Map based representation of navigation information for robust machines learning

Much research is being carried out in autonomous driving of vehicles under various disciplines but very few autonomous navigating vehicles are developed till date. This paper presents a novel technique, with a practical example, for the representation of navigational information under real-time considerations. Machine learning algorithms can easily be trained with the data sets developed from the representation and to testify this, an Artificial Neural Network is trained with a represented data set. This technique is independent of driving capabilities of vehicle and provides simple directional instructions for navigation. Using this method, an autonomous driving vehicle can be made to learn to navigate between locations in a known region. Moreover, the usage of ANN makes it completely adaptive and any changes or modifications in the trained region can easily be updated into the knowledge base.