Visual servoing for an omnidirectional mobile robot using the neural network - Multilayer perceptron

This paper presents a simple design of a visual (IBVS) designed for an omnidirectional mobile robot. Intends to work with an algorithm that allows the robot navigation and exploration in an unknown environment which will go from a start point to end point avoiding obstacles presented to it along the way and developing a work that can be used as part of a route planner high level, this being a critical and decisive factor in the field of robotics. We have chosen a mobile robot developed by a company called Rovio WowWeeTM as robotic platform for experiments. We utilized computer vision, image processing and sensor readings from an infrared IR, to detect obstacles of various shapes and colors characteristic. The robot is able to analyze information about the obstacles thanks to its integrated camera and data obtained from the image therefore decide in which direction to move through the design and training of a neural network “multilayer perceptron”. The general algorithm uses CGI commands, to communicate with the Mobile robot “Rovio”, such as LabVIEW graphical programming environment Matlab project and as a programming environment and training of the neural (Neural Network Toolbox).t.