RRT-SLAM for motion planning with motion and map uncertainty for robot exploration

Abstract

We address the motion planning (MP) subproblem that arises in a robotic exploration and mapping task. We consider sensing, localization and mapping uncertainties in the motion planning subproblem. The robot is holonomic with known size and shape, and is equipped with a laser range sensor. We use a rapidly exploring randomized tree (RRT) in conjunction with a simulated particle based Simultaneous Localization and Mapping (SLAM) algorithm to expand the tree. The simulated SLAM explicitly accounts for sensor, localization and mapping uncertainty in the planning stage. Moreover, the RRT itself is represented in the augmented configuration space where an extra dimension of uncertainty is used. The collision likelihood along a planned path is explicitly computed and is used to select a planned path. Preliminary simulations show the effectiveness and benefits of our integrated approach.