Improvements Just In Time.
Few weeks ago, Explorer-27 was a short, dog-sized robot:
and after weeks of further development..
fast prototype using sticky plasty to form a tablet holder.
Afterwhich, I also built the new paper model dog head(Husky) 
And installed the LED eyes into this new paper Husky head.
Re-code the UI in Unity and also connect it to the Arduino system.
continuation from the previous part 2 post with some new ideas after the talk about the past FYP showcase from Assoc Prof Laura Miotto, I’ve been inspired to do something that focuses solely on the showcase, since the showcase will mark the end ADM life of all current year 3, I would like to do something for many of the peers that would enhance student’s FYP showcase capability as well as the visitor’s experience of the visit.
As the founder of Kimchi and Chips, Elliot Woods asked during the Emergent Visions Symposium,
“What is the work that only I can make?”
This question hit me like Tiger Woods golf clubbed me in the head…
Yes, what is the work that only I can make in this coming FYP showcase?
back to my first post, I’ve already know what my strength is at this point of time, I would really want my FYP to have an impact on others, why not use my strength into making other’s FYP presentation to potentially having slightly more possibility? Wouldn’t that be great?!
so, this will be an idea which I know that I would enjoy spending a year to complete.
Just imagine, every student have their booth while a robot moving around in the FYP show displaying their project on its screen, when a visitor saw it on the screen, got interested and wanted to see the booth, they simply click a “Bring me to this booth” button and the robot will do so. How much more potential exposure will that bring to that student?
And on another aspect viewing this project, what will be realer than an actual application of an FYP project during the FYP show? What the visitor sees is what was actually made specially for that FYP show, if it doesn’t work or spoiled halfway, it simply meant that I did not plan well/have a backup or do a good enough job but there will not be an imaginative scenario where “This FYP was supposed to work in another location” *I am not saying that that kind of project isn’t good, but just what I would not like to do*
To design a set of Robots that have the ability to move around while avoiding people/obstacles during the FYP show. The purpose of the Robots is to expand the presentation possibility of student’s project while enhancing the experiences of the visitors. Safety of the people around the robot should be of the utmost importance and under no circumstances whereby injuries caused by the robot should happens.
Although it will be really hard to do so, but I aimed to complete a project of this standard but the task which my robot will do is much simpler.
There were many considerations made for this aspect of the project, firstly, the Cost and Value Analysis of the system must be made, of course, if money isn’t a problem, then this part would be really easy as I will just get the suitable parts which I need for the FYP, but sadly that isn’t the case… Money is a problem in FYP.
So, how can I maximize the output of my project by using the minimum amount of money? I have to also consider about how strong/fast the system is, and how would I like to have the user interface (touch screen) to be incorporated into the system.
These are the types of computation system which might be possible for my project at reasonable price –
This is the latest Raspberry Pi 3. It features a 1.2GHz 64-bit quad-core ARMv8 CPU with 802.11n Wireless LAN and Bluetooth 4.1. Like Raspberry Pi 2, it also has 1GB RAM,4 USB ports,40 GPIO pins,Full HDMI port and Ethernet port. This third generation Pi is an excellent tool for hackers, makers and educators because of its small footprint, low power consumption, and low price.
Pi 3 comes with build in WIFI system so that would save me about $30 for WIFI module, output of screen is easy to setup(like a normal computer), runs on Linux system, which I have a slight experience on and as a bonus, I own one of this so that would save me some money too.
A LattePanda is a complete Windows 10 single board computer. It has everything a regular PC has and can do anything that a regular PC does. It is compatible with almost every gadget likw printers, joysticks, cameras and more. Any peripherals that work on PC will work on a LattePanda.
In terms of processing strength, completely overpowered Raspberry Pi 3, the price is however much steeper than Pi 3, also as an added bonus, I already own a Lattepanda and I would rather work on Lattepanda over Raspberry Pi as it runs on Windows system and that would save a huge amount of time for me, plus overall it is much faster at processing video and even runs Max MSP(a programming software) which is what I will probably be using. I am sure Latte Panda will be more responsive than Raspberry Pi 3 at the cost of more $$.
Yes, I praised so much about Lattepanda, but the cheap CPU nowadays is already cheaper than a LattePanda, although bigger,heavier and more power hungry, it is way cheaper than a Lattepanda at relatively same processing power and also, a cheap CPU usually comes with Internal HDD(atleast 128GB) while Lattepanda only has a mere 64GB(minus operating system installed, leaving about 40GB). Best part, Desktop CPU comes with built-in cooling fan, which overheating is a problem for Lattepanda.
*REFURBISH is the word here hence the cheap price*
since I don’t really need a brand new computer to get the job done. This is the normal sized CPU, which is rather big but I am sure it is small enough to be build into a robot.
For me, I think that this is the best system i could think of for now which I will need, firstly, touchscreen laptop meant that I do not need to purchase another separate touchscreen(which is usually really expensive) secondly, it comes with all the benefit of a CPU like cooling fan and such, also, laptop have a additional battery which reduce the load of the power supply, like if the main power supply ran out, there will still be time for the system to send out notification to the control system which then battery could be replaced. lastly, laptop will be slightly lighter and uses lesser power in comparison to a CPU, which make it really good over a normal CPU.
These are really good deal I think, I have been also browsing on carousell to find cheap and well condition touchscreen laptops, will update when I found some.
Since I don’t have sufficient knowledge in programming for windows system or app, so I also looked up for some online courses which I think I may follow during the holiday.
https://www.udemy.com/the-complete-windows-10-c-course-and-build-2-apps/
This will be useful for my own personal knowledge as well as for this FYP, courses are from Udemy and they usually have offer on their courses, I could just wait for the $5.99 or $9.99 sale.
If I were to use a car battery, I would get the deep cycle/marine battery as it is made for continuous discharge when connected to my computer within the robot/motors.
after the previous process, process 3
Thanks to Tiffany and Joan for the awesome snippets that were used as the featured image as well as in the videos, and helping with lighting up all the candles + documentations.
Also Thanks to Bao, Fabian and Suhwee for helping me to place all the candles onto the candle platform.
I’ve seen the actual project with real candles in the dark room the first time as the class saw it.
Light & Darkness was created to give a sense of alternative space within an ordinary space, in this project, I wanted to create a space so that the audience will forget that they were in the IM room, the room was build with pure darkness and stability in mind as nothing should collapse as I wanted to hang the candles platform instead of building a table-like structure to give a sense that all the candles were floating in space(which is also the reason why everything, including the candles platform was black), this worked really well as the candle platform would swing gently which increased the magical feeling to my space, Also, all of this will be enhanced in the digital version as I would add a series of stars(LED, not projection) to the room’s wall to increase the feeling of enchantment. The smell of the candles, as well the burnt smell created when extinguishing the candles worked much better than I expected, as I wanted to have smell that is not too overpowering, I’ve used 239 unscented candles with 60 scented candles, this combination worked like a wonder because there is only a faint smell outside of the dark room. another point that worked better than I expected was the random placement of the scented candles created a colourful candle array when they were all lighted up and this increased the overall aesthetic of this project’s analog version greatly.
After this analog version, I’ve noticed a few points to take note of, it is actually quite hard to light up all the candles so more lighters will be needed and the current lighter that I was using appears to be rather difficult to be used, so I’ll probably buy some more lighter which is much easier to be operated and uses lesser strength. Also, I would have to paint the metal wires that was used to hang the candles platform into matt black as the metallic would reduce the magical floating aspect of this project.
Continue from my previous Process 2.
again, I had progressed much further since that post, mainly in designing the workable electrical and mechanical system that could fit into the Telephone and to write the Arduino code up.
Since the previous post which I’ve roughly did the belt system that drives the music card into the laser reader, I had added the motor to the system and tried it, at this point, it seemed to work as I thought all I need was to slow the motor down and it will be alright.
After I cut the hole, I proceed to modelling the card slot and i took inspiration from the ATM just to have the user to have something they’ve experienced and know how to put the Music Card in without instructing them, since subconsciously, I assumed that they interacted with a ATM at some point in their life.
After the modelling which i am really happy with, I proceed to print it.
Since it was looking good, I went ahead and make a nicer LED system for it by soldering 4 LED(3 on the bottom and one for the top).
Next, I Epoxyed the speaker onto the bottom of the front belt drive since there is already a hole in the bottom shell for the speaker.
This is a 8 Ohm 0.5watt speaker that will be plugged directly into the Arduino.
I also Epoxyed the 4 LED into the card slot to prevent them from sliding around.
After an afternoon of panicking and finding alternative motor or even thinking to redesigning my whole belt system….
And then I found that I have a longer DC motor with metal gears built into it and i tried to figure our if I can incorporate this gear box into my current system, which is also rather impossible as the ratio for this gear box is about 1:45. when I only need about 1:5 to 1:8. if i use this, I will have the belt driver running too slow. 
same goes for this, but this is 1:250… even slower.
So to solve this problem, I tried to get the medium speed which is faster than what the song should be and will stuck about 30% of the time and removed the buttons (which detects card when user insert into it that trigger the motor to turn the belt.) that caused more friction. And I also jump start the motor by making it to spin at full speed for half a second to break the initial force required when the motor is starting.
It took me some time to sort out these messy wiring and make sure that none of the wires interfere with the track that the Music card is going through.
after trying out the workable speed of sound and getting stuck by removing the buttons.
and after this, I tried to code the majority of the code together.
For this, I did not expect to work this well and I am really excited about it!
I copied a few from online like Demons by Imagine Dragons, Harrypotter’s Hedwig Theme, and Pokemon Theme song, These were labeled on the card and those that weren’t labeled was What I composed myself. Since I have no music background, I did it by trial and error to give it a tune.
This was screen recorded when I tried to compose my 4th tune for this project:
after this was completed, I screen shot it and import into Illustrator to trace it into the Card layout which I made.
and this was how the cards were made.
Laser raster and cut in school on 2mm acrylic.
well, this is relatively simple, just type whatever I want on Webbased Text to speech reader and have it read it out in different accent and edit them in premiere pro to cut them up to the exact same length(9 seconds) and put them into the SD card within the Obseleting Instrument’s MP3 Decoder.
I really like the Japanese and Korean accent, its really funny!
Why did I made it to speak different accent? It was to engage the user and make them feel like there was really life in the system where they called/receive call from a real person, like if they discussed with their friend and their friend said that there was a Indian accent while what they heard was the British accent, they might want to try Obseleting Instrument for a few more time. The accent there is there to add variables in the system.
Throughout this Project, I’ve learnt many things like how to model objects in Tinkercad and make measurements properly, there are always failures in everything that I modeled before it works, and this is why 3D printing is a good prototype process where I printed it out and tested it to know if it work or not, if it doesnt, I will shave off some piece to see if it fits, if it does, I will make new measurements for the edited model.
Also, to have a Telephone case in the start and scale everything into the Telephone case was really a challenge especially at the start when I could not measure how big the internal was and could only make a guess and print some test print to try it out.
In this project, I realized that if I were to do a project that require multiple fields of knowledge like mechanical and electrical, It was better if I did not know how hard it will be, if I were to know that every part of the project will be something that I don’t know, I will be too afraid to jump into this project. I did something, realized that it doesn’t work and find solution to that single problem and proceed to work on the project and faced another problem, solving and learning one problem at a time lead me to the completion of the project.
Now that I had completed the project and looking back. Obseleting Instrument is really a complicated project as a whole, but thinking about it, I am just putting many small system into one project- like using one laser diode and a photo resistor as a switch, playing a tune when triggered, a physical button to sense if the phone was picked up, using a relay to control circuits of different voltage, running two DC motor at the same time and so on… Obseleting Instrument is just a collection of small systems, which I personally thinks was what made my journey of doing this project really interesting because I explored the basics of these components and learnt a whole lot through it.
Continue from my part 4 of this project updates, This is the final part which leads to the completion and is rather long as many things from getting the power supply, modelling, testing, painting, assembling to coding had been done in the past 2 weeks.
And then, into the process….
At the end of last post, I was waiting for the power supply to reach and it did during these 2 weeks so I’ve tested it out, The one that I bought is a AC-DC converter at 6V 20A when all I needed was 6V 8A, i decided to buy the higher Amp one just if i need more amp in the future, and I could also share the supply to my Interactive device project during the End of sem show.
I did some wiring afterwards.
Since the last post that I’ve decided to change the system in the legs to save space and shorten the overall length by stacking a smaller MG90S on the MG966R to act as a lever system to pull control the legs.
After testing this system out and seemed to be working, I merged it with the turtle leg that I’ve modeled in Zbrush, I don’t know about Zbrush before this project and it tool a long time just to model the shell, the legs and the feet.
I merged the leg with the previous test “rod-like” leg because I’ve already gotten the dimension in there so i just need to scale the Zbrush modeled leg accordingly to fit the “rod-like” leg.
At this point, I was wondering if I should change the MG966R(metal gears) into SG-5010(plastic gears) due to weight issue I might face later after adding the shell and so on, so I weigh the motors and decided that I should change the back leg to SG5010(but I changed it back to MG966R one day before submission due to the internal gears got loose in SG5010)
Major changes were made for whole layout of the base due to various reasons – the backleg will move differently from the front and since my project is basically a turtle with round shell instead of a flat one, it made more sense to use a smaller but higher layout rather than a flatter and wider one to make use of all the space within the shell.
Since my Zbrush was cracked, it crash rather often and I did the same thing over and over if I forgot to save frequently and hence it took quite long to model anything, but well~~
As for the shell… it took so much time and crashes to get it right, because the shell cannot be too thin(unable to print later) and cant be too thick (too heavy) and I cant find a function that could allow me to see the thickness (like reduce opacity of the material to see through), so everytime to check the thickness, I have to export into STL, and import into Tinkercad to check the thickness.
Everytime I made major changes to the shell, I’ll have to import to Tinkercad to check the thickness and shape.
this is the liner slider an an aluminium rod for the head system.
Printing for the base shell:
Printing for the top shell:
testing out of the head slider and to mark the length I need it to move and cut.
The mechanism to slide the head that I will use after cutting the rod to almost the size I need.
metal rod were epoxyed into the head to tie the elastic thread to control the tile of the head.
and a elastic thread was added to counter the tension created by the elastic thread to turn the head
After this, I added a few lines of code to substitute the button pressed with sound sensor to make the turtle more intractable and a few more actions that made it more timid looking (head peeking out to check the environment before the whole body come out.)
Overall, I really liked this project although I kind of regretted trying to make a turtle during the process because the shell is giving a lot of layout and movability problem made me kept thinking that if I were to make something without shell, it will be so much easier, but the reason that I first wanted to make a turtle is the best is because the shell naturally hide all components that break the user’s perception that this is a robot and a turtle is my spiritual animal, now that I finished the project, I am really glad that I sticked to my initial idea of making a turtle and persevere through all the problems I faced(mainly hardware and mechanical problems which I changed my system and design so many times.).
This was what made into my final turtle with many component being edited into the next version.
With more time, I am sure that I could code the turtle be able to be scare back into the shell by another sudden loud noise – I tried to change the delay in my code to Millis multiple times and it do every action in the sequence twice so I sticked to using delay for now, which disabled me to write the activation code for using the sound as a trigger since it is in a delay loop. But, it looked really nice for now even if there is only one sequence of movement and I am glad that people were thinking that the turtle is cute when they saw it.
After the Study model of the last part, where I stopped at the lever system and realize that it is not as strong as it should be, my improved method is to use 2 servo to do the same action, After a long time trying, I found out that the overall height of the servo will still be as high and as big as a standard MG 966R servo motor, and the MG966R will provide significantly higher torque than 2x SG90 at roughly the same height.
So my new solution to this is to use the heavier and stronger MG966R as the feet motor. Since they are relative big, I need a system to fully utilize the height and the movement of motors to make it as compact as possible. Since the center of the turtle shell is usually the tallest part, it make sense to stack the first rotating joint (the shoulder) above each other to greatly reduce the total size of the turtle by making it slightly taller.
Improved prototype of the system which reduces the shakiness of the Shoulder Motor(which is important as Shoulder is the first joint, if this vibrates, it will be amplified become very obvious in the feet.)
after countless of testing to get the perfect balance of height in horizontal and vertical, this system should be able to progress into what I will use in the final, there are still improvements to be made in the base height of the Shoulder Motor as when the Arm spins, it will still touch the ground.
for now, it probably can walk and hide like this..
Next up is to further refine this system and find the suitable system for the Elbow joint.
This is the first time I’ve seen the butterfly emerged from chrysalis, these chrysalis of lime butterfly were given to us by our friend YaYu.
These 3 different species were given by the conservationist Mr.Foo, there were a total of 7 Chrysalis, 5 Leopard Lacewing, 1 Plain Tiger and 1 Lemon Emigrant. All 5 Leopard Lacewing had emerged this week and the other 2 will be coming out tonight or tomorrow morning.
Did a lot of butterfly related task this week which includes getting the chrysalis from Mr.Foo, setting the chrysalis up for documentation, waking up way too early to get the video of the eclosing process. laser cutting and build the final casing for the butterfly and presentation, Researched on what butterfly eats and how to feed them. buying Curry leaves from supermarket with vegetables for the butterfly to live with. And tried to get a second hand monitor screen from an online seller, the seller did not turn up and did not reply. Our 160 degree wide angle web camera had arrived from china and we will test it out really soon with Max.
While I was writing this post, The Lemon Emigrant eclosed! (and I diden see it nor video it!) now that the Lemon Emigrant came out, the last one that will be coming out will be the Plain Tiger.
Also for coding, we further researched about our max patch and spend most of our time studying and trying to merge/tweak the standard patch as well as what other people did for their project, some of the patch were too hard to understand until today’s lesson when LPD explained most of what the patch does then we understands(slightly) what the patch does.
From this week to next week we will focus mainly on the max patch and get something that could roughly work in terms of butterfly/motion tracking, visual generation and output to control a real life appliance using UDP or the wireless I-Cube X.
Planned Milestone for Week 10-11:
Actual Milestone Done in Week 10 -11
We went to Dama in Ubi to purchase our acrylic to build the box for this project, the box have the length of 2.5ft by 1ft by 1( 76 x 30 x 30 cm ) and net will be attached to one side of the box for ventilation.
Helping the Chrysalis to shift the house from the insect container to a proper container for better documentation process and prepare for the timelapse during the emerging process.
The lid of the container was replaced with a net hotglued in place for proper ventilation for the Crysalis.
Also, I’ve joint a butterfly community in singapore and tried to ask for people to help me to get caterpillars of chrysalis of butterfly, and Mr Foo, a conservationist in Singapore agreed to give us some chrysalis for the project!
Our Project will take the idea of “The Butterfly Effect”
Basically for this project, we will be having many butterflies in a container with a habitat that allows them to survive for weeks, the butterflies will be used as an input for the camera’s tracking system and every time a butterfly fly over certain area, a visual will be generated through Max MSP.
The whole container will be placed on a specially made table with a projector in the table and a square hole directly below the container of butterflies, the bottom of the container will be opaque and the projector in the table will be projecting up to the bottom of the container through the square hole, making a screen like appearance at the base of the container.
For our project, we need to have real butterflies, after our research, we found out that Oh’ Farms will be selling caterpillars, 
on Saturday, we went there trying to purchase the caterpillars, however they were sold out and said the primary school purchased all of their caterpillars and they will restock in 2 weeks time, the shopkeeper also said that it will take around a month for the caterpillars to fully grow into butterfly, so unless we got the caterpillar this week, it will be not possible for us to have butterfly for the submission. So… Alternative plan is to go catch caterpillars/cocoon of butterfly ourself
This is Stick insect Oh’ Farms sell, I think its kind of interesting.
As for the butterfly, I think the best way is to find for butterfly in the park and instead of catching the butterfly straight, it will be easier to catch caterpillar or cocoon, therefore we went to Parks and look for butterfly, since there is butterfly of the species there, there will be its caterpillar nearby.
Acraea violae (Tawny Coster)
| Eurema hecabe contubernalis (Common Grass Yellow) |
Zizula hylax pygmaea (Pygmy Grass Blue)
The above photo was taken in Yewtee Park, I also found 4 cocoon from the trees and bush there, it might be the cocoon of moth, but nonetheless, if it look like butterfly, it will work
For this week Milestone,
We’ve got our idea out, found where to find butterfly/cocoon/caterpillars and got ourself some cocoon to try wether it will become a moth or butterfly, we also purchased some cheap mini webcam from the web and will use it for our project.
our planned timeline for our project will be:
Week 10-11:
Week 11-12:
Week 12-13:
Week 13-14:
Before 21st April:
upload everything we have onto OSS.
https://www.udemy.com/windows-iot/
https://www.udemy.com/the-complete-design-course/
to make use of the original button on the phone, I’ve figured that the 12 buttons runs on 2 different circuit which I could simply solder these number together and make all the 12 buttons into one button, so nomatter which buttons the user pressed, it will be registered as one button pressed.
Because I cut off the Redial button on the phone to make space for my belt driver system, I epoxyed the Redial button back to the case as there are no PCB supporting it.
I am really glad that this many piece worked well together and this was the biggest challenge.. since there are so many components working together (electrical and mechanical), even if one of the parts failed, it would not work as well as it is now. So I considered myself really lucky that the parts happened to work well even when there are misalignment everywhere.
And after the wiring, I did power test, it could easily run 8 servos simultaneously with nothing overheating. which is a great news for me!
This was quite the final base before I added the mount for the small backleg servo and the servo driver mount which will be attached it using screws.
Nothing is more satisfying in removing the support in one whole chunk(I did alot of cutting before this video so I can pluck out in one piece.)
this is the final mechanism for the head after I printed a small piece to prevent the wire from tilting too much when pushing the rod.
the head could be push and pull out nicely even before adding the string to control the tilt of the head.
The final test of the turtle before I finally started coding it. The head uses elastic thread because it will go forward and retract, since I dont want anything to be loose and might interfere with the shoulder servo motor, I decided to use elastic thread so that the thread wont be loose when the head is retracted.
This is the almost completed sequence of action,the turtle’s movement is quite restricted due to the shell and the back leg is unable to push the turtle forward because of the shell’s restriction as well as the weaker servo(MG90S) which is responsible for the forward and backward thrust, while the MG966R is strong enough to lift the turtle up, so the turtle could do movements up and down but not walk.
Prototype of stacking the Shoulder motors, I’ve realized that the left motor should be at the bottom and right motor should be at the top as one will spin clock wise and other will spin counter clockwise to make it the same height when their pivot are not centralized.
