Research Trends in AI Maze Solving using GA

1
Research Trends in AIMaze Solving using GA

• Muhammad Younas 2005-02-0110
• Hassan Javaid 2005-02-0304
• Danish Hussain 2006-02-0225

2
Flash Back

• In the last presentation we showed our
  implementation of two functions. One of them was
  the mutate function and the other was the cross
  over function.
• In this presentation we have identified many
  other functions that we will be needing for
  project and we have made the pseudo code of many
  of them and have implemented some of them.

3
Functions

• MutateTakes a genome, returns the same genome
  with some of the bits flipped.
• Cross Over Takes 2 genomes, performs one-point
  crossover on them to produce two new genomes.
• Mate
• Takes 2 individuals and performs crossover on
  their genomes to get 2 new genomes. It then
  mutates the new genomes. Finally, it makes 2 new
  individuals with dummy values for fitness and
  phenome.

4
Functions continued

• Selecting an IndividualTakes a population of
  individuals. Chooses a single individual randomly
  and returns that individual. This random choice
  is based on the fitness value of a genome in the
  population
• Random Population
• Takes a population size and number of genes in
  each genome and generates a population of
  individuals with random genomes
• Random Genome
• This function takes a genome-length and returns
  a random genome of that length.

5
Functions continued

• Make and Access individual's componentsThis
  function will takes a fitness, genome and phenome
  and returns a list containing these items.
• Run Genetic AlgorithmIt creates an initial
  population with random-pop and then does the
  following things for each generation
• Creates the phenome for each individual in the
  population.
• Evaluates the fitness of each individual.
• Selects individuals and mates them to produce
  the new generation

6
Simulation Plan

• Simulations would be run on a variety of mazes to
  ensure the reliability of the algorithm
• Fitness value would be determined by the number
  of steps taken in the maze
• This fitness value could be changed to obtain the
  optimal results in the simulation

7
Testing Genetic Algorithm

• It will be good to make sure that there are
  enough genes to at least allow the robot to
  finish the maze.
• We have researched that it is always useful to
  have large populations. How much it is useful we
  will see once we have implemented the whole GA.

8
(No Transcript)
9
(No Transcript)
10
Reference Information

• Please refer to our previous presentation for a
  tentative GUI
• The Mutate and Crossover function are also
  present in our previous presentation
• Simulations results would be presented next week
  when implementation is complete

11

• Q A