Localization And Map Building Algorithm For Autonomous Robot

With continuous advances in technology, robots are being used to address increasingly difficult problems in the real world. Robots can be used in situations where manpower is limited, or in environments where humans are unable or unwilling to go. Fundamental robot skills in these environments include the ability to navigate intelligently, to represent what the robot observes, and to share learned information with other robots or humans.Map building is a common tool that is widely used for representation of environments, sharing such representations, and for performing navigation. Unfortunately, despite advances in map-building and robotics, maps produced from typical environments are often bulky and the resulting representations are unintuitive. In this thesis, we present a new behavior-based mapping algorithm that enables efficient navigation and intuitive representation of environments. The research we present is useful in situations where inexpensive robots are required and computational resources are limited.The solution we present uses a measure of the open area around the robot derived from sonar readings to determine the set of afforded behaviors at that place in the environment. The afforded behaviors are loosely based on navigational primitives such as "turn left", "turn right", or "go forward" and are used to create a landmark that represents the place in the environment. By following afforded behaviors between landmarks, the robot is able to build a topological representation of environments.We verify our research by performing localization and map-building experiments in simulation. We show that in comparison to state of the art algorithms, we are able to reduce computational requirements because we use topological algorithms in comparison to occupancy grid algorithms. Furthermore, we are able to autonomously classify landmarks and build accurate topological maps in real-time.