My first 3D printable model!

I recently purchased an RX5808 Pro Diversity FPV receiver and wanted to mount it on my FPV box-style goggles in a manner that didn't just involve a metric ton of double-sided tape.

I've been interested in 3D printing for a long time, but the cost for a "turnkey" 3D printer has remained just out of reach. On top of this I hadn't done any 3D modelling since my Year 12 CAD class over fifteen years ago.

Thankfully, that first problem can be resolved by places like Shapeways and the second by a truly awesome program called OpenSCAD - a programming-driven 3D modeller.

A day's furious measuring and coding, followed by two weeks delivery time from the US of A and look what arrived:

Even better, it works:

I've tweaked the control slot a touch since these photos, so if you just want the STL file, it's in my drones repository, or you can support me by purchasing it from my Shapeways shop.

Installing OpenCV 3.3.1 with Python 3.5 bindings on a Raspberry Pi Zero running Raspbian Stretch

I've been experimenting with some basic computer vision concepts on a Raspberry Pi as part of a small mobile robot project. Fundamental to this was getting OpenCV compiled and set up in Python, a process that took 5 or 6 tries over the majority of my Christmas/New Year break. Hopefully this guide will save you some proportion of that time when you have a go at it...

Assumptions
I'm working off the assumption that you are using Python 3.5 and are working within an activated virtual environment for your project. I am also going to use "~/myproject" as a stand in for your specific project path. This means you will have already done something like:

	cd ~/myproject
	python3 -m venv venv
	source venv/bin/activate

view raw assumptions.sh hosted with ❤ by GitHub

You need to stay in that activated virtual environment for the duration of this guide.
 
Caveats
This install will be missing ffmpeg due to a bug in the current version.

Virtual memory
I needed to add more virtual memory to get this to work, using the following steps:

Change swap size to 1 GB temporarily by setting CONF_SWAPSIZE to 1024 in:

/etc/dphys-swapfile

Then running:

sudo /etc/init.d/dphys-swapfile stop
sudo /etc/init.d/dphys-swapfile start

view raw Virtual Mem.md hosted with ❤ by GitHub

Installation
Now, the fun part. These are the steps you need to run, in order. The 'make' step will take 6-ish hours so is a good candidate for leaving overnight.

	pip install numpy==1.13.3
	sudo apt-get -y update
	sudo apt-get -y upgrade
	sudo apt-get -y install build-essential cmake pkg-config
	sudo apt-get -y install libjpeg-dev libtiff5-dev libjasper-dev libpng12-dev
	sudo apt-get -y install libavcodec-dev libavformat-dev libswscale-dev libv4l-dev
	sudo apt-get -y install libxvidcore-dev libx264-dev
	sudo apt-get -y install libgtk2.0-dev libgtk-3-dev
	sudo apt-get -y install libatlas-base-dev gfortran
	sudo apt-get -y install python2.7-dev python3-dev
	cd ~
	wget -O opencv.zip https://github.com/opencv/opencv/archive/3.3.1.zip
	unzip opencv.zip
	wget -O opencv_contrib.zip https://github.com/opencv/opencv_contrib/archive/3.3.1.zip
	unzip opencv_contrib.zip
	cd opencv-3.3.1/
	mkdir build
	cd build/
	cmake -D CMAKE_BUILD_TYPE=RELEASE \
	-D CMAKE_INSTALL_PREFIX=/usr/local \
	-D OPENCV_EXTRA_MODULES_PATH=~/opencv_contrib-3.3.1/modules \
	-D BUILD_opencv_java=OFF \
	-D BUILD_opencv_python2=OFF \
	-D BUILD_opencv_python3=ON \
	-D PYTHON_DEFAULT_EXECUTABLE=$(which python3) \
	-D INSTALL_C_EXAMPLES=OFF \
	-D INSTALL_PYTHON_EXAMPLES=OFF \
	-D BUILD_EXAMPLES=OFF \
	-D BUILD_TESTS=OFF \
	-D WITH_FFMPEG=OFF \
	-D BUILD_PERF_TESTS=OFF ..
	make
	sudo make install
	sudo ldconfig
	ln -s /usr/local/lib/python3.5/site-packages/cv2.cpython-35m-arm-linux-gnueabihf.so \
	~/myproject/venv/lib/python3.5/site-packages/cv2.so

view raw Install.sh hosted with ❤ by GitHub

Virtual memory - again
It is important that you undo the virtual memory change above or you will risk corrupting you SD card. Just revert the value to 100 and start and stop the service.

From here you should be able to import 'cv2' from within the virtual environment's Python, and get on with actual development :)

This guide couldn't have been made without the efforts of:

• https://www.pyimagesearch.com/2017/09/04/raspbian-stretch-install-opencv-3-python-on-your-raspberry-pi/
• https://github.com/ys7yoo/PiOpenCV
• https://github.com/opencv/opencv/issues/10080#issuecomment-343976885

NERFrame build write-up

My NERFrame FPV drone has generated enough interest that I've decided to do a build write-up - so without further ado here it is!

Drones drones drones

I've got a new hobby! A couple of months ago I purchased my first FPV drone, the entry level Eachine QX95.

This glorious little machine came with a Betaflight-compatible flight controller, and I've since learnt how to build the ARM-targeted firmware (a first for me) after adding a software failsafe.

I'd played with an FPV camera on a basic toy drone before but this is my first foray into a FPV-specific drone. So far I've been learning fairly quickly



And then I pranged it.



Which became the perfect opportunity to build my first drone "frame", out of old NERF rounds and hot glue :)

The aptly-named Pink Rocket flies suspiciously well, though it's a touch docile compared to the original due to all the wind resistance from the cross-sectional area. It'll certainly do until the replacement frame arrives.

Like most of my hobbies, there's a GitHub repo.

Temporarily switching Wi-Fi networks from the command line on a Raspberry Pi

I recently had a scenario with my Raspberry Pi (Zero, running Raspbian) where I needed to temporarily switch away from the primary Wi-Fi connection to another access point (to drive a camera) then reconnect to my original network. Turns out wpa_cli can do the trick:

	#!/bin/bash
	# Worked example of temporarily switching from a configured network in
	# wpa_supplicant.conf to a new network to perform an operation, then returning
	# to the original network. An example of this configuration would
	# be:
	# https://learn.adafruit.com/pi-wifi-radio/raspberry-pi-setup-1-of-3#configure-wireless-networking
	#
	# Assumes that running `wpa_cli list_networks` will return a single network at
	# index 0, the one you want to always be connected to at boot up.
	#
	# You'll need to replace [SSID] and [PASSWORD] (including square brackets)
	# with appropriate values, you must keep the single and double inverted
	# commas though.
	#
	# The `wpa_cli add_network` will add a new network every time, but we ignore
	# this and just use the first one after the pre-configured one at index 0.
	# Connect to new network
	sudo wpa_cli add_network
	sudo wpa_cli set_network 1 ssid '"[SSID]"'
	sudo wpa_cli set_network 1 psk '"[PASSWORD]"'
	sudo wpa_cli select_network 1
	# Do stuff on that network
	# ...
	# Return to the original network
	sudo wpa_cli select_network 0

view raw switch_network.sh hosted with ❤ by GitHub

The best thing about this approach is that it is non-persistent, so if it all goes horrible wrong you get your original networking configuration back upon reboot.

Open sourcing all the things

A while ago I was blogging about GenghisIO - a project with the goal of removing all the mess involved in programming a small range of real-world robots - the IDE, drivers, cables etc etc.

The project eventually stalled due to increasing complexity and decreasing resources (time...) but quite a few neat problems were solved along the way - so I have open sourced it!

If you're interested in using Google App Engine applications, Python Flask with WebSockets, empythoned, Python QR code generation, Android WebViews and JavaScript -> Bluetooth communication then go and have a look.

I'd love to hear from you below if you find something useful in there :)

Wireless headless Raspberry Pi Zero with APC220 radios

As a bit of a weekend project and as a way to (finally) use the really cool APC220 Radio Communication modules that I bought ages ago I've put together a wireless and "headless" setup for my Raspberry Pi Zero. This allows me to have a remote-controllable Pi Zero placed anywhere within 1 km (line of sight) without wires! I'm hoping to use this as a robot backpack initially, probably for the Jumping Sumo.

I won't go into technical detail as the build was simply connecting the power and TX/RX from the Pi UART to the APC220 module (I've used protoboard to make a "hat") and telling the Pi to use ttyAMA0 for the console in /boot/cmdline.txt.

If you have any specific how-to questions, let me know in the comments. In the meantime enjoy the photos :)

The top of the hat

The bottom of the hat and Pi header

At the local end I am using an USB-TTL adapter and Putty to connect to the Pi.

It's all in the (domain) name

After taking a crash course in CNAME entries and CloudFlare's free SSL I've moved dnstwister to a shiny new domain:

dnstwister.report

Using CloudFlare has restored 99% of the performance-helping caching that the old Google App Engine site had - the only negative from the migration.

The DNS entries are still propagating at the time of posting so if you have any issues you can still use the old domain.

DNSTwister is now on Heroku

Well, the title says it all. I've moved dnstwister to Heroku.

It didn't take a lot of work and the benefits are huge - the killer for me was the ability to control exactly how much the application scaled up cost-wise. The costing for hobbyist use on Google App Engine was, in my opinion, completely unmanageable. You couldn't limit the performance of your app under load, the only option was to just pay more and more until you covered the quota exhaustion. With Heroku if I get a surge in interest it'll just get a bit slower, and I can choose to scale up at that point as necessary.

I also like deploying through pushing a branch to GitHub and the whole admin dashboard interface is really slick.

Bookmarked report links will automatically redirect from the old domain so you shouldn't feel a thing and the performance is right on par with Google App Engine.

Enjoy!

Driving a DFRobot 3-Wire LED module from a Raspberry Pi

Here's some very simple test code that you can use to drive one of these LED modules from a Raspberry Pi:

	# For Raspberry Pi.
	# Code based on: http://www.dfrobot.com/wiki/index.php?title=SPI_LED_Module_(SKU:DFR0090)#Sample_Code
	import re
	import spidev
	# 0-9
	NUMS = (0xc0,0xf9,0xa4,0xb0,0x99,0x92,0x82,0xf8,0x80,0x90)
	# a-z
	CHARS = (0xA0,0x83,0xa7,0xa1,0x86,0x8e,0xc2,0x8b,0xe6,0xe1,
	0x89,0xc7,0xaa,0xc8,0xa3,0x8c,0x98,0xce,0x9b,0x87,
	0xc1,0xe3,0xd5,0xb6,0x91,0xb8)
	class LED(object):
	def __init__(self):
	self._spi = spidev.SpiDev()
	self._spi.open(0, 0)
	def send(self, chars, clear=True):
	# Clear
	if clear:
	self._spi.xfer2([0xff]*8)
	# Expects string of [a-z0-9. ]
	out = []
	for c in chars.lower():
	if re.match('[0-9]', c) is not None:
	idx = int(c)
	out.append(NUMS[idx])
	elif re.match('[a-z]', c) is not None:
	idx = ord(c) - 97
	out.append(CHARS[idx])
	elif c == '.':
	# Try to efficiently include the '.'
	# using AND with previous char
	out[-1] = out[-1] & 0x7f
	elif c == ' ':
	out.append(0xff)
	self._spi.xfer2(out[::-1])
	if __name__ == '__main__':
	led = LED()
	led.send('Raspbery')

view raw dfrobot lcd.py hosted with ❤ by GitHub

(The misspelling was intentional, I figured "Raspbery" looked better than "Raspberr" :D)