Ausarbeitung/experiment1.tex

   1 \chapter{Experiment 1: Original Movement Behaviour}
   2 \label{sec:exp1}
   3 In the first experiment, the Roomba's original movement behaviour is
   4 measured to get an overview of the errors that occur while moving, and to
   5 establish a pool of data for correction approaches to work on later.
   6 There was no error correction involved (apart from any possible error
   7 correction the Roomba itself implements in its underlying logic, and which is
   8 not known to anyone except iRobot). To achieve this, the Wiselib Roomba Control
   9 is used to access the Roomba over the \ac{ROI}\index{Roomba Open Interface}.
  10
  11 \section{Setup}
  12 \begin{figure}[htbp]
  13  \centering
  14  \includegraphics[width=0.45\textwidth]{./images/IMAGE_00079.jpg}
  15  \caption{Roomba with netbook\label{fig:roombasetup}}
  16 \end{figure}
  17 \begin{figure}[htbp]
  18  \centering
  19  \includegraphics[width=0.6\textwidth]{./images/IMAGE_00148.jpg}
  20  \caption{Measuring turn angles with laser pointer\label{fig:laserpointer}}
  21 \end{figure}
  22
  23 The test equipment consisted of a small x86 \index{netbook} netbook which was
  24 mounted on an iRobot Roomba~530\index{Roomba} robot, as seen in
  25 Figure~\ref{fig:roombasetup}. The netbook controlled the Roomba over a
  26 \acs{USB}-to-serial converter plugged into the \ac{ROI}\index{Roomba
  27 Open Interface} port on the Roomba, and hosted as the environment for
  28 executing the Wiselib \index{Wiselib} code.
  29
  30 In this experiment, the Roomba started right off with the full velocity the
  31 movement was executed with; so there was no control to adhere a constant
  32 acceleration when starting or stopping the movement. As mentioned before, due to
  33 limitations in the \ac{ROI}\index{Roomba Open Interface} it is only possible to
  34 explicitly start and stop the Roomba's movements at different times, so the
  35 Wiselib's implementation of the Roomba control code first starts the Roomba's
  36 movement, keeps track of the turned angle and dcovered distance, and then stops
  37 the Roomba if these values exceed the target values.
  38
  39 The tests were done in two atomic drive modes: letting the Roomba walk a
  40 specific straight distance with a specific velocity in its viewing direction and
  41 letting it turn on the spot with a specific velocity about a specific angle.
  42 Each of the two modes was carried out on two different floor types\index{floor
  43 type}, a laminated floor and a carpet floor, to see if the movement behaviour
  44 significantly depended on the floor type. The side brush was removed, since the
  45 Roomba tends to turn slightly towards the left when driving straight on a
  46 carpet floor. Without the side brush, this was not the case.
  47
  48 The actual travelled distance of the straight drive tests were determined using
  49 a measuring tape with an accuracy of 1~mm. Only the distance in the Roomba's
  50 original viewing direction was considered, as it turned out that the offset
  51 perpendicular to the viewing direction and a possible shift in orientation were
  52 too small to be measured precisely.
  53
  54 The actual turn angles of the turn tests were determined using a
  55 \acs{ISO}/\acs{DIN}~A0 sheet of paper with a printed polar coordinate system in
  56 which a circular hole was cut in the center to let the Roomba's wheels touch the
  57 floor. The sheet was fixed on the floor, and the Roomba was aligned in the
  58 center of the paper. A laser pointer\index{laser pointer} attached to the Roomba
  59 pointed to the current orientation on the paper, as shown in
  60 Fig.~\ref{fig:laserpointer}. The accuracy for these tests was 1~degree.
  61
  62 After the initial setup, the application \prog{roomba\_test} (see
  63 section~\ref{sec:impl:measuring}) was started on the netbook for half-automatic
  64 testing. It used a predefined array of nominal distances, angles and velocities
  65 and for each pair of distance (for straight drive tests) or angle (for turn
  66 tests) and velocity it used the Wiselib implementation to move the Roomba. Then
  67 it asked the user to input the measured distance or angle the Roomba had moved
  68 and repeated the process until each combination of distance/angle and velocity
  69 was used. The nominal and input values were written to a log file, as well as
  70 the floor type, the Roomba ID, the \ac{SVN} revision the program was built
  71 from, the battery status\index{battery status}, and other internal
  72 implementation-specific values.
  73
  74 For the straight drive tests, the arrays with predefined values were:\\
  75 \begin{tabular}{@{}ll@{}}
  76 Distances: & 20, 50, 100, 200, 500, 1000, 2000, and 4000~mm \\
  77 Velocities: & 20, 50, 70, 100, 150, 200, 300, and 400~mm/s
  78 \end{tabular}
  79
  80 For the turn tests, the arrays with predefined values were: \\
  81 \begin{tabular}{@{}ll@{}}
  82 Turn angles: & 5, 15, 30, 45, 90, 120, 180, 360, 530, and 720~degree \\
  83 Velocities: & 20, 50, 70, 100, 150, 200, 300, and 400~mm/s
  84 \end{tabular}
  85
  86 According to the implementation of the Wiselib Roomba control, the velocities
  87 were given in mm/sec and referred to the distance the wheels travelled when the
  88 Roomba turned on the spot, which was a circle of 230~mm in diameter.
  89
  90 \section{Results}
  91 \label{exp1:results}
  92 The following graphs show the difference from the measured value to the input
  93 value for driving or turning with different velocities. Positive values mean
  94 that the Roomba had turned too much or travelled more than the target value,
  95 negative values mean that the Roomba had turned or travelled too less. The
  96 plots show multiple test runs; given are the minimum, the maximum and the
  97 arithmetic mean\index{arithmetic mean} of all results for a data point.
  98
  99 \todo{clearpage, cleardoublepage?}
 100 \begin{figure}[p!]
 101  \centering
 102  \includegraphics[width=\textwidth]{images/iz250flur_drive_data.pdf}
 103  \caption{Original behaviour on laminated floor, straight drive movements
 104   \label{fig:orig:lam:drive}}
 105 \end{figure}
 106 \begin{figure}[p!]
 107  \centering
 108  \includegraphics[width=\textwidth]{images/iz250flur_turn_data.pdf}
 109  \caption{Original behaviour on laminated floor, turn movements
 110   \label{fig:orig:lam:turn}}
 111 \end{figure}
 112 \begin{figure}[p!]
 113  \centering
 114  \includegraphics[width=\textwidth]{images/seminarraum_drive_data.pdf}
 115  \caption{Original behaviour on carpet floor, straight drive movements
 116   \label{fig:orig:carpet:drive}}
 117 \end{figure}
 118 \begin{figure}[p!]
 119  \centering
 120  \includegraphics[width=\textwidth]{images/seminarraum_turn_data.pdf}
 121  \caption{Original behaviour on carpet floor, turn movements
 122   \label{fig:orig:carpet:turn}}
 123 \end{figure}
 124
 125 Figure~\ref{fig:orig:lam:drive} shows that the error becomes greater with
 126 increasing input distance when driving straight on the laminated floor, however,
 127 in Figure~\ref{fig:orig:carpet:drive} we see the opposite effect on the carpet
 128 floor, the error decreases with greater input distance. This could happen due to
 129 imprecise measurement of distances in either the Roomba's sensors or the Wiselib
 130 implementation that controls the Roomba, or both, adding up over the time the
 131 movement continues. Also slippage of the wheels on the laminated floor could be
 132 possible, as well as slowdown through the carpet floor, explaining why the error
 133 increases on the laminated floor, but decreases into negative values on the
 134 carpet floor when the distance grows.
 135
 136 On the other hand, rising the velocity always seems to cause the
 137 error to increase.
 138
 139 The same effects also apply for turn tests. Additionally, there could be errors
 140 resulting from false assumptions about the Roomba's wheel base\index{wheel base}
 141 diameter, resulting in false calculations of the circle the Roomba's wheels
 142 describe while turning, and therefore leading to false results.