19th Annual Student Research Conference
Mathematics and Computer Science
Robot Navigation using Artificial Neural Networks
Spondon Saha
Dr. Alan Garvey, Faculty Mentor
The aim of this research was to engineer an Artificial Neural Network (ANN) that could learn the behavior of Collision Avoidance for robots. A Collision Avoidance algorithm was therefore used to train this ANN. The algorithm was initially designed for a class project during the Fall of 2004 but was then later modified for this research so that it could train the Neural Network to embody the said behavior. Once the Network learned the behavior with the highest possible accuracy, it was tested on a robot simulator and a field-robot. Testing on the field-robot however required me to design a driver for the robot that could communicate with the Pyro API. But in the end, the project was successfully completed with a working ANN that could be “plugged” into a robot and a driver that also enabled the ANN to communicate with the field-robot.
Keywords: Neural Network, Collsion Avoidance, Robotics, Pyro
Topic(s):Computer Science
Presentation Type: Poster
Session: 60-53
Location: OP Lobby and Atrium
Time: 4:15