Thesis-Hector-VR/Concepts for operating ground based rescue robots using virtual reality/chapters/introduction.tex

\chapter{Introduction}

In recent years, natural disasters such as earthquakes, tsunamis and potentially nuclear explosives have seriously threatened the safety of human life and property. While the number of various disasters has increased, their severity, diversity and complexity have also gradually increased. The 72h after a disaster is the golden rescue time, but the unstructured environment of the disaster site makes it difficult for rescuers to work quickly, efficiently and safely.

Rescue robots have the advantages of high mobility and handling breaking capacity. They can work continuously to improve the efficiency of search and rescue. Also, those robots can achieve the detection of the graph, sound, gas and temperature within the ruins by carrying a variety of sensors.
When rescue robots can assist or replace the rescuers, the injuries caused by the secondary collapse could be avoided, and risks faced by rescuers might be lower. Thus, rescue robots have become an important development direction.

In fact, rescue robots have been put to use in a number of disaster scenarios. 
\gls{crasar} used rescue robots for Urban Search and Rescue task during the World Trade Center collapse in 2001 \cite{Casper:2003tk} and has employed rescue robots at multiple disaster sites in the years since to assist in finding survivors, inspecting buildings and scouting the site environment etc \cite{Murphy:2012th}. Anchor Diver III was utilized as underwater support to search for bodies drowned at sea after the 2011 Tohoku Earthquake and Tsunami \cite{Huang:2011wq}.

Considering the training time and space constraints for rescuers \cite{Murphy:2004wl}, and the goal of efficiency and fluency collaboration \cite{10.1145/1228716.1228718}, the appropriate \gls{hri} approach deserves to be investigated. Some of the existing \gls{hri} methods are Android software \cite{Sarkar:2017tt} \cite{Faisal:2019uu}, gesture recognition\cite{Sousa:2017tn} \cite{10.1145/2157689.2157818} \cite{Nagi:2014vu}, facial voice recognition \cite{Pourmehr:2013ta}, adopting eye movements \cite{Ma:2015wu}, \gls{ar} \cite{SOARES20151656} and \gls{vr}, etc.

% VR and robot
Among them, \gls{vr} has gained much attention due to its immersion and the interaction method that can be changed virtually. \gls{vr} is no longer a new word. With the development of technology in recent years, \gls{vr} devices are gradually becoming more accessible to users. With the improvement of hardware devices, the new generation of \gls{vr} headsets has higher resolution and a wider field of view. While \gls{vr} are often considered entertainment devices, \gls{vr} brings more than that. It plays an important role in many fields such as entertainment, training, education and medical care.

The use of \gls{vr} in \gls{hrc} also has the potential. In terms of reliability, \gls{vr} is reliable as a novel alternative to \gls{hri}. The interaction tasks that users can accomplish with \gls{vr} do not differ significantly from those using real operating systems\cite{Villani:2018ub}. In terms of user experience and operational efficiency, \gls{vr} headsets can provide users with stereo viewing cues, which makes collaborative \gls{hri} tasks in certain situations more efficient and performance better \cite{Liu:2017tw}. A novel \gls{vr}-based practical system for immersive robot teleoperation and scene exploration can improve the degree of immersion and situation awareness for the precise navigation of the robot as well as the interactive measurement of objects within the scene. In contrast, this level of immersion and interaction cannot be reached with video-only systems \cite{Stotko:2019ud}.

However, there remains a need to explore \gls{hri} patterns and improve the level of \gls{hrintergration}\cite{Wang:2017uy}. Intuitive and easy-to-use interactive patterns can enable the user to explore the environment as intentionally as possible and improve the efficiency of search and rescue. The appropriate interaction method should cause less mental and physical exhaustion, which also extends the length of an operation, making it less necessary for the user to frequently exit the \gls{vr} environment for rest.

% What I have done (overview)
For this purpose, this paper presents a preliminary \gls{vr}-based system that simulates the cooperation between ground rescue robots and humans with four different operation modes and corresponding test scenes, which imitate a post-disaster city. The test scene simulates a robot collaborating with Unity to construct a virtual 3D scene. The robot has a simulated \gls{lidar} remote sensor, which makes the display of the scene dependent on the robot's movement. In order to find an interactive approach that is as intuitive and low mental fatigue as possible, a user study was executed after the development was completed.


% Paper Architecture
In Chapter \ref{related}, related work involving the integration of \gls{vr} and \gls{hri} is presented.
Chapter \ref{implementation} provides details of the proposed system, including the techniques used for the different interaction modes and the setup for test scenes.
Chapter \ref{evaluate} explains the design and procedure of user study.
Chapter \ref{result} presents the results of the user study and analyzes the advantages and disadvantages of the different operation modes and the directions for improvement.
Finally, in Chapter \ref{conclusion}, conclusions and future work are summarized.