2D Simultaneous Localization And Mapping

A common challenge for autonomous robots is the Simultaneous Localization and Mapping (SLAM) problem: given an unknown environment, can the robot simultaneously generate a map of its surroundings and locate itself in this map? In this project, a solution to the SLAM problem was implemented on a Pioneer 1 robot equipped with a SICK laser scanner. Extended Kalman filtering was used to continuously estimate the robot's position within the map and the associated covariances. Although landmark updates were not fully implemented, heading updates were performed using a structural compass. Despite lacking landmark updates, our solution produced a reasonably accurate map of the indoor environment. Introduction: Autonomous robots are becoming increasingly ubiquitous. Most autonomous robots require a map of their surroundings and an estimate of their location within this map. It is also required that this map be generated autonomously. This problem is known as Simultaneous Localization and Mapping (SLAM). The solution to the SLAM problem employed in this project has three main steps: propagation, compass update, and landmark update. The propagation step uses a kinematic motion model with Kalman Filtering to predict the state and covariance of the robot in the next time-step. Next, laser scan data is used to detect the walls around the robot, which can be used to improve the estimate of the robot heading. Finally, landmarks are detected and an Extended Kalman Filter is employed to further improve our state estimate.