Learning a Map
ECE Ph.D. Student
About the project
For Robot Localizing and Mapping, a robot is driven through the hallways of a building to make a floor map of the building. Using an IMU, acceleration and angular rates for the robot are collected to determine the robotís state, such as its orientation and position relative to the start location. The robotís state is recorded during the run and used to create a map of the building from the start location to the end. After the map is created, another robot follows the map autonomously using its inertial and sonar measurements from 8 sonar sensors on the robot. The robot centers itself in the hallway using sonar measurements from its sides, while monitoring the map for upcoming turns. The robot checks for opening on its side, indicated by max sonar readings on its side, then takes a turn to go into the next map segment. It repeats these steps until it reaches the end of the map. Using this algorithm, multiple robots can be given the task of mapping an entire building from different starting locations. Their shared map of the environment can be used to perform a faster localization or search within the mapped environment.
Professor Gilmer Blankenship, Director
J. Karvounis, J. Patel, M. Stanley graduate assistants
Office: 1427 A.V. Williams Building