naresh/master-thesis
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases 11 Wiki Activity

implemented feedback and changes

Browse Source
This commit is contained in:
Nareshkumar Rao
2022-02-18 21:44:14 +01:00
parent 5e44f7535f
commit d13b09beb7
6 changed files with 156 additions and 102 deletions
Download Patch File Download Diff File Expand all files Collapse all files

37
4-Validation/Validation.tex
View File

@ -1,22 +1,28 @@
\chapter{Validation}\label{chap:validation}
As was briefly touched upon in \autoref{sec:developmentphases}, the development of this project was carried out in iterative phases. After each iteration, the developed features were validated for their functionality and suitability. This chapter will go over these stages---sandbox, laboratory and field-testing---in depth.
\section{Sandbox Stage}
\begin{figure}[h]
\centering
\includegraphics[width=0.75\textwidth]{prototype/mug}
\caption{Visual representation of the sensor data plotted as a color-graph.}
\end{figure}
The fundamental objective of the sandbox stage of development was to investigate the suitability of the Intel RealSense L515 LIDAR sensor.
In this stage, the testing was mainly to determine
\begin{enumerate}
\item if a connection to the sensor can be established
\item if the data received can be visually represented and manipulated
\item if the data received can be visually represented and manipulated---see \autoref{fig:mug}
\item if the accuracy of the data was within a tolerable range
\end{enumerate}
\begin{figure}[h]
\centering
\includegraphics[width=0.75\textwidth]{prototype/mug}
\caption{Visual representation of the sensor data plotted as a color-graph.}
\label{fig:mug}
\end{figure}
In order to accomplish this, a precursor to the FlowDAR software was developed that was connected to the sensor directly over USB. The aim of the software was to process the raw sensor data in order to determine the cross-sectional area of an object upon a flat plane. In this case, a small cardboard box was placed against a wall.
As represented in \autoref{fig:prototype_program}, the software firstly calibrates itself to the flat plane---the wall---using linear regression to generate a straight-line. Then upon placing the object on the plane, using the techniques discussed in \autoref{chap:design}, the cross-sectional area of the object could be measured.
As represented in \autoref{fig:prototype_program}, the software first calibrates itself to the flat plane---the wall---using linear regression to generate a straight-line. Then upon placing the object on the plane, using the techniques discussed in \autoref{chap:design}, the cross-sectional area of the object could be measured.
An object with the cross-sectional area of \SI{3150}{\milli\meter\squared} was used in this validation. The software measured \SI{3288}{\milli\meter\squared}, yielding an error of \SI{4}{\percent}.
@ -56,9 +62,9 @@ An object with the cross-sectional area of \SI{3150}{\milli\meter\squared} was u
\end{figure}
\section{Laboratory Stage}
The development of the fundamental features of this project was done in an iterative process of rapid prototyping and testing. In order to accomplish this, a controlled environment that can be easily accessed and modified must be established. This setup was realized in the Telelaboratory\footnote{Laborator for the development of remote systems at the Faculty of Electrical Engineering, University of Applied Sciences Düsseldorf}.
The development of the fundamental features of this project was done in an iterative process of rapid prototyping and testing. In order to accomplish this, a controlled environment that can be easily accessed and modified must be established. This setup was realized in the Telelaboratory\footnote{Laboratory for the development of remote systems at the Faculty of Electrical Engineering, University of Applied Sciences Düsseldorf}.
The setup consisted of miniature looped conveyor belt system. The looped nature of this conveyor system was advantageous, as it could be loaded with material that would continuously circulate. This allowed development to be carry on uninterrupted and even remotely if necessary. The objects used to simulate material on the belt were miniature cars that were chosen simply for their availability and simple geometry.
The setup consisted of a miniature looped conveyor belt system. The looped nature of this conveyor system was advantageous, as it could be loaded with material that would continuously circulate. This allowed development to be carry on uninterrupted and even remotely if necessary. The objects used to simulate material on the belt were miniature cars that were chosen simply for their availability and simple geometry.
The LIDAR sensor was mounted using its ISO 1222 tripod mounting point on a regular camera tripod and positioned over the conveyor belt. The Raspberry Pi was connected to the laboratory network which allowed for configuration and testing to be done over the network. Using a VPN tunnel, further configuration and testing could also be done remotely from outside the laboratory network.
@ -97,7 +103,7 @@ The values sent to the controller by the FlowPi over the Profinet interface were
A simple PLC program was written to activate an output---in this case, turning on an LED---whenever the Cross-Sectional Area was over a certain threshold value.
The FlowPi software as well as the Profinet interface were shown to be functioning as the LEDs lit up in a robust manner whenever a miniature car passed under the scanning area of the LIDAR sensor. A rigorous measurement of the latency was not carried out, however the latency was deemed to be under a second.
The FlowPi software as well as the Profinet interface were shown to be functioning as the LEDs lit up in a robust manner whenever a miniature car passed under the scanning area of the LIDAR sensor. A rigorous measurement of the latency was not carried out, however the latency was deemed to be under one second.
\subsection{Linux RT-Patch}
In order to test the effect of the Linux RT-Patch, a simple test comparing the jitter values of Profinet-IO communications was conducted.
@ -107,7 +113,7 @@ The system was connected over Profinet to a Virtual PLC running on Codesys, and
The results show that the RT-Patched kernel had a maximum jitter of \SI{2166}{\micro\second}, which was \SI{26}{\percent} lower than the normal kernel. This lower jitter may be indicative of higher-determinism of the system.
\section{Field-Testing Stage}
The field-testing stage was carried out at the Siep Gravel Quarry\footnote{Siep Kieswerk GmbH \& Co. KG in Jülich. See \nameref{chap:ack}.}. There, a bucket loader was being used to excavate gravel into a hopper. The hopper first filtered out larger rocks and boulders through a set of evenly spaced rods. Acting as a buffer, the hopper would continuously load a conveyor belt with gravel.
The field-testing stage was carried out at the Siep Gravel Quarry\footnote{Siep Kieswerk GmbH \& Co. KG in Jülich. See \nameref{chap:ack}.}---see \autoref{fig:photo_overview}. There, a bucket loader was being used to excavate gravel into a hopper. The hopper first filtered out larger rocks and boulders through a set of evenly spaced rods. Acting as a buffer, the hopper would continuously load a conveyor belt with gravel.
The quarry currently uses a conventional belt scale system that was placed under the conveyor belt. This simultaneously measured both belt velocity and the material mass flow---in tonnes per hour---delivering the values to a PLC in a nearby control box.
@ -115,16 +121,19 @@ The quarry currently uses a conventional belt scale system that was placed under
\centering
\includegraphics[width=0.75\textwidth]{photographs/overview}
\caption{Siep Kieswerk GmbH \& Co. KG in Jülich where the field testing was carried out.}
\label{fig:photo_overview}
\end{figure}
\subsection{Setup and Testing}
The LIDAR sensor was attached to a walkway that went over the conveyor belt---see \autoref{fig:beltview}. The sensor must be placed at a minimum distance of \SI{0.5}{\meter} from the belt, in addition to clearance accounting for the height of the gravel on the belt as well. In this case, the sensor was placed at a height of \SI{1}{\meter} from the belt. The housed Raspberry Pi and various connections were also attached to the walkway.
\begin{figure}[h]
\centering
\includegraphics[width=0.75\textwidth]{photographs/beltview}
\caption{The view of the conveyor belt and material as seen by the sensor.}
\label{fig:beltview}
\end{figure}
\subsection{Setup and Testing}
The LIDAR sensor was attached to a walkway that went over the conveyor belt. The sensor must be placed at a minimum distance of \SI{0.5}{\meter} from the belt, in addition to clearance accounting for the height of the gravel on the belt as well. In this case, the sensor was placed at a height of \SI{1}{\meter} from the belt. The housed Raspberry Pi and various connections were also attached to the walkway.
A standard home-grade wireless access point was used to provide a local network, through which the configuration of the system could take place.
Once all the devices were connected and turned on, the pre-configured Raspberry Pi connected itself to the wireless access point that was reachable by the engineering laptop. A connection between the FlowRemote configuration software and the FlowPi processing software could be established and configuration could begin.