Pi-based LED Sign

I took some time this Thanksgiving break to work with Raspberry Pi and a flexible LED display. I love the Pi, as it is a familiar environment, but with opportunities to work with different hardware.

Working with some instructions online, I was able to get the board up and running quickly.

Wired up via breadboard
Lighting up in sequence with random colors.
Note that the way the LEDs fill in sequentially, which does not map to an XY grid.

You’ll notice above that the LEDs are filled in top-down, then bottom-up. This is because the LEDs are arranged in series, producing a serpentine pattern. In order to display images easily, I’ll need to map those IDs to an XY grid. I’ll get to that more in the Software section below.

Hardware and Mounting

After thinking about how I could mount the flexible, I settled on a flat, length-wise orientation. I initially thought about mounting it curved in a half-circle, so that part of the display could be viewable from multiple orientations. I knew I wanted to mount the sign on the wall, but having it stick out away from the wall was not ideal for my home. I then looked into mounting the display in a way that only the display (and power cord) is visible when mounted to the wall.

After finding some great resources online to work with WS2182B LEDs and Raspberry Pis, I first started with a breadboard to make sure I had the wiring correct. I then started to connect up to a Raspberry Pi Zero, but I realized (thankfully before soldering) that my Pi Zero was a 1.3, not a Pi Zero W. For now, I’ll use a Model 3B+ until a Zero W is delivered. In the interim, I soldered the sign’s three leads to a 5 pin molex connector so that I can simply slide it on to the GPIO pins.

Looking at the backside of the display, the wiring was soldered in at approximately 5/6cm from the other edge. I had some birch plywood around from a previous project. I ripped it down to 4cm wide and then cut it into three pieces, one 30cm and two 6cm. I then glued the two 6cm pieces to the 30cm piece to the longer piece. I then used some heavy double-sided tape to affix the display to the wood. As I wouldn’t be directly soldering to the Zero W at this point, the GPIO pins are closely grouped, so I wired up the display connector to a small molex connector and connected the board to the LEDs.

Software

Code accompanying this write-up is available here: https://github.com/johnjreiser/pisign

One issue I encountered is that the LEDs are identified in series. I needed to make a translation between a 32-wide by 8-high matrix to
an individual LED’s ID. I put together matrix.py to build the array that would map to the LEDs. After building the matrix, I can then map a list of lists in Python (my XY grid) to the display.

Displaying a rainbow sequentially 
(x then y)
Now the LEDs are lighting left-to-right, top-to-bottom instead of serpentine from bottom-right.

Using the rpi_ws281x module, I put together a few demos of what can be done. I’m sure that they are not the most efficient or elegant code, but they will hopefully give you some ideas for your own projects. Feel free to open an issue on GitHub or submit a pull request if you make changes to the demos.

One item in the repo is a quick and dirty web application using Flask. It allows someone to view what’s displayed on the LEDs and manipulate it. My family has gotten pretty adept at making some cool designs with the sign while I’ve been away:

This was a fun one-day project and it spurred a lot of other ideas. Pushing alerts from the home automation software out to the occupants is one idea. A ticker with time and weather is one common offering. What would you put on your sign?

Posted in Uncategorized | Tagged , | Leave a comment

Meandering Meetings and Open Data

I love meetings. When the right people are in the room, with a clear agenda and desire to work together towards a common goal, magic can happen. Ideas are put out there and refined as a group, and an action plan is developed, with next steps and clear accountability. They end 5 minutes early and everyone leaves ready to tackle big problems.


Of course, this rarely happens. We spend a ridiculous amount of time in meetings that meander through a rough (or non-existent) agenda, no direction given nor accountability assigned, and then we depart to the next meeting feeling demoralized.

I am an employee at a public University. My colleagues and I have a responsibility to not spend taxpayer money and students’ tuition and fees inappropriately. I wanted a way to display a rough estimate of the cost of a meeting, to help remind us to keep focus and make progress.

New Jersey, with its strong freedom of information law (the Open Public Records Act, or OPRA) has salary information for all State Agency and Authority employees that is available on YourMoney.nj.gov. Also available are records containing information on those non-State Government employees contributing to public retirement systems, such as local government employees and educators. I have compiled this information and built a clock that will track time in dollars, based on the members in the meeting.

CostClock.com will allow you track a meetings cost based on the participants in the meeting. You can quickly add individuals from a list of public employees. If you are tracking a meeting of individuals not in the list of public employees, simply click “Add New Participant” and enter the person’s name (or title) along with annual salary or hourly rate.

As it is a clock, the amount and time are displayed prominently at the top of the page. I also wanted to reduce the navel-gazing that may be introduced with showing annual salaries, so only the total cost of the participants is displayed, per minute and per hour.

And while I’ve been mentioning the dollar amount as a “cost” costs aren’t necessarily a negative concept. An hour long meeting with several participants might yield a large dollar amount, but if it was productive and brings focus to the group, then that “cost” becomes a well-made investment.

I’m hoping that this can get some use within New Jersey and that others may adapt it and use it for their own meetings. I’m going to follow up this post with a technical explanation of my development and deployment workflow. If you’d like a copy of the code, head over to Github where you can use the code as you see fit.

Posted in Data, Government, OPRA, Privacy, Technology, Visualization | Tagged , , , | Comments Off on Meandering Meetings and Open Data

Presentations at MAC URISA 2018

I’m currently down at the MAC URISA 2018 conference in Atlantic City. This year, I conducted a full-day workshop on Spatial SQL. I also presented on using Docker to explore open source GIS offerings. If you’re still around at the conference and interested in what I presented above, let me know and I’d be happy to discuss! If you can’t make it to MAC URISA and are curious about databases and software deployment strategies, feel free to comment here or message me on Twitter.

Posted in Conferences, PostgreSQL, Teaching, Technology, Tools and Scripts | Comments Off on Presentations at MAC URISA 2018

What I’m Reading – July 2018

Here are a few articles I’ve read over the last month that you might find interesting.

“I keep it because I have not heard a voice I like better,” he once said, “and because I have identified with it.” He could change to a smoother voice, but then he wouldn’t sound like himself.

The Quest to Save Stephen Hawking’s Voice
Recreating Hawking’s hardware voice in software.


… of a production run of 10,000 transistors for example maybe 2-3 percent were “defective” and he purchased these rejected parts which became the source of the “sizzling” sound in the TR-808.

The mysterious heart of the Roland TR-808 drum machine
Recreating “defective” transistors to preserve an iconic sound.


When reading the copy, the line where the two little pieces of paper met looked like a stroke and was added to the character by mistake.

A Spectre is Haunting Unicode
IRL bug makes its way into a digital specification, living on indefinitely.


While Gargac did have a small sign on his car that theoretically provided “consent” to “recording,” many passengers did not notice it, and it did not indicate at all that people were being livestreamed. He openly advertised his livestreaming on Twitter.

Uber, Lyft driver booted after newspaper reveals he was livestreaming passengers
There’s a big difference between what’s legal and what’s ethical.

 

Posted in In the News, Privacy, Technology | Tagged , , , , | Comments Off on What I’m Reading – July 2018

Leveraging the Power of Spatial Databases

I’m out at the ESRI DevSummit this year, and tonight I presented the following talk on using SQL to perform spatial analysis tasks. The presentation slides are included here.

I tried to squeeze in several examples, including mapping 2,199 farms for the State of New Jersey, studying habitat change, and monitoring use of a crowdsourcing application.

I also use SQL extensively in NJParcels.com on processing the available public data, as well as to produce the data products available on the site. You might also find my LearnSpatialSQL.com site helpful if you want to learn more about using only SQL for performing GIS tasks.

Posted in Conferences, ESRI, Events, PostgreSQL, Technology, Tools and Scripts | Comments Off on Leveraging the Power of Spatial Databases

Location-based Search for Property Information

This is third in a series of posts on a side project of mine, NJ Parcels. Read the first and the second parts.

In the summer of 2015, we decided to get takeout from a local Greek place. We were apparently not the only folks with the same idea. When I arrived, there was a line outside and all the seats in this small place were occupied. They apologized and told me my order would take about 20 more minutes. So I hopped back in the car and took a short drive around the neighborhoods south of Glassboro.

After passing several houses either for sale or just plain interesting from the curb, I realized that I needed an easy way to search my site based on my phone’s current location. The county roads in South Jersey all have variable names, often changing mile by mile. For example, Woodbury Road becomes “Woodbury-Glassboro Road” somewhere between Glassboro and Woodbury, and then eventually changes to Glassboro Road in Woodbury. The numbering resets at one point as well, which makes searching for a specific house difficult at times.

After dinner, I wrote down my thoughts on what I would need to put a location-based search together. That weekend in June, I implemented it as “Properties Around Me” and I included a cookie to conduct some basic tracking of how the search was used. The cookie has an expiration date of two weeks from the last request and it is only set if the user agrees to share his or her location. It’s a somewhat simple unique identifier, an MD5 hash of several user attributes. The two week expiration will allow me to link users that access the page at least once every two weeks and see their usage patterns. If you use the search less frequently (or use different devices or delete/block cookies) you’ll register as a new user with no history. Continue reading

Posted in NJ Parcels, Privacy, Technology | Tagged , , , | Comments Off on Location-based Search for Property Information

Open Data and Personal Privacy

This is the second in a series of articles regarding a side project of mine, NJParcels.com. Read the first article here. Read part three here.

Once GlassboroMap and NJ Parcels started to get considerable traffic, I began receiving emails asking for the owner information to be removed from the site. While this was public data and, I imagine, has always been available, the easy ability for a search by name to bring up this information caused some individuals concern. Having known about assessment and parcel data for a while, I was initially surprised at the level of concern some individuals had about the site.

Of the emails received, most were courteous.

redactemail01

Others were angry.

Continue reading

Posted in Data, NJ Parcels, OPRA, Privacy, Technology | Tagged , , , , | 3 Comments

Update on a side project

This is the first in a series of posts about a side project I’ve been developing and maintaining for the past few years. This introduction will be accompanied by subsequent posts dealing with the technical, commitment, and emotional issues associated with running a project on the side. Read part two here.

Back in 2009, I moved down to Glassboro, New Jersey to be close to work. I had spent that summer contributing to OpenStreetMap and wanted to spend more time working with local data. In searching for a new home, I wanted to use property information and other points-of-interest data to get (re)acclimated to the area.

At the time, I recall web GIS falling into two categories: Google Map mashup or ArcIMS/ArcGIS Server desktop-on-the-web GIS. Typically, the map mashups were tailored to a specific region or data set. The clunky web viewers set up by county and regional governments had valuable data, but an absolutely abysmal user experience.

So, in the Summer of 2009, I planned on making a web site with some local GIS data for Gloucester County. I originally planned to pull in information from OSM and other sources to paint a picture of the community, highlighting what would be most useful to someone considering moving to or visiting the area. Setting it up on Dreamhost shared hosting, I eventually bought GlassboroMap.com that fall and began building a web map, starting with property information for the County.

Screenshot of GlassboroMap.com

GlassboroMap.com – perpetually in beta

Continue reading

Posted in NJ Parcels, Technology | Tagged , , , , , | 6 Comments

Optimize Your Indexes and Selections

I have been working on a project that is driven by a series of functions written in PL/pgsql. The functions look at New Jersey’s land use data over several different time periods and selects out certain areas if they have the right mix of land use types and they meet an overall size threshold. The project requires the same selection process with approximately 60 different selection criteria and it operates over 5 different time periods of land use data. An efficient, programmatic approach to the problem is absolutely necessary.

Last night, I finished writing up a function to perform another set of steps in this selection process. As land use changes over time, the function creates 5 different sets of polygons based on the five time periods (1986, 1995, 2002, 2007, 2012) to represent the size thresholds the land use data must meet. After selecting the non-compliant areas, the function then marks the underlying land use as “not selected”, using a bitmask which represents selection/rejection for each time period. For example, I need to only include land use polygons where their contiguous area is greater than 50 acres. Individual polygons are going to be smaller than 50 acres, but should be included if they are part of a contiguous fabric of polygons that exceeds that size.

Before leaving work, I ran the function against some test data and when I arrived this morning I found the query was still running. It took 7 hours to complete the one step!

NOTICE: Generating patch polygons...
NOTICE: Applying constraint to patch polygons...
NOTICE: Applying bitmask to polygons that fail to meet size threshold...
NOTICE: Patch Size Requirement Constraint Complete.
NOTICE: Query execution time: 07:00:54.204277
query result with 1 row discarded.

The function that was causing the delay was that final update to the base data.

EXECUTE $$
 UPDATE $$||tblbase||$$ b 
    SET transition_mask = transition_mask & '11110'::bit(5)
   FROM $$||tblpatch||$$ patch 
  WHERE patch.period = 1986 AND patch.size_threshold = 0 
    AND ST_Intersects(patch.shape, b.shape)
;$$;

Keep in mind this is one of 5 functions for each of the time periods. What was causing it to run so slow? Well, the land use data is a conflation of all five time periods, weighing in at about 2.5 million polygons. I calculated a spatial index on both tables, but clearly that was not enough. Luckily, this data also has a region identifier. The land use data was split up into about 8,000 regions, each with its own unique region identifier. As the “patch” data was generated from the same land use, we could include the region identifier in that as well to help optimize the query. It was safe to use the region id, as no two polygons with different region ids would touch and be contiguous.

I modified the “patch” creation portion of the function to include the region identifier and then modified the function, like so:

EXECUTE $$
 UPDATE $$||tblbase||$$ b 
    SET transition_mask = transition_mask & '11110'::bit(5)
   FROM $$||tblpatch||$$ patch 
  WHERE patch.period = 1986 AND patch.size_threshold = 0 
    AND patch.newregionid = b.newregionid
    AND patch.shape && b.shape AND ST_Intersects(patch.shape, b.shape)
;$$;

I also realized I was missing a step I used elsewhere in this project. The double-ampersand (&&) operator performs a minimum-bounding rectangle comparison between the two geometries. This operation can also be performed using only the spatial index. So I added it as well in the hopes that it would improve the results. Running the function again, the process now only takes about 15 minutes, which is approximately 28 times faster than before.

Here’s the EXPLAIN on the first version of the function:

Update on baseland b  (cost=16.47..20972348.39 rows=4192474 width=526)
  ->  Nested Loop  (cost=16.47..20972348.39 rows=4192474 width=526)
    ->  Seq Scan on curr_patch patch  (cost=0.00..48093.89 rows=49087 width=1546)
         Filter: ((period = 1986) AND (size_threshold = 0))
    ->  Bitmap Heap Scan on baseland b  (cost=16.47..425.72 rows=34 width=520)
         Recheck Cond: (patch.shape && shape)
         Filter: _st_intersects(patch.shape, shape)
         ->  Bitmap Index Scan on sidx_baseland  (cost=0.00..16.46 rows=101 width=0)
              Index Cond: (patch.shape && shape)

And with the changes to the where clause:

Update on baseland b  (cost=0.41..458738.64 rows=1 width=526)
 ->  Nested Loop  (cost=0.41..458738.64 rows=1 width=526)
   ->  Seq Scan on curr_patch patch  (cost=0.00..48093.89 rows=49087 width=1553)
        Filter: ((period = 1986) AND (size_threshold = 0))
   ->  Index Scan using sidx_baseland on baseland b  (cost=0.41..8.36 rows=1 width=520)
        Index Cond: ((patch.shape && shape) AND (patch.shape && shape))
        Filter: ((patch.newregionid = (newregionid)::text) AND _st_intersects(patch.shape, shape))

No longer are we relying on a Heap Scan, instead we use the spatial index, with both the text comparison and the more thorough ST_Intersects() to validate the results returned from the Index Scan. I’m still amazed that the query planner (this is on PostgreSQL 9.3.9 with PostGIS 2.1.7) doesn’t use the MBR/Index Scan when comparing features using ST_Intersects. It’s always good to run EXPLAIN against your queries and test them in isolation. Just because you generated an index, doesn’t mean the database is actually using it. And it might seem redundant to type a.shape && b.shape AND ST_Intersects(a.shape, b.shape), but I’m happy to do it if it saves hours of time waiting for tasks to complete.

Posted in PostgreSQL, Technology, Tools and Scripts | Tagged , , , , , , | Comments Off on Optimize Your Indexes and Selections

Using Tableau to visualize land use change

Map of Land Use Categories by County. From the Tableau dashboard.

Map of Land Use Categories by County. From the Tableau dashboard.

I was home sick on Wednesday, so while I was on the couch I decided to dive into Tableau Public, a free desktop visualization tool. At IERP, we use Tableau for some of our public dashboards. Other than some minor playing around, I really did not use tool too much prior to this. It’s really great for working with complex data and seeing results quickly. Seeing that the 2012 Land Use data was released a few weeks ago, I wanted to see if I could bring it into PostgreSQL and produce some graphics of how the land use has changed over time.

View the Tableau Dashboard, the code to reproduce the data, and read on.

Continue reading

Posted in Data, Planning, PostgreSQL, Technology, Tools and Scripts, Visualization | Tagged , , , , , , | 1 Comment