Skip to content

Commit

Permalink
format inner html
Browse files Browse the repository at this point in the history
  • Loading branch information
@AndrewDTR
AndrewDTR committed Nov 25, 2024
1 parent 56c80c8 commit 23e3a8f
Showing 1 changed file with 96 additions and 15 deletions.
111 changes: 96 additions & 15 deletions src/pages/blog/upl-people-counter.md
View file
Original file line number Diff line number Diff line change
Expand Up @@ -77,40 +77,121 @@ It was around this time that I stumbled upon the homepage of [MITERS](https://mi
<span class="callout-icon">ℹ️</span>
<span>Note - Added 11/25/2024</span>
</div>
<p>In response to this article's release, I've received a <i>lot</i> of questions asking me why I made the choice to track the status of the doors, rather than other attributes of the room. Many of these questions came from <a href="https://news.ycombinator.com/item?id=41907360" target="_blank">my post on Hacker News</a>, as well as the article posted on <a href="https://hackaday.com/2024/10/24/keeping-tabs-on-an-undergraduate-projects-labs-door-status/" target="_blank">Hackaday</a> -- where one particularly amusing comment accused me of "...not looking at [the project] from a systems point of view." Here's my response to some of those questions, in no particular order:</p>
<p style="margin-bottom: 3px;">Why not...</p>
<ul style="margin-top: 0; margin-bottom: 3px;">
<p>
In response to this article's release, I've received a <i>lot</i> of
questions asking me why I made the choice to track the status of the doors,
rather than other attributes of the room. Many of these questions came from
<a href="https://news.ycombinator.com/item?id=41907360" target="_blank"
>my post on Hacker News</a
>, as well as the article posted on
<a
href="https://hackaday.com/2024/10/24/keeping-tabs-on-an-undergraduate-projects-labs-door-status/"
target="_blank"
>Hackaday</a
>
-- where one particularly amusing comment accused me of "...not looking at
[the project] from a systems point of view." Here's my response to some of
those questions, in no particular order:
</p>
<p style="margin-bottom: 3px">Why not...</p>
<ul style="margin-top: 0; margin-bottom: 3px">
<li>...point cameras at the doors?</li>
<ul>
<li>That could work too! However, it would just be far more complex than simply attaching the two Zigbee sensors on the doors -- I would have to train a model to recognize the difference between viewing the interior side of the UPL versus the CS walkway.</li>
<li>
That could work too! However, it would just be far more complex than
simply attaching the two Zigbee sensors on the doors -- I would have to
train a model to recognize the difference between viewing the interior
side of the UPL versus the CS walkway.
</li>
</ul>
</ul>
<ul style="margin-top: 0; margin-bottom: 3px;">
<ul style="margin-top: 0; margin-bottom: 3px">
<li>...point the camera at the lights of the room?</li>
<ul>
<li>One of the main things that people seemed to be confused about was the difference between tracking the <i>state</i> of the room versus its <i>occupancy</i>. The key distinction is that people being in the room doesn't necessarily indicate that it's open to the public. There could be coordinators having a meeting (or a maintenance person in there cleaning, and so on). Yes, it would be possible to have other ways of knowing whether there are people <i>in the room</i> (like the other suggestions mentioned here), but none indicate its status as well as the doors (other than people just writing "the room is closed to the public" on Discord).</li>
<li>
One of the main things that people seemed to be confused about was the
difference between tracking the <i>state</i> of the room versus its
<i>occupancy</i>. The key distinction is that people being in the room
doesn't necessarily indicate that it's open to the public. There could
be coordinators having a meeting (or a maintenance person in there
cleaning, and so on). Yes, it would be possible to have other ways of
knowing whether there are people <i>in the room</i> (like the other
suggestions mentioned here), but none indicate its status as well as the
doors (other than people just writing "the room is closed to the public"
on Discord).
</li>
</ul>
</ul>
<ul style="margin-top: 0; margin-bottom: 3px;">
<ul style="margin-top: 0; margin-bottom: 3px">
<li>...use a technological solution to detect devices in the lab?</li>
<ul>
<li>I saw a few suggestions about this -- namely, <a href="https://news.ycombinator.com/item?id=41908772" target="_blank">this one</a> by zdw mentioning the <span style="font-family: monospace;">finger</span> command, as well as <a href="https://news.ycombinator.com/item?id=41912287" target="_blank">this comment</a> by zimpenfish about using <span style="font-family: monospace;">rlogin</span> and <span style="font-family: monospace;">who</span> to determine who's logged into the network. Well -- the UPL doesn't have desktop computers available for students to come in and use anymore. Most students just bring their laptop, as computers aren't necessarily hard to come by anymore...! We don't have the infrastructure (at least, right now) to allow students to ssh into our servers, and even if we did, there would be no way of telling if they were doing it on-premises. As for using WiFi/DHCP leases to detect devices (as recommended by q3k in <a href="https://news.ycombinator.com/item?id=41913831" target="_blank">this comment</a>): UPLians use UWNet or eduroam, not the UPL's own WiFi network. There's also some privacy-related concerns about snooping/logging/aggregating users, even if anonymized, that highly dissuades us from implementing anything like that.</li>
<li>
I saw a few suggestions about this -- namely,
<a href="https://news.ycombinator.com/item?id=41908772" target="_blank"
>this one</a
>
by zdw mentioning the
<span style="font-family: monospace">finger</span> command, as well as
<a href="https://news.ycombinator.com/item?id=41912287" target="_blank"
>this comment</a
>
by zimpenfish about using
<span style="font-family: monospace">rlogin</span> and
<span style="font-family: monospace">who</span> to determine who's
logged into the network. Well -- the UPL doesn't have desktop computers
available for students to come in and use anymore. Most students just
bring their laptop, as computers aren't necessarily hard to come by
anymore...! We don't have the infrastructure (at least, right now) to
allow students to ssh into our servers, and even if we did, there would
be no way of telling if they were doing it on-premises. As for using
WiFi/DHCP leases to detect devices (as recommended by q3k in
<a href="https://news.ycombinator.com/item?id=41913831" target="_blank"
>this comment</a
>): UPLians use UWNet or eduroam, not the UPL's own WiFi network.
There's also some privacy-related concerns about
snooping/logging/aggregating users, even if anonymized, that highly
dissuades us from implementing anything like that.
</li>
</ul>
</ul>
<ul style="margin-top: 0; margin-bottom: 3px;">
<ul style="margin-top: 0; margin-bottom: 3px">
<li>...just use an online reservation/booking system?</li>
<ul>
<li><a href="https://www.upl.cs.wisc.edu" target="_blank">The UPL</a> isn't a conference room. While there <i>is</i> a list of the coordinators' individual hours, the actual occupancy of the room doesn't necessarily follow that schedule. Sometimes coords miss their hours, and other times <a href="/images/upl-pc/five_am.jpg">the room remains open until 5am.</a> All that to say: it is <b>not</b> possible to simply do a schedule lookup to determine the status of the room, and having preset times for determining occupancy would be inaccurate fairly often. </li>
<li>
<a href="https://www.upl.cs.wisc.edu" target="_blank">The UPL</a> isn't
a conference room. While there <i>is</i> a list of the coordinators'
individual hours, the actual occupancy of the room doesn't necessarily
follow that schedule. Sometimes coords miss their hours, and other times
<a href="/images/upl-pc/five_am.jpg"
>the room remains open until 5am.</a
>
All that to say: it is <b>not</b> possible to simply do a schedule
lookup to determine the status of the room, and having preset times for
determining occupancy would be inaccurate fairly often.
</li>
</ul>
</ul>
<ul style="margin-top: 0; margin-bottom: 3px;">
<li>...use <a href="https://arxiv.org/pdf/2301.00250" target="_blank">WiFi signals</a> to get an accurate number of the people in the room?</li>
<ul style="margin-top: 0; margin-bottom: 3px">
<li>
...use
<a href="https://arxiv.org/pdf/2301.00250" target="_blank"
>WiFi signals</a
>
to get an accurate number of the people in the room?
</li>
<ul>
<li>Uh... wow. That's insanely cool. But it still doesn't expressly convey the <i>intent</i> of the room's use. It would still be neat to have a count for the number of people as <i>well as</i> the status of the room, so I'm definitely stashing this away in my head.</li>
<li>
Uh... wow. That's insanely cool. But it still doesn't expressly convey
the <i>intent</i> of the room's use. It would still be neat to have a
count for the number of people as <i>well as</i> the status of the room,
so I'm definitely stashing this away in my head.
</li>
</ul>
</ul>
</ul>
<p style="margin-bottom: 4px;">Hopefully that can answer some of the most common questions I've gotten about the project. Thanks for reading!</p>
<p style="margin-bottom: 4px">
Hopefully that can answer some of the most common questions I've gotten
about the project. Thanks for reading!
</p>
</div>

I considered using similar components for a door status checker for the UPL -- it wouldn't have been too much effort to buy WiFi enabled ESP32 modules and off-the-shelf door-mountable reed switches. Then, I would have the chips simply send a POST request with their status every time the door was opened or closed.
Expand Down

0 comments on commit 23e3a8f

Please sign in to comment.