Automatically navigated cars that drive on their own and robots that can perform any manual work science fiction authors predicted that it would all be possible in the 20th Century (see Asimov, 1951 "I, Robot"). In the dawn of artificial intelligence, the main obstacle that has to be overcome was creating robots that are able to solve logical problems. Humans, as far as we know, are unique in being able to solve abstract logical problems. Thus, creating a robot to mimic this ability is an extremely complex task.
What researchers initially perceived as a much easier task - having a robot navigate and interact with the physical environment, however, proved equally difficult. Robot vision and localization are just two of the tasks that need to be solved effectively before robots can truly do what Asimov predicted.
Robot localization specifically is at the heart of any artificial intelligence system that needs to navigate in the physical world. It is not possible for a robot to carry out tasks that involve moving in the environment if every time it moves, it gets lost.
In this paper we examine an implementation of the Monte Carlo localization algorithm for the Pioneer 2 robots. Our implementation of the algorithm utilizes the robot motion sensors, sonar sensors and camera to gain as detailed a picture as possible of its environment to allow it to navigate successfully.
Stamenova, Stiliyana, "Solving the Maze: Robot Localization Using the Monte Carlo Localization Algorithm and Shape Context" (2009). Honors Projects. Paper 15.
© Copyright is owned by author of this document