implementation.tex

\subsection{Working Model}
    The project group is divided into two subgroups; Hardware and Software, and assignments are distributed respectively.\newline
    The workflow is an implementation of the waterfall model, where different tasks are to be done within a set time. These tasks are distributed via a Gantt schedule that is set by the project leader.
    \subsection{Implementation - Gitflow workflow} 
    The gitflow workflow defines a strict branching model designed around the project release.
    Instead of one main branch called master, we will work with two; master and develop.\newline
    The master branch is used to store the release history while {\it develop} is used as a feature integration branch for the software developers. 

    \begin{figure}[h!]
        \centering
        \makebox[\textwidth][c]{\includegraphics[width = 1.2\textwidth]{Pictures/gitWorkflow.png}}
        \caption{GitFlow Workflow}
        \label{fig:gitflow}
    \end{figure}


    
    \subsection{Resource and activity planning}
    \subsection{Time monitoring}


\subsection{IR Sensor development}
    Two kinds of IR sensors were chosen as a backup, to cover for the sonar should it encounter problems in the mineshaft.
    As the sonar has a nominal range of $\SI{2}{}$ to $\SI{400}{\centi\meter}$, both of the IR sensors were required to fully cover this range.
    The short range IR, however, has a minimum range of $\SI{20}{\centi\meter}$.
    Sensor response within this range coincide with response from targets much further away, leading to unreliable measurment.
    This was deemed a non-issue, as the minimum distance overlaps with the total copter width.
    Thus, any measurment values that could be ambiguous can be ignored, as the copter would already have crashed.

    Six clusters of IR sensor pairs (long and short) were to be mounted at $90^{\circ}$ angle relative each other, pointing nominally up, down, forward, backward and to each side.
    Thus, the double mounted IR clusters could cover the range should the sonar fail.

    As development progressed, issues were found with the IR sensors.
    Both sensors hade issues with accuracy at range, as well as low update rate, in comparison to the sonars.
    They were also difficult to implement in software.
    The decision was thus made that all ADC and IR development were to be put on halt in favour of SPI.
    Additionally, the focus was to make the short-range IR sensors work first, and have them act mostly as a short-range collision avoidance system.
    The long-ranged sensors were to be implemented either if the sonar performed poorly or if time could be spared from other parts of the project.