Dynamic neural network approach to integrated autonomous path planning and navigation.

摘要

A new method for autonomously navigating a robot in an indoor dynamic environment was developed. A fuzzy inference system (FIS) and a neural network were developed and tested, and the FIS was better able to navigate the robot in a complex environment. The FIS system was implemented in the embedded hardware of the robot, thus eliminating a laptop computer running MATLAB and LabVIEW. A wrapper for the FIS was made to accept human directions over a TCP/IP protocol. The protocol can be used to navigate a robot autonomously from one location to another in the same building using the FIS.