Computer, linux, Raspberry Pi

Setup a Network Camera with a Raspberry PI

Leave a comment

In this post I am setting up a network available RTSP video camera on a Raspberry Pi 4 Model B with the latest Debian Buster installed, you will also want another ‘remote’ computer with VLC installed on the same network. I have used a camera module connected via a ribbon cable to the Pi’s video port. The performance is more like what you would see from a security camera rather than some fancy dedicated device. I will be making the configuration with a bash shell using SSH rather than the GUI.
A Camera for the Pi

Installation

If you haven’t already, enable the camera on the Pi with sudo raspi-config and go to Interface Options > Camera. To reduce the load on the computer I also set the Pi to boot to the Console in System Options > Boot / Auto Login and selecting Console, you should also check that at least 256MB of memory is allocated to the GPU in Performance Options > GPU Memory, when complete reboot to set the updated configuration.

Install some prerequisites, you may have some or all of these installed already:

Next add the Video For Linux (V4L) Repository, this adds the drivers for the video on the hardware side

and add the following line to your apt sources with $ sudo nano /etc/apt/sources.list

We can now install the V4L packages:

We now need to install the RTSP server from Git, this server connects to the V4L drivers above:

Configuration

To setup V4L, edit the configuration file $ sudo nano /etc/uv4l/uv4l-raspicam.conf and update the driver options and uncomment the h264 options:

Restart the service once you have saved the file:

Useful screen Resolutions; designation, ratio: width x height:

  • 1080p, 16:9: 1920 x 1080
  • 720p, 16:9: 1280 x 720
  • 720, 4:3: 1296 x 792
  • Standard Definition, 4:3: 640 x 480

I’ve chosen 720p 16:9 widescreen, this gives a good image quality without stressing the Pi 4B. The framerate is a compromise between quality and usability, its unlikely you would want to go above 25fps.

Connecting

If you don’t know it already, find the IP address of your Pi with $ ip a. Look for the address in eth0 or wlan0 depending on if you are using ethernet or wireless. In this case I will be using 192.168.2.33 in the examples.

Start the RTSP server with:

Replace user:password with a suitable username and password. I have assumed that the camera is on /dev/video0, you can test this with:

and look for a line such as Driver version: 5.10.17, if this is 0.0.0 then try a different device, /dev/video1, /dev/video2 etc.

On a successful load it will tell you the connection:

Now on your ‘remote’ computer open VLC and choose Media > Open Network Stream and place the given stream address into the network URL, you will then be asked for the username and password. For me it took less than five seconds for an image to appear, there is a fairly high latency on the image shown, the picture lags by about a second from reality.

Using OpenCV

You can access the stream using Python scripts and OpenCV on your ‘remote’ computer for AI/Machine Learning. For me this is a Debian Linux box with FFmpeg as OpenCV uses this to access the stream.

You can check OpenCV is working with FFmpeg with:

and looking in the output for FFMPEG saying YES:

Links and Sources

history, Sheffield

History: The Watermills of Sheffield

Leave a comment

There are 149 watermills and dams shown on this interactive map, dating from the 1200’s to the 1930’s all using water power to drive tilt hammers, blowers for forgers, grinding wheels for sharpening knifes and scythes as well for the milling of corn and other grains. Each mill has a brief chapter of the places history taken from the book ‘The Water-Mills of Sheffield’ by W.T. Miller, 2nd Issue, published in 1947.

The map has been designed for use with mobile phones, and the GPS can be used to find your location when you are out and about, on a desktop computer the GPS will not work. Click on a dot to see the mill name, and then click on the name for some interesting facts. On your mobile use the menu (top right) to turn GPS on to show your position.

Link: https://www.g7smy.co.uk/history/watermills/

The Watermills of Sheffield
Computer, Jetson, linux, Programming

Jetson Nano Install Notes

Leave a comment

The Jetson Nano Developer kit – B01 is a small computer comprising of an NVIDA Maxwell GPU, Quad-Core ARM Cortex-A57 Processor and 4GB of Memory along with four USB 3 ports, Gigabit Ethernet, HDMI and Display Port output, main storage is on a MicroSD card and there is a variety of selection of expansion available via GPIO, I2C and UART. On the software side NVIDIA provide their JetPack SDK – a customised version of Ubuntu. This development kit has been produced to provide an entry point into Machine Learning, for which I will be using Python programming language. I got my board from Pimoroni

Jetson Nano
Jetson Nano
GPU 128-core Maxwell
CPU Quad-core ARM A57 @ 1.43 GHz
Memory 4 GB 64-bit LPDDR4 25.6 GB/s
Storage microSD (not included)
Video Encode 4K @ 30 | 4x 1080p @ 30 | 9x 720p @ 30 (H.264/H.265)
Video Decode 4K @ 60 | 2x 4K @ 30 | 8x 1080p @ 30 | 18x 720p @ 30 (H.264/H.265)
Camera 2x MIPI CSI-2 DPHY lanes
Connectivity Gigabit Ethernet, M.2 Key E
Display HDMI and display port
USB 4x USB 3.0, USB 2.0 Micro-B
Others GPIO, I2C, I2S, SPI, UART
Mechanical 69 mm x 45 mm, 260-pin edge connector

These notes cover my process of setting one up and links to the documentation, it not intended to repeat those install guides but to provide an install sequence and any additional commentary as needed. I’m going to assume you have a little experience of using the terminal and am familiar with using the bash command line – I’ve no idea how this would be done through the GUI.

jetson nano ports
Jetson Nano ports [source: NVIDIA]

Initial Startup

I followed the instructions for downloading and installing JetPack 4.4 on https://nvidia.com/jetsonnano-start I used a 64GB Class 10, UHS-I, U3, V30 SanDisk card. I formatted the card in a camera before using balenaEtcher to write the JetPack SDK image, this creates a partition of about 16GB on the card formatted to ext4, during installation the volume is resized to fill the card.

jetson nano connections
Jetson Nano connections [source: NVIDIA]

Despite using a good quality USB power supply with an output of 3 Amps at 5 Volts into the Micro USB port the computer would only boot long enough for the NVIDIA logo to appear on screen but after a few seconds the green power LED would go out and it would be off, the same happened when I tried a variety of USB power supplies used. I got round the problem buy using a 5 Amp power supply connected to the barrel jack J25 on the left (centre pin positive) and connecting the jumper J48 located just behind this connector.

Customising the Setup

There are a couple of thigs to do, get a network volume mounted and set the default version of python.

For the network share install samba, some network utilities and the nano text editor:

create a text file: sudo nano /etc/samaba/videoserver with the following:

And set the permissions sudo chmod 600 /etc/samba/videoserver. In this example I have a network share on my server; 192.168.1.30, called video. Create a mount point for the share: sudo mkdir /mnt/video now you need to edit fstab, sudo nano /etc/fstab and add your network connection to the end:

Reload fstab with sudo mount -a and check for any errors. Because of the way that Jetpack boots it does not appear to wait for the network so the share needs to be set to automount and this causes it to only appear in drive listings when accessed. Further reading can be found in this excellent guide to fstab: https://wiki.archlinux.org/index.php/fstab.

Jetpack 4.4 comes with two versions of Python, 2.7 and 3.6, I want it to default to 3.6 and while this is rather out of date I don’t want to go down the hole of upgrading just yet, you will also need to install pip and set pip3 as the default too.

and test its worked:

I did get an error later on, a crash was reported on the desktop when an occasional python 2 script ran. I fixed the error in /usr/sbin/l4t_payload_updater_t210 by changing the first line of the file from !#/user/bin/python to !#/user/bin/python2

Post Install Problems

A recent update occured, so I did the usual sudo apt get update && sudo apt get upgrade but one of the files gave a script error, this turned out to be with nvidia-l4t-bootloader, like so:

after some head scratching, it turns out this script is python2 only. I reverted back to the python 2.7 as the default, and tried again:

Afterwards I set the default back to python 3.6 again.

Setup the Machine Learning

There are three Machine Learning packages, OpenCV for Computer Vison, Tensorflow for machine learning models, and TensorRT – accelerated deep learning networks for image recognition.

OpenCV is already installed in Python:

Tensorflow installation can be found here: https://docs.nvidia.com/deeplearning/frameworks/install-tf-jetson-platform/ I found this to be straight forward but it did keep the computer busy for a while.

TensorRT can be found at: https://github.com/dusty-nv/jetson-inference/blob/master/docs/building-repo-2.md I used the instructions in the Quick Reference section and installed the default models without any errors, I have not yet had the opportunity to test it.

Further Reading

Computer, linux, Raspberry Pi

Copying and Resizing a Raspberry Pi SD Card

The Micro SD Card I use in a Raspberry PI ran out of space so here is how I copied the contents of the 15GB drive to a new 64GB card and resized the partition. I used a separate computer running Debian and as the machine does not have a monitor or keyboard attached this will be being completed through the bash command line using SSH.

1. Making a copy of the SD Card.

Insert the old card into your computer, if the computer attempts to mount the drive then unmount it. We need to find which mount point has been used, do this with the lsblk command:

The device sde matches our SD card. So we will use that. The dd command is used to create the ISO image, I am creating the file in my home directory:

once complete, check that the records match:

Swap the SD cards and check the mount point again:

Now copy the iso image to the new card:

this can take a while, the write speed is rather slow:

and lsblk shows us the new partitions:

2. Resizing the partition

The copied partition is now the same size as the original. If you have space remaining, the new card can be put back in the Pi and use the raspi-config utility, and using the Expand Filesystem option in the Advanced Settings section. However if the drive is completely full you won’t be able to login as there won’t be enough space available for the temporary files created at login, to get round this you can use parted to resize, start with:

if you get the following message:

then quit from parted, if you are using a full size SD Card, check the Write Protect tab on the side of the card and try again, otherwise try:

If problem persists try formatting the new card in a camera, as these have a simple file system, and write the ISO image again.

We need to resize the larger partition /dev/sde2 with the ext4 file system, the smaller is used to boot the Pi and can be ignored. In parted, list the partitions with the print command:

Using resizepart set the new size, I set this to larger than the 15GB but smaller than the unallocated space, this to save me having to accurately work out the remaining space manually:

Now update the boundaries to grow and resize the partition into the freshly allocated space:

Now boot the Pi with the new card, login and use the raspi-config utility then in Advanced Options choose Expand Filesystem and follow the onscreen instructions. Once rebooted you should now be set to fill up your new card.

Links and Sources

Computer, linux, Programming

Extracting GPS data from the GoPro 7

2 Comments

While extracting the telemetry data from the GoPro is reasonably well documented I have found some gaps for getting the extracting utilities installed and when extracting and combining data from multiple files. These notes are for a Debian/Ubuntu installation in a BASH Shell.

Installing the gopro-utils

As I couldn’t find any straightforward instructions for installation, I’ll be going through everything I needed to do to get it working, you may have some of these packages installed already.

Now to get the gopro-utils and install them, I’m placing the source files into my Downloads directory, as well as the GPS data extractor we’ll be adding the other telemetry tools too, this is all a bit long winded.

Extracting the Data

You will need to find which stream in the video recording the data has been saved to, to find this use ffprobe to examine the recording and look for the stream that contains GoPro MET, for example:

You can see that what we are wanting is on stream 3, as far as I can tell this stays the same every time, I don’t know if it is different for other GoPro models.

This bash script extracts the GPS data in GPX format from all the GoPro GX recordings in the directory, other options have been commented out, if you are using Garmin VIRB edit there is also an option for use with that. The script creates two files, one that contains the raw data and another with the desired GPS data, the GPS output file has the same name as the recording, but in lowercase with a .gpx extension.

Merging GPX files

As the GoPro splits recordings into 4GB blocks, when extracting you will get a single GPX file for each recording. Many pages found by Google say that to create a single track from these all you need to do is append the files into one big file. This is wrong, what you end up with is a single file with many short tracks, when what you are after is one long track covering the entire journey. This bash script uses gpsbabel to create single merged file from the extracted GPX data, it creates a file called “gpsoutput.gpx”.

The next stage will be to write a script that combines all these and completes the job in one easy process.

Links and Sources

CNC, Computer, Electronics, Hardware

Using Inkscape for CNC designs

Inkscape is a free vector graphics editor for all major platforms, generally it is aimed at art and design users but it does have an option for generating G-Code for use in your favourite CNC software. While Inkscape doesn’t have many of the functions of proper CAD/CAM software it is an relatively easy place to start for creating basic designs, I have been using it to make boxes out of 3.5mm plywood.

These notes are based around my cheap CNC machine sold as an CNC3018 by a variety of Chinese manufacturers on Amazon and eBay, the included controller is a Woodpecker CNC board (Ardunio clone) I have upgraded to GRBL v1.1 and I am using version 0.92.4 (April 2019) of Inkscape with the included Gcodetools.

CNC Wood Cutting

This post focuses on setting up Inkscape for the CNC machine and producing the g-code from your drawing, it is not intended to be an Inkscape tutorial.

Document Setup

With a new drawing set your Document Size, this should be the same as your CNC bed, in my case this is 300 x 180mm. From the Inkscape menu go to File > Properties and in the Page Tab set the Display Units (millimeters in my case), the Orientation to Landscape and Page Size width: 300 and height: 180. In the Grids Tab set the Grid Units to mm and the Spacing X and Spacing Y to 1.0. Back on your main page, turn the page grid on with: View > Page Grid.

Inkscape Document Properties

By default Inkscape scales the stroke/line width when you resize a shape, to prevent this click the the fourth box from the right in the top icon bar “when scaling objects, scale the stroke width by the same proportion”

Turn Line Scaling off to prevent the line width changing when resizing

You can save this as a template, such as: CNC3018.svg or as the document default with: default.svg by saving the file to your templates directory:
On Linux and OS X: ~/.config/inkscape/templates/
On Windows: C:\Users\<username>\AppData\Roaming\inkscape\templates

The lines you draw will need to be the same width as the bit you are using in the CNC machine. Draw a rectangle, Right mouse click on the rectangle and select Fill and Stroke…. In the Fill Tab click the X – no paint box and on the Stroke Style tab set the width to that of the bit you are using – 1.5mm, subsequent rectangles will be in the same style, other shapes will need to be setup this way too. The colour of your lines should be black, there is some functionality for different colours to represent different depths but I have not yet worked out how to do this.

Setting the line width

Layout Tips for G-Code Routing

Remember to check the dimensions of the cuts, with an outside cut such as the width and height of a box side you need to measure for the inside of your rectangle, for holes in your box measure to the outside edge, Inkscape sets distances to the outside edge.

For positioning holes for switches and the like, I add thin lines 0.1mm thick as guides and make use of the width/height settings as well as the Object > Align and Distribute options. A pair of digital vernier calipers are a great aid to discovering the required sizes. Remember to delete these before G-Code encoding.

Using guidelines for accurate layout

When generating the G-Code each shape will be seen as an individual object, so lets say you want to have two sides of your box cut from a single sheet of plywood, this would be two rectangles abutting each other with a side to be cut overlapping. As it takes four passes to cut each shape 1mm at a time, this means it’ll take six passes down the centre. To fix this select both rectangles and then Path > Combine followed by Path > Difference to make a single object.

Outputting to G-Code

Now that you have completed your drawing, save your work then convert your objects to paths by selecting all objects then Path > Object to Path. You may also want to place your drawing near the bottom left of the document, as this is where the CNC router starts. Now using Gcodetools there are three things you need to do to produce the G-Code file. None of the Gcodetools windows close automatically when apply is clicked, you will need to do that yourself. From the Inkscape menu:

1. Extensions > Gcodetools > Tools Libary…
Select Tools Type: cylinder and click apply In the overlarge green box that appears you will need to set the tool diameter and feed speed.

Gcodetools Tools Library

This can be a bit fiddly as the text can become detached from the box and the settings lost, what seems to work most reliably for me is to change to Text Objects (F8) click on the numbers you want to change and once done go back to Select and Transform (F1). Resize the box afterwards to check that it is still working – if the green box moves but the text does not then Ctrl-Z a few times and try again.

Gcodetools Tools Library – Green background is detached from the text – this won’t work.
setting default used description
Diameter 10 1.5 tool bit diameter in mm
Feed 400 300 speed while cutting through the material in mm/second
Penetration angle 90 90
Penetration feed 100 100 Plunge speed in the material in mm/second
Depth Step 1 1 Depth of cut on each pass in mm

2. Extensions > Gcodetools > Orientation Points
This tells the g-code where to start, normally bottom-left on the CNC Set the following:
– Orientation type: 2-points mode
– Z Surface: 0mm – this is the top of your surface
– Z Depth: -3.4mm – this is the thickness of material to cut, a negative number

With the Orientation Points added.

3. Extensions > Gcodetools > Path to Gcode
This creates the G-code file, in the Preferences Tab set the following:
– File: output filename 
– Directory: output directory
– Z safe height: 5mm – height above the work surface when moving between cuts
The filename once set doesn’t change, an incremental number is appended to the output filename. Click the Path to Gcode Tab before clicking apply (this appears to be a bug).

Your image will be updated to show the g-code routing, give this a visual check to ensure that all objects have been coded and that it looks right, the path to be taken should be in colour and contain arrows showing the direction of the router.

If there are too many arrows or if a line has arrows pointing in different directions then there may be an object underneath, check on your original artwork, in the image with the three circles below you see that A has not been converted to a path with Path > Object to Path, B has a duplicate object underneath and C is correct.

Gcode check; A and B have problems, C is correct.

The generated G-Code does not appear to include the Spindle Motor Start command – So remember to start the spindle manually in your CNC software before running the G-Code – its interesting how easily these bits break with a sideways load. Remember if you are cutting trough rather than engraving, don’t forget to put a layer to sacrifice between whatever you are making and the CNC’s bed, I use 5mm MDF/Fibreboard.

Links and Sources

history, Sheffield

History: Colliery Road is Weird

2 Comments

Colliery Road in Sheffield Brightside is an odd throughway that passes under the Midland Mainline Railway providing a 290m (317 yards) route between the junction of Brightside Lane and Weedon Street to Holywell Road. It is a narrow single track road and the the Brightside Lane end has been cut into the hillside to accommodate the railway junction and sidings above. Nowadays access is limited to pedestrians and cyclists as at 2.32m (7′ 6″) high the bridges are far too low to be of any use for general traffic, the road was closed to motor vehicles in around 2016 and is now part of the Lower Don Valley Cycle Route1. In this short essay I am looking a little into its history, the modifications that have changed the road to accommodate the railway and to speculate why it exists at all.

The Bridges of Colliery Road

Two rail bridges run over the road, one for the Midland Mainline, originally the Sheffield to Rotherham Railway, constructed in 1838/39, this bridge is referred to in Drake’s Road Book published in 1840 where he says “We now cross a private road by a beautiful stone bridge – one which is infact the only ornamented bridge upon the line, and also the only one that is graced with parapets”2. The other is now disused, it was part of Brightside Junction giving access to the various steel works as well as a line towards Catcliffe and the south as part of the Sheffield District Railway. As well as the two we now see, sometime between 1880 and 1955 there was also a third single track bridge on the north western side used for the gas and chemical works.

Colliery Road, Brightside [Ordnance Survey]
The 1854 map (Yorkshire 289) shows a footpath or track approximately following the route of the road today, leading from Brightside Lane under the railway and up to The Grange farm at the end of Wincobank Lane, on the 1892 map this is shown as a footpath, it looks to have been around a 620m (678 yards) walk from the farm to Brightside village.

Colliery Road highlighted in yellow

By 1892 Brightside Junction had been added to the railway providing access to the steel works in that part of the Don Valley, at that time Colliery Road was moved to start from near near Naseby Street and run alongside the south-eastern side of the tracks for a short while before ducking under the line, this was to accommodate the embankment built for the junction and the goods line into Brightside Steel works. It wasn’t until 1896 when construction started on the Sheffield District Railway that the road was moved to go between the new railway and the existing goods line, this is when the roads route was fixed to that we see today.

Ordnance Survey Maps – 1854 to 1949 [National Library of Scotland]

In the 1930’s on the north-eastern side of the tracks Holywell Road was extended from Upwell Street alongside the Gas Works to connect to Limpsfield Road and replacing Burslem Street as part of an expansion of housing into the area. Before then this part of Wincobank was mostly farmland with some industry alongside the railway, most notably the 1906 map (CCLXXXIX.SW) shows the Grimesthorpe Gas Works taking up a large amount of land between Upwell Street and Colliery Road. Looking at the 1926 aerial photo below, the exit of Colliery Road from under the railway can be seen on the far left of the image. Today this land is used by a mail order distribution centre as well as some smaller wholesalers and trade suppliers.

The Grimesthorpe Gas Works, Sheffield, 1926 [Britain From Above]

Brightside Colliery was also situated on the north-eastern side, the mine was worked from 1855 to 1886 and the shafts were located at the end of Colliery Road, the colliery was owned by Unwin & Shaw Co. and the pit had two shafts, No. 1 from 1855 to 1871 and No.2 from 1868 to 1886, they worked the Parkgate seam3. From 1883 until around 1955 the site was occupied by the Brightside Chemical Works which was a part of the gas works dealing with sulphur produced from gas manufacture [Sheffield History]. Nowadays the land is used by a self storage facility and a motor vehicle repair company.

Extra width has been added to the mainline bridge over time

The purpose of Colliery Road?

The road must of had a high cost of construction, the changes that needed to be made to have it run under the railway including cutting it into the hillside at the Brightside Lane seems a lot of effort has gone into maintaining a private a road that gives such poor access.

A reason for this could be that the colliery supplied coal to the works and foundries on the south-eastern side of the railway, this would justify a bridge for the Sheffield to Rotherham Railway but by the time Brightside Junction was built the mine had been closed for around five years so why go to all the trouble to modify the road? Or is Colliery Road a legacy of the landowner insisting they have access from their farm at The Grange down to Brightside Lane? I can imagine a track existing before the railway and it being kept as a right of way as a route to Brightside village and the bridge across the Don which had been there in one form or another since before 16504.

Comparing access for the colliery 1854 and 1892

By comparing the 1854 and 1892 maps above we can see similarities between the buildings being shown, the map also shows there is no roadway from the colliery site into Brightside as a field is in the way. The coming of the railway would have cut access off from Naseby Street and maybe the landowner did not want to sell part of his field, or set the price too high, to permit access to the nearest bridge over the line on Jenkin Road and forcing further development of Colliery Road, the bridge and its parapets.

More of my Colliery Road photos can be found on flickr

Dates of Interest

1838/39 Sheffield to Rotherham Railway constructed (Midland Mainline)
1838 Brightside Station opens
1855-1871 Brightside Colliery – Shaft No.1 in operation
1868-1886 Brightside Colliery – Shaft No.2 in operation
c.1855-c.1955 Grimesthorpe Gas Works in operation
1883-c.1955 Brightside Chemical works – part of Grimesthorpe Gas Works
c.1890 Brightside Junction added – for the steel works
1896 Sheffield District Railway construction starts

Links and Sources

My photos of Colliery Road on flickr

  1. Sheffield City Region Combined Authority, Transport Committee, 26 Sept 2016. p. 17
  2. Drake’s Road Book of the Sheffield and Rotherham Railway, London 1840, p.33
  3. The Coal Mining Industry of Sheffield and North Derbyshire, Ken Wain, 2014: ISBN-13: 978-1445639635
  4. Hallamshire. The History and Topography of the Parish of Sheffield. Joseph Hunter, London 1819, p.227

Ordnance Survey maps at the National Library of Scotland: