Design and development of an autonomous duct inspection and mapping robot

Booysen, Tracy

Design and development of an autonomous duct inspection and mapping robot

Master Thesis

2008

Abstract

Just a few years ago, the idea of having robots in factories and households was science fiction. But, as robotic technology develops, this is becoming reality. Nowadays, robots not only perform simple household chores, but are used in most production lines and are even employed by the army. Visual inspection robots are very common and are used in many industries, including inspecting the interior of duct systems. Duct systems are in place in almost all large buildings and require ongoing maintenance and cleaning. Systems that are not properly maintained can pose a health risk as dust and mold form and are then blown throughout the building. In some cases, access holes have to be cut to allow access for inspection to occur. A robotic system, small enough to enter a duct through any existing access panel, would be advantageous. An autonomous robot would be even more useful as no operator would be needed thus reducing operating costs. To this end, a robot was developed that could autonomously navigate through a duct system, recoding video images and mapping the internal profile. The development of which is discussed in this thesis, included the design of the robotic platform, the inclusion of appropriate sensors and accompanying circuitry, generation of a simulation to test the control algorithm and implementing embedded software to control the robot. From the testing of the entire system the following conclusions were drawn. The robot as a whole performed well and navigated autonomously through the duct with a success rate of 90%. The system tests were repeatable and the odometry data closely matched the actual paths for straight line travel. The sonar data closely corresponded to the duct walls but was hard to interpret when the odometry and actual paths diverged. These paths diverged from each other due to wheel slip caused as the robot turned. The simulation developed showed that the control algorithm would ensure that the robot recursively inspected any duct system and provided information about the system as a whole. Further work should concentrate on improving the correlation between the odometry path and the actual path, perhaps by adding in a bearing measurement system. Sensors with greater range and accuracy should be implemented and the entire system re-tested. The embedded controller allowed for expansion should additional requirements be needed and was more then adequate for the task.

Keywords

Mechanical Engineering

Reference:

Collections

Masters

Full item page