The purpose of this project is to develop a method for learning
  and motion planning for robot arm. TRN(Topological Representing Network)
  is the key concept in this implementation. TRN is used to learn the
  movement of the robot arm. With the result of learning , motion planning
  can be done using diffusion algorithm.  Also we can use the learning
  result to control the movement of robot arm. That is, given the
  coordinates as the input to TRN, TRN will generate the corresponding 
  values to control the robot arm to move. 
     Theoretically, the dimension of TRN can be arbitrary. This means
  the input to the TRN can be any dimensions and generate any dimension
  output. That is an advantage of TRN over some other kind of ANN. 
     With diffusion algorithm, the path found from source to destination
  has the fewest number of movement, namely passing fewest intermediate
  neural nodes.