Lightning Talks

Luka Čubrilo

University of Novi Sad

Table of contents

• Intro
• The problem (and context)
• The solution
• User needs
• Implementation
• Benefits and impact
• Conclusion
• Future steps?
• On Arduino ecosystem
• Takeaways

2

Lightning Talks - Making old Lab Machines Smart

Intro and motivation (why should you care)

AUTOMATING AND OPTIMIZING WORKFLOWS

• Strong effects of AI on productivity, many niche industries and applications…
• … but mostly digitally (software)
• Is it possible to get some benefits physically too?
• This is just a case study (specific problem/solution)
• Motivation? Give an example/inspiration for better, bigger and more impactful optimizations and automations

4

Lightning Talks - Making old Lab Machines Smart

THE CONTEXT

• Late ‘22 - early ‘23, Data Science internship at JSI in Ljubljana, Slovenia (Department of Electronic Ceramics - K5)

5

Lightning Talks - Making old Lab Machines Smart

• Made a desktop software for standardizing/combining data from different experiments; data manipulation, analysis and visualization

THE CONTEXT

• Late ‘22 - early ‘23, Data Science internship at JSI in Ljubljana, Slovenia (Department of Electronic Ceramics)
• Made a desktop software for standardizing/combining data from different experiments; data manipulation, analysis and visualization

6

The project this talk is based on

THE PROBLEM

10

Lightning Talks - Making old Lab Machines Smart

THE PROBLEM

11

8-12× a day

Lightning Talks - Making old Lab Machines Smart

THE PROBLEM

• Why were junior researchers going through that 8-12 times a day?
• Machine has no scheduling system or remote control.
• This and other limitations => plans to get a new one in the next months.
• The lab technician was complaining about it, but didn’t think of a solution apart from pushing for new one to be bought
• Having overheard the interaction, I offered to make a tailored solution for them.

12

Lightning Talks - Making old Lab Machines Smart

BRAINSTORMING

• I was warned not to modify the machine itself or to risk breaking it
• It should be temporary and easy to add/remove if needed
• Having prior Internet of Things (IoT) experience and willingness to learn 3D modelling, I got to work.

13

Lightning Talks - Making old Lab Machines Smart

The solution

14

Lightning Talks - Making old Lab Machines Smart

Here is how it looks like

• Detachable, lightweight, cheap “robotic finger”
• Easy to install on any other machine for pushing buttons
• Easily upgradeable for multiple buttons, levers, dials
• Potential to connect to a smartphone for remote work

15

Lightning Talks - Making old Lab Machines Smart

Structure of the project

1. Off-the-shelf electrical components
2. 3D modelled and 3D printed components
3. Software

16

Lightning Talks - Making old Lab Machines Smart

1. Off-the-shelf electrical

components

17

Lightning Talks - Making old Lab Machines Smart

1. Electrical components

Electromotor - MG90s

• Tracking position, metallic gears, cheap

​

LCD Display + buttons

• Most user-friendly, clear and intuitive option, cheap

18

Microcontroller - NodeMCU

• Very versatile, has WiFi capabilities, cheap

Lightning Talks - Making old Lab Machines Smart

1. Electrical components - wiring

19

Electromotor MG90s

Microcontroller

Voltage converter

Lightning Talks - Making old Lab Machines Smart

• How did I do the wiring?

Things can get confusing.

20

Pinout schemes�& documentation.

Lightning Talks - Making old Lab Machines Smart

2. 3D modelled and 3D printed components

21

Lightning Talks - Making old Lab Machines Smart

2. 3D modelled & 3D printed components

22

5:1 gear reduction

2 casing interlacing panels

LCD cover & buttons

Optional panel

Lightning Talks - Making old Lab Machines Smart

2. 3D modelling process

• Necessary to take into account every “piece of the puzzle”
• Measured to the fraction of the milimeter
• Even the trivial things such as the ball bearing

23

Lightning Talks - Making old Lab Machines Smart

2. Buttons

For navigating the user menu:

+- Increase/decrease numerical amount in the menu (num of cycles, length of cycles)

→ Next step

↻ Reset the robot

24

Lightning Talks - Making old Lab Machines Smart

2. Mechanism

• Simple 5:1 reduction
• Increase pressing force, decrease range of motion
• Ball-bearings significantly reduce friction
• Characteristic hole design to account for print errors and snug fit

25

Lightning Talks - Making old Lab Machines Smart

Mechanism after printing

26

Lightning Talks - Making old Lab Machines Smart

3. Software - C++ code

27

Lightning Talks - Making old Lab Machines Smart

3. Software

• Based on the Arduino ecosystem (hardware and software standards)
• Using Arduino IDE, C++^* programming language
• Can directly address specific pins on the microcontroller, measure received voltage, send specific voltage
• Hardware components support high level of flexibility programmers are used to (instead of being single-purpose)
• Big open-source software development community that broadens compatibility of off-brand products with the ecosystem

28

*It’s not pure C++, has some Arduino specificities

Lightning Talks - Making old Lab Machines Smart

Outcome?

29

Lightning Talks - Making old Lab Machines Smart

THE IMPACT:

• Saved 8-12 daily 5 minute trips
• (40-60 minutes each day, ie 20hrs a month, 240hrs a year)
• Delayed the necessity of buying a new $20,000 “smart” machine (with IoT capabilities)
• => ability to direct resources to more urgent and impactful matters first
• Reducing electronic waste of an otherwise perfectly fine machine

30

Lightning Talks - Making old Lab Machines Smart

IMPROVEMENTS WE DISCUSSED�(but never ended up implementing)

• Internet connectivity => having a mobile/desktop app that tracks everything => changing schedule remotely
• A vibration sensor for error-checking to confirm if a click caused a machine powerup => attempts to re-click several times => reporting if it fails
• Documented schedule => seeing which ones worked and when => seeing which actions are upcoming => exact status (fail/pass), elapsed time, remaining time

31

Lightning Talks - Making old Lab Machines Smart

… more improvements

• Obviously it could be made to click multiple buttons, levers etc
• Ability to respond to environment intelligently (not just follow schedule blindly)
• Camera to detect any visible changes in the experiment
• Microphone to hear any changes
• Hall’s detector for electromagnetism
• Movement detectors
• Basically any detector

32

Lightning Talks - Making old Lab Machines Smart

ECOSYSTEM

• Why is this all this expandability possible and easy?
• Because Arduino created a vast open-source ecosystem
• Countless producers of microelectronics (sensors, motors etc) worldwide that comply with these standards
• => plug and play
• => ability to program in Arduino’s flavor of C++
• (maybe MicroPython on RaspberryPi or something)

33

Lightning Talks - Making old Lab Machines Smart

TAKEAWAYS

• The main conclusion : automations of repetitive tasks is always possible…
• … and can often be easy
• You can do it beyond the virtual

34

Lightning Talks - Making old Lab Machines Smart

CHATGPT’S IMPACT

• With the advent of ChatGPT - even more so!
• Ability to get accurate responses:
• brainstorming how to resolve some inefficient procedures
• answering question about which components to buy for specific problem
• analysis of documentations (of each component), which are otherwise tedious and dense
• getting code snippets for coordinating it all
• getting advice for setting up your work environment (Arduino IDE, libraries etc)

35

Lightning Talks - Making old Lab Machines Smart