Modelling and Simulation of Fast Autonomous Exploration and Mapping System for GPS Denied Unknown Environment Using ROS

Abstract

This paper describes a multi-robot coordinated heterogeneous vehicle which employs cooperative unmanned ground vehicles (UGVs) and unmanned aerial vehicles (UAVs) to explore and map a GPS-denied unknown terrain. Exploration processes like cave exploration, tunnel inspections required an accurate and quick mapping of the environment in emergencies for rescue or maintenance operations. This multi-robot system’s main objective is to use the Dijkstra algorithm to produce a 2D and 3D map of an unexplored region. Each robot features a border identification system that uses image processing to derive border locations. The autonomous exploration problem is a partially observable Markov decision process (POMDP) to the local planner exploration component. The robot is directed to the uncharted region in the map after recovering boundary points in the region separating known from unknown space. The global planner graph, introducing flexible values, is constructed during the discovery phase to examine the robot’s border positions at the end. The two methodologies are combined with speeding up the search for border sites. Closed-loop control ensures that the robot is always pushed to its greatest advantage following achieving the desired advantages. Finally, the 2D and 3D maps obtained from each robot are combined using conventional map merging techniques to form a map.