robolab @ unipv

Autonomous Navigation of a Wheeled Mobile Robot in Ventilation Ducts

Moses Koledoye

15/07/2015

The aim of this Master’s thesis is to develop a suitable algorithm for the autonomous navigation of a 4-wheeled
mobile robot tasked with the exhaustive exploration of a network of ventilation ducts. Following the development
of the algorithm, there must be a corresponding implementation in C++/Python code that will satisfy the timing
and resource constraints of the underlying operating system – ROS.

Contents

• Autonomous Robots
  • The MaaT-One Robot
  • The Mapping problem
  • The Path Planning problem
  • The Localization Problem

• The ROS framework

• Modeling and simulation

• Autonomous Navigation

• The Navigation Task: Finite State Machine

• The Navigation Task: Algorithms

  • Map Centroid Approach
  • Map Edge Approach

Results
This thesis was aimed at addressing the problem of driving a robot autonomously
through a network of ventilation ducts for complete exploration.
The sole objective was to develop an algorithm that could help achieve this
and provide a corresponding implementation in C++ code.
Two approaches were presented in this writing as the outcomes of this
undertaking: the Map Centroid and Map Edge approaches.

Future Work

• The robot is fitted with a camera that will be used for video documentation,
  capturing details about features in each part of the duct network.

• Set up landmarks at strategic points in the tunnel to help alleviate
  the problem of localization error accumulation.

---