Cooperative SLAM using fuzzy Kalman filtering for a collaborative air-ground robotic system

摘要

This paper presents a three-dimensional (3D) cooperative simultaneous location and mapping (SLAM) method for a collaborative air-ground robotic system, designed to manage an indoor quadrotor flying done together with a Mecanum-wheeled omnidirectional robot (MWOR) in indoor unknown and no GPS environments. An ORB (Oriented Fast and Rotated BRIEF)-SLAM 2.0 (ORB- SLAM 2.0) approach is used to produce a 3D map and discover the position of the indoor quadrotor simultaneously, and a particle-filter SLAM (FastSLAM 2.0) approach is employed to build the 2D map of the global environment for the MWOR. A more accurate 3D quadrotor position estimation (QPE) method for the quadrotor is proposed with the assistance of the MWOR. A cooperative SLAM using fuzzy Kalman filtering is proposed to fuse the outputs of the ORB-SLAM 2.0, FastSLAM 2.0, and QPE approaches, in order to localize the quadrotor more accurately. Both SLAM approaches, quadrotor position estimation method and cooperative SLAM have been implemented in the robotic operation system (ROS) environment. Moreover, the cooperative SLAM method is exploited to achieve landing of the quadrotor atop the MWOR. Five experiments are conducted to show the effectiveness and superiority of the proposed cooperative SLAM method.