Filtering-RRT for Autonomous Indoor Navigation

This paper proposes a modified mapping approach for an autonomous navigation system to traverse unknown territories. The proposed method uses the features of Light Detection and Ranging (LiDAR) to self-explore its environment and create a map without human assistance using precise algorithms like Rapid Random Trees (RRT), GMapping, and ROS Navigation Stack. Further, a web interface is provided to control the robot from a non-Linux machine. Thus, the proposed solution is effective and self-sufficient for a flat space environment. The working of the developed robot is also demonstrated through autonomous mapping in a physical lab space.

I. Introduction

Autonomous robots have become indispensable in numerous industries worldwide, spanning manufacturing, services, and goods sectors. Conventional robots typically use depth or visual information to map their surroundings for accurate navigation. Traditionally, mapping a robot's environment was manual, with humans providing control commands for the mapping process [1]–[3]. This method proved time-consuming and physically demanding, mainly when mapping extensive spaces. Many solutions have emerged in recent years to tackle comprehensive exploration in unfamiliar territories, encompassing pattern-based, frontier, and entropy-driven approaches. Nevertheless, the Rapidly-Exploring Random Tree (RRT) method stands out for its computational efficiency [4]–[6].