03/08/18

D-DAY

02/08/18

We set up a second prototype as a spare. We worked on slides and also set up our display table for the presentation.

01/08/18

We tried to make a balloon for the final prototype but there were microleaks. The suspected reason was the tear near the seam that we ironed over to incorporate into the seam to avoid making a new one. However, this is not a large issue as we already have a working prototype and hopefully can fix the second one tomorrow.

We also printed a lighter version of the buoyancy mechanism basket to make the buoyancy mechanism more secure than some bluetack on a stick.

One of the Raspberry Pis fried after we connected it to powerfor 10 minutes while testing. The reason is unknown but suspected to be a short-circuit on the Pi internal board as the soldered portions were properly soldered. Hence, we had to set up our spare Pi today as well.

31/07/18

We worked on presentation slides and blog.

30/07/18

We made a lighter buoyancy mechanism so that the servo which is used for the tail can be used without frying. We also attached a capacitor onto the buoyancy mechanism which reduced the frequency of crashing.

27/07/18

The new servos arrived and we began to test them. However, the Raspberry Pi kept crashing and restarting. We discovered that the new servos drew too much power from the Raspberry Pi, resulting in the constant crashing. Hence we had to roll back to the older 95E servos as they did not draw as much power. However to address the frying of servos, we decided to make the CG mechanism lighter so that the torque load could be reduced. We acknowledged that this will decrease the effectiveness of the tilting of the CG.

23/07/18

We made yet another balloon, but this balloon was more aerodynamic with the same lift of 120g. We also printed new servo supports for the new servos which will be delivered soon. However servos have not arrived so we cannot test them. We also recorded footage while creating the new balloon for the video. We are taking a break for the next 3 days due to being in the organising committee of Freshmen Orientation Programme.

20/07/18

Today we built yet another blimp, correcting our method, but it was too small, lifting only 60g. We also fixed the buoyancy mechanism and attached it onto the previous working set.

Today was our first full test flight. We flew the HoverFish to the 5th floor. The buoyancy mechanism and tail is working fine. There is a slight lag occasionally but overall the UI is still usable. Unfortunately, after about 10 mins, the buoyancy mechanism servo fried due to suspected overload. We may have put too much weight on the buoyancy mechanism which the servo could not support. A decision was made to get new servos which can have greater torque.

19/07/18

Today we built a new blimp, which was completely working except for the buoyancy mechanism. We tried to make an ellipsoid shape but failed because we did not use the method from Euler correctly due to a misunderstanding in one of the . We also tried to estimate volume but calculations were way off. The new blimp has a lift of 120g.

We also edited the code to make the buoyancy mechanism change less abruptly.

18/07/18

Today was the progress meeting where we finally had a prototype to display. We demonstrated how our UI and blimp worked.

We created the tail from the 12 micron mylar instead of trash bags as planned. The mylar is more durable, lighter and aesthetically nicer.

We decided to rethink the balloon shape, maybe to make it more aerodynamic. We were going to print new hinge and support to better fit the tail servo. Also going to refine the buoyancy mechanism. We also decided to print a mount for the Raspberry Pi so we don’t need to use the provided Pi box, which is heavy and does not let heat escape as easily.

17/07/18

We decided to try the 12 micron Mylar again, in case the leaks last time was due to human error. We changed the design of the balloon into an octagonal shape. We cut out a template using the 12 micron Mylar and seal with the HeatnBond. We tried again with the sock on the hobby iron which only sealed the HeatnBond without melting the Mylar. However, when we inflated the balloon, the seam gave way along one edge. We removed the hobby iron and sealed again. The HeatnBond did not give way but the same problems with leaks occurred, where small holes appeared near the HeatnBond where we accidentally went over with the iron onto the Mylar.

Luckily, the new 30 micron Mylar arrived, and we got to work with the new dimensions. The new Mylar balloon works but the valve is faulty. The balloon only leaks significantly from the valve, which we temporarily sealed with tape. It managed to fly! Due to the octagonal balloon shape, the balloon was not aerodynamic enough, hence the tail could not push the balloon satisfactorily fast enough.

16/07/18

We realised that the MAC address for each individual device is different so need to edit the MAC address each time.

We calculated new dimensions for the new balloon, meant to cut down 50g of payload from the old balloon. We also decided to make the balloon rounder. As for the actual making of the balloon, we decided to use a rectangle of Mylar so that we can tape down the sides. We also decided to HeatnBond before cutting out the template. We tried to use the sock to prevent accidental melting of the Mylar, but found that the sock prevents the HeatnBond from binding together. We inflated the new balloon but it had a lot of leaks, which is due to the micro tears near the HeatnBond, which could be due to stress or abrasions.

We also resoldered wires.

15/07/18

We reinstalled Rasbian and created a second ap0 interface with the same mac address.

After some troubleshooting, the Pi was able to connect to networks registered in the wpa_supplicant.conf while hosting the access point.

However, the status of the wireless properties available through the GUI stated that DHEPED was lost and wireless interfaces were unavailable, meaning networks had to be added through the wpa_supplicant.conf.

14/07/18

Converted the second Pi to a wireless access point. However, as the wlan0 interfaced was used, the Pi was unable to connect wirelessly to other networks.

We concluded that this was mildly successful – the second Pi works as an access point, however, it acts like a local area network as it is unable to connect to the internet.

Purchased the hobby iron for use in heat-sealing the Mylar balloon.

13/07/18

Based on the approximate weight that we needed the balloon to support, we made approximate measurements for mylar balloon

Cut out blimp shape from mylar sheet

Used fabric product “Heat-n-Bond” to stick 2 sheets of mylar together, as outlined in the guide supplied by Tony Avak, a mylar supplier, as well as an online forum page named “RC Groups” which specialised in the construction of blimps.

Used a normal clothes iron to seal the heat n bond

Difficult to control temperature using a clothes iron

Accidentally melted some parts of mylar

Heat and bond is successful with no leaks

However, when inflated, the melted mylar tore and we had to tape it back together to prevent leaks

The balloon was also too big, half inflated can lift 120g

Not completely filled means the tail is not working properly

Buoyancy mechanism not working

Decided to get a hobby iron which enables precise temperature control

12/07/18

• Installed NOOBs on second pi
• mylar finally arrived – will make the tail tomorrow

11/07/18

• Soldered new Pis
• Attached new servos to the Pis
• Tested a miniaturized Pi on a smaller balloon to show that the buoyancy mechanism is able to tilt the balloon
  • it is a success!

10/07/18

• Printed support for buoyancy structure
  • Much simpler design compared to the tail servo
• Attempted to clone the working Pi SD card onto the new SD card
  • Copying and pasting the contents of the SD cards did not work
  • Image of the original SD card was too large to fit onto the new card

06/07/18

• Joken measured the various components which would be mounted with the tail, prototyped and printed the tail support

04/07/18

• Worked on presentation slides
• Presentations took the rest of the day
• Received advice on 3-D printed a support structure for the tail from Hanyang

03/07/18

• Joken popped the balloon 🙁
• Went to source for tail materials
• Realised that the airswimmer uses a simple DC motor as opposed to our servo
• Disassembled the airswimmers DC motor and discovered a gear system used to step down the RPM and increase the torque of the motor
• Found that the 4.3 g servo was able to flap the supported tail
• Compared the flapping action of the DC motor to our Pi controlled servo
• Found that the 4.3 servo was able to generate enough torque to lift the battery and DC/DC booster at a distance of 10cm

02/07/18

• Joken bought plastic rods to try to use to support the tail
• Trying to used the plastic rods to support the styrofoam tail
• Attempted to move it with the 4.3g servo
• Result:
  • Servo was unable to handle the load of the tail
  • We decided that the weight of the tail needed a support structure and could not be placed on the servo
  • Styrofoam was deemed to be too heavy for the tail
  • Plastic rods were deemed too flexible and heavy
• Opted for carbon fibre rods and a 3-D printed support structure

01/07/18

• Joken bought plastic rods to try to use to support the tail

29/06/18

• Bought new batteries as backup

28/06/18

• Went to Mnt lab to find the  battery
• Talked to kanesh about the frying servos
• Suggested to us to get larger servos at 6g weight
• Kanesh said servo load was too much
• Kanesh said to cancel the already ordered 1.7g servos but too late
• But actually servos fried because too high voltage
• Joken was cutting tail shapes from styrofoam

27/06/18

• Soldered the USB
• USB keep getting fried
• Went down to MnT lab to measure the weight of everything
• Joken designing the battery holder
• Fixed the soldering of USB cable at night
• Tested the tail
  • Worked but keep dropping off and motor fried
• Bought more servos

26/06/18

• Touch up the code to fix minor bug issues
• Redefined http communications
• Make the smartphone registers that the pi has received the request
• Recoded the request time

25/06/18

• Designed early User Interface (UI) prototype – It is working!!!
• Major breakthrough – the touchscreen application can communicate with the pi to command the servo motor
• Identified issue:
  • servo keeps jittering – probably need to switch to another servo library
  • make the UI nicer – add joystick appear when the screen is touched
  • no helium
  • no balloon

21/06/18

• Went to Pioneer garage to do soldering of cables to our Pi
• Tested Lipo Battery capacity – 17 mins with script and camera running
• Probably need battery with bigger capacity

20/06/18

• Next progress meeting we need to have the mass measurements and a preliminary blimp prototype
• Realised that it is hard to achieve neutral buoyancy – any helium leaking out or undocumented change in weight of the component tips the balance over
• Kanesh asks us – “What is wifi?”. This prompts us to understand wifi and http protocol better
• Estimated the total mass of item
• Figured out how to work with Lipo battery

19/06/18

• Realised previously that battery will be an issue for our project (Source 1, 2, and 3)
• Pi W Zero has certain voltage requirements that need to be met, but the project requires us to use as light a battery as possible

18/06/18

• Decided to revamp design to reduce need of long cable to connect to camera module
• Using clear packing tape to attach tail to the end of AirSwimmer
• To buy longer wires and a measuring scale to enhance preliminary calculations in estimating the required volume of helium
• Worked on Powerpoint Presentation for Wednesday presentation
• Brainstormed on more purposes/useful features of the AirSwimmer (child-friendly, safer, cheaper, advertisement, entertainment as toy, flying times (battery life), flying altitude)
• Updated Blog
• To make presentation video as well as a poster
• Pi is working, but only when connected to a powerbank. This might be a huge issue if not resolved as soon as possible

13/06/18

• Met Dr Ho in the morning to discuss the progress of the project
• Updated him about how our components have yet to come
• Assembled the Airswimmer and inflated the balloon
• Realised that the DC motor does not work (to get refund )
• Learnt how to stream video to a browser using the raspberry pi camera
• Connected the Pi to NTUSECURE so we can work on it more easily
• Borrowed the pin adaptor and breadboard to test motor script
• Reprogram script for servo (tutorial link)

12/06/18

• Worked on blog (scientific principles)
• Attempt to set up python libraries for blynk on the pi (github link)
• Got VNC viewer to work so we can work with raspberry pi using laptop as the interface instead of connecting screen, keyboard and mouse to the Pi
• Tried to make Pi connect to NTUsecure but cannot
• Borrowed PiCamera to test

08/06/18

• AirSwimmer was delivered
• Opened AirSwimmer and discovered that the components used were very small
• Need to source for equally small and light components
• Installed raspbian on the borrowed Raspberry Pi 3 board
• To transfer the logic from the Arduino scripts for the motors to python so we can test on raspberry pi

07/06/18

• Combined servo and dc motor scripts in arduino
• Tried out Arduino Wifi Mod but need soldering of the parts
• Learnt Blynk through tutorials

06/06/18

• To do:
  • Millis instead of Delay in Arduino script
  • Arduino wifi
  • Blynk
  • Learn raspberry pi
  • RaspiCamRemote
• To source for:
  • DC Motor – Pull a load that consists of Raspberry Pi, Batteries, breadboard (maybe)
  • Servo motor – Tail size? Tail material? Needs to be strong enough to flap tail
  • Balloon – To use Mylar and check up its properties

05/06/18

• Updated the logbook
• Realised that we need to order the components we intend to ahead of time:
  • Servo motor
  • dc motor
  • premade balloon

• Worked on the powerpoint for Wednesday’s progress meeting
• Also learnt about the various related scientific principles
• Created a Gantt chart which let us realise the limited amount of time and hence came up with a more concrete plan
• Jiahe to complete risk assessment and Damian to complete the purchasing spreadsheet in time for Wednesday (06/06/18)

31/05/18

• Read up some tutorials on how to use Raspberry Pi

30/05/18

• Updated Dr Ho on our progress thus far with programming the Arduino
• Despite Arduino not being usable in our actual prototype, it was a good experience and we also learnt more about the considerations we had to make
• Updated Dr Ho on our decision to utilise Raspberry Pi instead
• Borrowed a Raspberry Pi set to test programming
• Motors
  • Look into brushless motor – heat up slower
• Flight controller
  • Pixhawk
  • Function – stabilise the machine despite gusts of air
• Learnt not to 3D print gears as stress points at teeth will likely cause it to break

25/05/18

• Programmed DC motor for directional control with the pressof 2 buttons:
  • As long as UP button is pressed, DC motor will rotate clockwise
  • As long as DOWN button is pressed, DC motor will rotate anticlockwise
  • If neither or both is pressed, DC motor will stop
• Learnt about Arduino “millis” command instead of delay
• Delay will pause the entire programme from taking input, millis will not prevent programme from taking input, necessary for simultaneous control of servo and dc motor, and also instantaneous change of direction of dc motor
• Did research integrating a camera into Arduino for the video feed, realised that Arduino cannot support image processing.
• Have to change to Raspberry Pi which is a similar microcontroller but with video capabilities.
• Plus side, raspberry pi circuit board comes with inbuilt wifi capabilities which removes the need for a wifi module, hence reducing the payload of the blimp which enables us to either reduce the size or incorporate better cameras for a better user experience
• Downloaded Blynk and started to read tutorials on how to use it. Necessary (at the moment of writing) for wifi module for Arduino (for testing purposes) and Raspberry Pi (for actual prototype)

24/05/18

• Updated Dr Ho on our plan
• TArduinouino kit, 1x servomotor, 1x strip camera
• Programme the servo motor for propulsion and directional control using a button as the input
  • Result: Success, able to move forward, left and forward, right and forward on demand
• Next: purchase wifi module, learn how to use wifi module, and able to transmit the button input through wifi to move the motor

23/05/18

• Sketched preliminary ideas for the design of blimp
• Discussion about next steps
• To use Wifi Module for transmission and receiving instead of the original plan of a separate video transmitter and receiver for remote signals to cut down mass required
  • Step 1) Connect Arduino to camera and motors
  • Step 2) Connect Camera and motors to microcontroller connected to wifi, connect a computer to the wifi
  • Step 3) Replace with a smaller microcontroller (Arduino Pro) and attach to balloon (Keep track of total mass to ensure that it will work)
  • Step 4) Optimise the balloon design and components used to minimize size, programming for the robot to be more responsive