engleski

Fast Active SLAM for Accurate and Complete Coverage Mapping of Unknown Environments

In this paper we present Active SLAM solution with active loop closing component that is independent on exploration component at the same time allowing high accuracy robot's pose estimation and complete environment mapping. Inputs to our SLAM algorithm are odometry estimates obtained from IMU and wheel encoders. SLAM is based on Exactly Sparse Delayed State Filter (ESDSF) for real-time estimation of robot's trajectory and vision based pose registration and loop closing. Active component ensures that localization remains accurate over long periods by sending a robot to close loops if criterion functions satisfies a predefined value. Our criterion function depends on the number of states predicted without update between predictions, information gain from loop closing and the sheer distance between a loop closing state location and the current robot location. Once a state in which loop closure should occur is reached and update is performed the robot returns to its previous goals. Since active component is independent of the exploration part, SLAM described in this paper can easily be merged with any existing exploration algorithm, the only requirement is that the exploration algorithm is able to stop exploration at any time and continue exploration after loop closing has been accomplished. In this paper, we propose active SLAM integration with LRF based exploration algorithm that ensures complete coverage of a polygonal environment and therefore detailed mapping. Developed Active SLAM solution has been verified experimentally which demonstrated its capability to work in real-time and consistently map polygonal environments.

active SLAM; exploration; path planning

nije evidentirano