reasonably link it to the two main topics of this exam. For instance, if you worked on HMM or Decision tree, how they are related to GA and FL?
  13. The structure of your final report, in Word format with figures, is the following.
    1. Title - Should be interesting.
    2. Abstract.
    3. Introduction. What is your thesis about this work? What is your claim of innovation? What is new? What is the structure of the report?
    4. Previous work by other authors. Previous work by you (if any). Short characterization of this type of robots. Short characterization of software used in this type of robots, especially evolutionary and fuzzy.
    5. Intended behaviors of the robot. Fuzzy information, fuzzy behaviors.
    6. Proposed software system of the robot.
    7. Description of Rules. What are the input and output variables.
    8. Description of Evolutionary Algorithm. Representation of knowledge, strings, vectors, matrices, trees, decision diagrams? Crossover and mutation. Selection. Additional knowledge-based operators.
    9. The fitness function. Complete discussion of the function variants and its importance.
    10. Experimental results. How the work will be evaluated. How will you know that you achieved the goal? How experiments will be conducted?
    11. Conclusions. What has been achieved. Future work.
    12. Complete list of references that you used.