Efficient exploration method for SLAM using random points sampling

Abstract

Developing fully automatic robot platforms to operate in dangerous environments or inaccessible place shown as figure \ref{robot-app} would be beneficial for humankind. Such robotic application is the necessity for the robot to collect sensor data about its environment and present this material to humans in an understandable manner. The robot might present the data to the human as a map, or model of the environment along with a path representing the robot’s trajectory. These correspond to \emph{Simultaneous Localization and Mapping (SLAM) problem}. The difficulty of manually constructing a highly accurate map has motivated the research community to develop autonomous and inexpensive solutions for robotic mapping. The robot needs to have some representation of its pose within the world for successful navigation. Absolute pose in world coordinates is not always necessary, but an accurate pose relative to the starting location is required for many applications. Moreover, It is necessary for robots to navigate such unknown terrain autonomously with decision on next destination by itself. It corresponds to \emph{Exploration problem}. In this thesis, an novel exploration method for simultaneous localization and mapping of multiple mobile robot is proposed, called the Random Point Sampling (RPS) exploration algorithm. Conventional exploration methods only consider moving toward unexplored region and it may lead to increase pose uncertainty of robot. Moreover, conventional exploration methods often have higher computational cost for goal assignment. Proposed RPS exploration algorithm reduce computational cost by sampling random candidate points in search space. Furthermore, using entropy of candidate goal points and importance weight of current pose, make decision to move toward unknown region or to go back to already explored region to reduce pose uncertainty. Simulation results show that RPS exploration algorithm reduce much time for computation and travel dist...