Autonomous Robot Navigation Based on Pattern Recognition Techniques and Artificial Neural Networks

The autonomous navigation of robots is one of the main problems among the robots due to its complexity and dynamism as it depends on environmental conditions as the interaction between themselves, persons or any unannounced change in the environment. Pattern recognition has become an interesting research line in the area of robotics and computer vision, however, the problem of perception extends beyond that of classification, main idea is training a specified structure to perform the classifying a given pattern. In this work, we have proposed the application of pattern recognition techniques and neural networks with back propagation learning procedure for the autonomous robots navigation. The objective of this work is to achieve that a robot is capable of performing a path in an unknown environment, through pattern recognition identifying four classes that indicate what action to perform, and then, a dataset with 400 images that were randomly divided with 70% for the training process, 15% for validation and 15% for the test is generated to train by neural network with different configurations. This purpose ROS and robot TurtleBot 2 are used. The paper ends with a critical discussion of the experimental results.