Skip to content

Meetings archive

Jump to bottom
Jacques Chartier-Kastler edited this page Apr 1, 2020 · 55 revisions

List of Meetings [GDocs List of Meetings]

2020.04.01 DIY Droners #3 [GDocs link]

Please join us for this meeting!

2020.03.25 DIY Droners #2 [GDocs link]

Background

Video archive on Youtube: https://youtu.be/q78I1be_ZmQ

Date: 23.3.2020 | Time: 7 pm | Place: Makerbay Tsuen Wan and Google Hangout

People:

  • Michael O’Brien #Industrial Design
  • Adel AKLOUL #cyber security
  • Cesar # maker #a million other things
  • Felix # maker
  • Davide # robotics
  • Mathew cua # hard core droner
  • Ming THS # teaching
  • Jacques # interdisciplinarity

RoadMap

Goals:

  • Features: we map coral reefs + collect metadata
  • Subject: we focus on shallow water coral reef <10m depth
  • Price: we want to make it affordable enough <1500 USD
  • Usability: deployable from the beach with an autonomy >90 minutes
  • Target users: ocean scientist, citizen scientist, coastal community, Marine protected area manager

We want to create an easy workflow for non professionals, to deploy and collect data: Dronedeploy concept is perfect, but in practice it crashes bad.

HARDWARE

Stages of development

Functionnal

  1. It floats, it's waterproof
  2. It moves! Radio-Controlled boat
  3. It has a high battery autonomy (> 90 min)
  4. It can be programmed! Ardupilot-controlled boat

Imaging

  1. It sees (and record on an SD card)! It's got a camera on board
  2. it sees IN REAL TIME (see what I see underwater from the shore / from the deployment boat)

Monitoring

  1. It can collect environmental data (temperature , pH, salinity, oxygen, luminosity, sonar, turbidity)
  2. It can collect depth-dependent environmental data (temperature , pH, salinity, oxygen, luminosity, sonar, turbidity)
  3. It can collect samples

Priority

Repeatability (it can come back to the same location)

SOFTWARE

Coral monitoring AI development

Goals:

  1. Recognise a feature VS No feature
  2. Classify rock, sand, coral, fish, ghost net, seaweed
  3. Identify the coral family (from shape & colour)
  4. Identify the coral genus (from shape & colour)
  5. Identify the coral specie (from shape & colour)
  6. Evaluate the coral health (from colour)
  7. Statistics of coral coverage, growth, health, biodiversity

Development steps

  • Get a lot of pictures of coral snorkeling
  • Build a database
  • Label the pictures for the AI to be trained
  • Train the AI

Software functions

2D Mapping: Photogrammetry, 2D photo into map tiles

  • Create a map of Coral Reefs
  • Identify corals
  • Evaluate the health of the reefs
  • Discover new reefs
  • Monitor and patrol to protect Reefs

3D Map: photogrammetry, 2D photo into point cloud, and point cloud into 3D model

Integration: multiple agents "Swarm" collaborating

2020.03.18 DIY Droners #1 [GDocs link]

Time: 20200318 | Place: Makerbay Tsuen Wan and Google Hangout People:

  • Michael O’Brien #Industrial Design
  • Sidhant #AI #Robotics
  • Anders #Design
  • Brad O’Dell #Art #Fish #Tank #Education
  • Abbas #Mechanical #Engineering
  • Kwok Wah #Facebook #Coding #Marketing #Solution Engineer #advertising #Hobbyist
  • Adel AKLOUL #security
  • Philippe COUTURE #Architect #Filmmaker #ECOSquared.org

Table of contents

Discussion Why not making it another platform Android-powered?

Pixhawk-powered?

Raspberry-Powered?

Pricing Start expensive with high-end applications CTD

Architecture Only surface? ROV conversion? Questions and Comments from Group

Philipe: Comments proving to the community that the fish communities ARE still there Baseline surveys in amount of fish colonies in reefs Fish are moving and counting fish is variable Laser quadrant would help count the fish Laser would be in the photo and that's no problem according to Philipe

H: Q: Can you make everything smaller ? A: With SEED manufacturing we will - see link : https://www.seeedstudio.com/

Anders: Lots of discussion on what the drone needs to be ? Then the technical aspects to reach the goals. 3 levels of Define what it needs to be Recruitment of the team of experts Develop it

Philipe: AI of the coral is not as useful as the AI features for identifying the fish Define the different use cases Keep one focus and try and try and meet to meet one of the requirements

Adel: Think about the end user and the marketing of this product

Michael to create a common google doc: and the everyone inputs their ideas for user scenario for what it needs to be Then we find the common points for what it needs to be and what it needs to be done Please scroll to the end of this document for this section

H: Suggests to develop and or refine a goal, this will help the whole process and organisation.

Note for Sanjeew: Sanjeew and or our AI team, is it possible to write a piece of software that views a picture of a reef from above, I.e a bunch of corals overlapping and the AI separate the coral into individual components?

Issues

Camera Getting the camera that we need and testing with it Camera: large image sensor, integratable, and affordable (This is hard to find) Go pro: we have used this but its closed the doors Arducam: large megapixel, with Auto focus Define the minimum requirements with a coral photographer, philippe is a diver and says any camera can do it Q: Why do you need real time info in the image? We can take pictures on a any normal camera and then do post processing:

Q: How do you calibrate the camera? Some sort of chart that you put in front the camera to calibrate the color of the coral The chart needs to be in the water, outside of the tube Attach the laser to the boat to help with calibration

We need to use DIY Droners to help us look for said camera ? Any one with any contacts to manufacturers ? Canon, Go pro, panasonic, sony, Leica etc Documentation Github https://github.com/MakerBay/Coral_Reef_Mapping_Drone

DIY Droners #1 Meet up slide show https://docs.google.com/presentation/d/1j4I_Rg5wkQaXkAAQc53EmIh80ZOjIIcQ9Yx8rvPSJEM/edit#slide=id.g818da5978f_0_101

2020.03.04 DIY Droners Knowledge Transfer Session with Ken Chew [GDocs link]

Background

Date: 04.3.2020 | Time: 7 pm | Place: Makerbay Tsuen Wan People

  • Ken Chew
  • Michael
  • Sidhant

Agenda

Electrical Assembly General overview of current build of electrical assembly Binding X8R receiver and Taranis x-lite radio Unbinding & Transmitter setup Lipo battery charging, and balancing Review Ardupilot settings. Optimal setting for control of drone boat. Flight Modes Ardupilot Waypoint Navigation setup & tuning for rover mode Testing current build in water bath and or in ocean Tuning (settings on Ardupilot) Ardupilot capabilities and limitation Can we output to Pi - how? Camera Advice

Notes

X8r Receiver-at this stage doesn't matter that much BUT in the future Best to use NON EU because the power / range will be limited. Tattu battery is a good brand to buy or aca brand. Just power the x8r first with a small battery in any one of the 8 rows and bind the radio receiver and controller Tararins xlite radio may need a firmware update - Each Model can be each be binded to a separate receiver LiPo is dangerous and easily damaged, get a voltage alarm https://cdn.getfpv.com/media/catalog/product/cache/1/image/9df78eab33525d08d6e5fb8d27136e95/l/i/lipo-meter-screamer.jpg https://www.getfpv.com/1-8s-lipo-battery-voltage-tester-low-voltage-buzzer-alarm.html 3.6 per call - lowest, 4.2 V highest HUD Setup Initial setup -> Optional Hardware -> Battery Monitor -> Analog Voltage & Current -> 3DR Power module then you can see the battery data on HUD Can connect directly instead of telemetry for fast Can do calibration on battery in initial setup -> battery monitor Never let it go below 13 or so. Below 3.2V will damage the battery. Do not keep it at full charge. While on charging, we try not to leave without supervision DO NOT STAB WITH SCREWDRIVER - will catch fire Flight mode current PWM is the channel which needs to be set on transmismitter. Current PWM variable can be set as a parameter. On controller, go to Trottle scrren, scroll down to current PWM number, select and bind to 3 position switch (ideally). Auto - when you start running auto-mission plan Flight mode - set by PWM value Loiter - same GPS location Hold - stop everything - hold is good Disable switch -> Variable in parameter (SAFETY_ENABLE)

MAV-Link Firmware on the Pi - can read sensor data

Go-Pro physical trigger, but quite hard and not happy. GPS Data -> Tagging into the photos -> Pixhawk is triggering

Initial Setup -> Optional Hardware -> Motor Test Should be C & D

Make a tank w tracks - > used that go to the roof waypoint right in front of me. Parameters in basic tuning within configuration - one upfront sonar possible Tank vs Ocean

Mount the telemetry the and the GPS up to a metre apart from each other Think about duo GPS so you have a back up GPS

Do maintenance on the thrusters every time after ocean run- run in freshwater and then grease components in vaseline every few times after ocean / salt water testing.

Rebuild the tank- or buy a new RC control tank for land testing

Actions:

Get LiPo Voltage Alarm : https://www.google.com/search?q=Get+LiPo+Voltage+Alarm&source=lnms&tbm=isch&sa=X&ved=2ahUKEwj-7dSb3YDoAhVGVN4KHQrrD10Q_AUoAXoECAwQAw&biw=1536&bih=722 Buy or rebuild the testing RC Tank: https://www.google.com/search?q=remote+control+toy++tank&tbm=isch&ved=2ahUKEwjl8_Wv3YDoAhVGa5QKHYv5AK0Q2-cCegQIABAA&oq=remote+control+toy++tank&gs_l=img.3..0j0i8i30j0i24.5239.5874..6102...0.0..0.121.423.3j2......0....1..gws-wiz-img.......0i67j0i7i30j0i8i7i30.gVK4UAESetQ&ei=0JJfXuXLPMbW0QSL84PoCg&bih=722&biw=1536

Link to Google drive for photos and videos https://drive.google.com/drive/u/0/folders/1An99r47O7Xg0rDhmxKo2N9BQBH2ernSN

A project by

https://www.makerbay.org Maker Space and community workshop based in Hong Kong. Innovation for social and environmental impact.

Clone this wiki locally