Her last weetabix

Abandoned. The signs there was no point in saving her, I found her alone in the empty ward, no other patients – empty beds. A bowl of lukewarm Weetabix sits next to her bed, sodden beige slime.

I nudge her.

She responds, unaware of who I am and I lift her head, to feed her the beige food.

I am dying with her, she accepts a little of the food and I scrape some more back in her mouth. She raises a faint polite smile, enforced into her by her French upbringing.

I realise now she has not long to go, years of trying to jolt her into a proper conversation, this is no different to recent years of our inability to talk.

I still say goodbye, see you later. She plops back on the pillow.

Next time she would be totally unconscious, the tumour blocking her lungs she will fall back into a drowned coma.

She used to tell me how she chopped off frostbitten toes of Athenians returning from frozen Greek mountains having battled Italian soldiers.

My grandfather, who summarised Italians as ‘marionettes’.

I used to wonder how the Italian toes managed in the cold.

This was a cold I could not thaw.

I was shocked at how empty I felt, losing the one person who gave me life and fed me. Waves of guilt as I walked back outside onto Fulham Palace Road.

Twelve years later, I am gently excavating the remains of Josette, brushing away crumbs of damp earth, deep in her grave, to reveal memories,  albeit painful.


New Control board for ODrive Robotics

Success with my ODrive board, replaced due to a fault. Video shows them working independently – not at full power yet!

Test of ODrive motors and Board

Motors and Prototyping

After some experimentation with tiny N20 5V DC motors, the same type found in MG905 servos – I realise that they run just too fast. For the DC motor to work I have used a small 2N7000 / 2N7002 / NDS7002A — N-Channel Mosfet.

I will use a few of these DC motors in the forthcoming installation for the final project, however, because the RPM is too fast,  I will be converting some of the servos I have into continuous and much slower, rotating servos.

This is achieved by cutting the physical obstruction on the main gear and disconnecting the potentiometer inside, replacing with  2 x 2.2K ohm resistors as voltage divider. I have also tested the more reliable MG906 metal gear servo with the PCA9685 16 channel controller board with an Arduino Uno. I drive the board with a 10A 5V fanless power supply with my own printed enclosure holding a dual pole 30A 240V switch.

I will now be looking at converting as many of the plastic geared servos as possible using surface mount 2.2K ohm resistors (+/- 1% accurate), on order from RS Components ; 100 for 70p!

Heres a short clip of 3 motors in action:

Its been a while…

Hello – just to let you know what I am working on…

I am trying out a few things and they include

Lidar – Neato’s Lidar sensor connected to..

Raspberry Pi

Robot Operating System (ROS)

cutecom to talk to the

Lidar Controller from GetSurreal.com

and tmux to facilitate dual screen operations and more

using rviz to display map of the room (up to 5 metres away)

prototyping control of Lidar unit using Arduino Mega

Aquired a Tripmate Router/NAS/Powerbank to avoid the university wifi

Installed different versions of Linux

(i) Ubuntu Mate (with ROS)

(ii) Raspbian Jesse (installed Processing and OpenFrameworks)

difficulty on combining OF and ROS! (help)

if I cant, I will use OSC and two Pi’s; one to sense the Lidar and the other to process in OpenFrameworks.

Any comments (not spam please) welcome! Thanks.



Final Show – notes pw: physical comp tutor


End of Year Project



I have looked into these areas so far;

  1. Autonomous robot with lidar sensor – no actual purpose, perhaps avoidance of objects in a room, mapping area etc.


Further extension of idea – automaton riding on the mobile vehicle, interactive with onlookers



I will learn a lot and it may result in an impressive installation

I ordered the parts anyway, I was curious and will continue with it in my own time anyway



Seems like a huge learning curve to throw all these paths into one project.

Is it art or an engineering project?

Terrifying learning curve

2. Small, not micro robot(s) large enough to mount small motors, scaled up into 5-10 robots to investigate flocking behaviours, emergence.



Good project to push computational content



Not very interesting, been done loads of times already, e.g. 1000 micro robot emergence flocking project by MIT[1] other methods using overhead sensor [1] ref on plans at Github here[1][1]

3. Using O-Drive and powerful brushless motors to animate a fast moving fully enclosed sculpture. E.g. dangerous sharp objects narrowly missing each other enclosed in a Perspex case.



minimal computational content, unlikely to progress with this.

Parts already sourced in any case.



Startling and highly dramatic piece, interesting.



Lidar and ROS operating system, investigate this with ODrive [1]motor driver board and Lidar control board [1]


1.2 Cartographer [1] and SLAM [1]library; Google’s mapping open source code for Lidar.



Useful course on SLAM and introduction to robot mapping


1.3 ROS [1]Robot Operating System (best installed in Linux Ubuntu[1])



Investigate further into I2C and AtTiny with small battery, continuous rotating modified micro servos for power.





[1] https://github.com/ShapeLab/SwarmUI

[1] https://github.com/swarminterface/Zooids

[1] https://odriverobotics.com/

[1] https://www.getsurreal.com/product/lidar-controller-v2-0/

[1] https://opensource.googleblog.com/2016/10/introducing-cartographer.html

[1] https://en.wikipedia.org/wiki/Simultaneous_localization_and_mapping

[1] http://www.ros.org/

[1] https://ubuntu-mate.org/raspberry-pi/

Prototyping for Term2 project

Testing the mechanism and stepper driver with an EasyDriver. I have used a LN298N H bridge board but I found the latter method less bulky and convenient. I am going to introduce a second stepper to incorporate movement about the y axis to move my object.

The supported object is heavy so I have to test the Nema17 stepper to see if it is strong enough.

AtTiny85 With mini Joystick

ATiny is a great little chip, low power, cheap to set up and compact. No Serial.prinln is available however so all testing is done on an Arduino Classic.

I tried out a little joystick I bought as a ‘bundle’ of sensors etc from China.

Fairly simple, using a more limited array of pins to connect to; two pulse width modulated pins and the rest as disgital pins (or use the PWM pins if you want.)

I used two digital pins and two PWM pins (see Fritzing diagrams)

I made a programmer board to plug into the Arduino as this was easier to carry about and re-use. Credit to Brian  Lough on YouTube. Really Good Video.

Video and more pics to be uploaded  very soon!


Examining the low ‘minified’ ATtiny85 chip, this provides a cut down version of the Arduino; using much less current and only about £1 in cost.

I connected up the ATTiny to an Arduino to burn the bootloader – this allows other programs to be loaded up afterwards. I also made a harness – dedicated, soldered circuit to simplify future programmingof the ATTiny without wiring up a breadboard each time.

Once the bootloader is loaded, I upload the ‘Blink’ sketch to prove it works.

Here is the pcb I made to program the ATTiny, it plugs on top of the Arduino.

underneath! Filed down solder to make a good connection to Ardiuino…

And Attached 3 LEDs to blink

Schematic for programmer shield



Proposal for End of Term Project: Sand Plotter


Many designs of sand plotters are found on the Internet. When I first thought of doing this, I was unaware of how well trodden the path is, however I am not discouraged because it will be a good test of fabrication and still leave plenty of scope for designing patterns that can also be interactive to external environmental factors.

Movement or sound could modify the emerging pattern of the plotter, which may run continuously for extended periods. Thus, patterns can be drawn in a circular (it could be square also) enclosed chamber with a steel ball or cylinder.


Under the enclosed chamber is a revolving double rail, powered by a high torque stepper motor. The double rail will have a magnet mounted, pulled in the y axis by another stepper motor. The control board will be an Arduino mega with a Ramps shield board (or similar) comprising polulu stepper drivers.

So, with one circular motion and one lateral motion, circular patterns can be drawn in the sand.

Software to control the Ramps board will be developed and also some runtime scripts developed to demonstrate the sand plotter.


First stage will be to construct a frame without enclosure, mounting the mechanical parts. A very simple test script to move the turntable and lateral axis with be the first stage.

Second stage is to provide accurate control of the stepper motors

Third stage is to fabricate the sand enclosure and outer box (laser cut)

Fourth stage (possibly too much in the time alotted) – to build some interactivity with external sensors, thus modulating the sand pattern in real time.


I do not want to build anything too large, it becomes impractical, however it must be large enough to look impressive in a gallery setting. So, at least 50 cms diameter.


I am not the first…



Main Parts list:

Laser cut enclosure

Bearings for circular movement

2 x Nema high torque stepper motors

toothed belt, possibly 3d printed gears, metal belt pulleys,steel rods and linear bearings (I have most of these parts from building a 3d printer)

limit switches – these may not be mechanical but optical sensors

Ramps shield and Arduino Mega, polulu stepper drivers

Steel rods and 3d printed supports

Neodymium magnet

15-20 mm diameter steel ball