Long thread and some musings. On the state of technology today + what one can actually learn with minimal investments. /0

I remember a time - way back in 1990, where my monthly school fee was equal to that of a DC hobby motor in an electronic shop. If we asked for a motor, we would get it - but we couldn't really experiment with it [which would burn it down] - for we won't get another. /1

Electronics such as controllers or servos or radios were out of question. Could only dream of such stuff. I would go every year to the Trade Fair in Chennai, hoping to watch that mini train exhibit + the control line driven aircraft. used to drool at the tiny engine. /2

Remote controlled toys were far and few in between - and were too costly or too rare [often both]. Also, it took an enormous amount of work to get them going. Something not possible for a high school kid.

Say what one may, toys have a special place in everyone's heart.

Cut short to 2006 - I was out of India and was exposed to the burgeoning field of Remote Controlled (RC) vehicles. As is my wont, I just plunged off the deep end with RC Helis. These are just a beauty of engineering. One has to spend hours setting them up precisely. Then, fun. /4

This went on for a couple of years, and I got bored of it naturally. I wanted to do more .. can it fly on its own? Of course, it is possible - but can I do it? Having neither the engineering nor the math background, this always nagged at the back of my head. Fast forward 7yrs. /5

Around 2013, the RC bug refused to go away though I had moved several places and phases in life. So, I started reading up a bit more. I was surprised by how much was available for free. Figuratively, of course. You still had to spend time on internet. However, all that info! /6

It had been almost 10 years since my last serious "course work". I signed up online for something that would help me with both my work and my long term interests: Machine learning with Andy Ngo. What clinched it? He had a video of a self flying RC Helicopter :-D [I love them]. /7

// Andy [Andrew Ng] -- not Ngo//

Needless to say, I failed to complete the course. :-D .. but took two important lessons from that failure. 1) I am not serious if my money is not at stake [auditing isn't for me], and 2) I wasn't prepared enought to complete a ML course. /8

So - back to square one, and three more years later - I had completed two courses on basic statistics. I wanted to strengthen my basiscs. I paid for them and completed them.

But then, I had also used ML ad-hoc in many of my work related projects. Whatever little I learned +/9

+/ from the 2013 course held me in good stead. By then, the itch to do something about toys and electronics became stronger .. More reading.

I invested in an Arduino kit. I went with a big kit - for about 2500 INR in 2016 .. looking back, could have done with a lot cheaper. /10

And I can say this with all my heart: Arduino was the most fun I had with anything - EVER. Like serious fun. Have you ever wondered if you can measure something? Like light? Like temperature? pressure? current? distance? Can you control a motor? A stepper? Gosh, the rush!! /11

And all that fun for the cost of four tickets of a 3 hour movie? Hell throw the movie in the dustbin! Honestly.

And then slowly I learned what can be done what can't be with arduino - well not entirely, but I wasn't earning my income from arduino :-) .. so, I just engaged. /12

Speaking of arduino and other stuff -
https://robokits.co.in/  - I always found them very good with routine stuff. They respond well too.

By 2017, I had also bought a RC car - a proper 4WD chinese Traxxas knock-off. Don't judge me. No one in India makes this stuff yet. /13

Now, I had gotten confident of my ability to understand robotics. There was a sound in the background. I now recognize it as the gods laughing. Anyway - this time, I signed up and paid for the course in Robotics [UPenn, no less]. Coursera of course. Darned great course. /14

And of course, I failed to complete it. What else did you think?

But, I did successfully complete a part of the course that dealt with 2D motion. In theory, the theory was perfect. In practice, with a Chinese made RC car - it sucked. Really sucked. Like a vacuum cleaner. /15

Then the real beauty of those RC helicopters I used to fly before a decade hit me. I had one of them still in pristine condition, so I dug a big into the machine. What I found was not surprising - but a revelation nonetheless. One word: tolerance. Chinese product sucked. /16

The tolerance on the Taiwanese RC Helicopter was so good that there was barely any play between the parts. It had to be so, else the helicopter would wobble and self destruct. OTOH, the car wouldn't go on a straight line. A wise person would have given up then :-)

Not me. /17

I had Arduino, and I had an electronic controller + a car. So, I hijacked the controller of the car with the arduino .. attached a sonar for obstacle sensing and make it run in fields. .. few months later, I recognized that this can be done in a bigger size. [another story] /18

Move a year ahead, and I had moved a few more places and phases in life. All this excitement had to take a back seat. And it started again in earnest, in 2019 [summer]. by now, I had a proper industry grade EV prototype - crowd funded + bootstrapped. And I was still learning. /19

I collected odd items - motors, wheels, controllers, and text books .. for another smaller robot. By not, RC cars had become fun but slightly less interesting stuff. I wanted nothing less than my own toys. But, I had also learned the importance of thinking through designs. /20

What I found was that at a certain level, you run out of specific components that you need. When you try to create something that is in your mind, there are only two ways ahead: 1) you compromise with the vision, and basically go with parts that are available, or 2) make them /21

No - I am not saying it is some ground breaking stuff. And for most part, I like to buy off the shelf stuff for two reasons: 1) cost, 2) quality. A 4mm-6mm coupling made by an industry is going to be of a much better quality than if it is made by self. And far far cheaper. /21

But there are irritants along the way. You want your camera and your SONAR to be turning around the same axis and see the world with the same angle. Guess what? You are screwed. This is a limitation for serious DIY robotics people! Also - the cost of small scale components. /22

Steering back to the theme -- along the way, I read. I read every scrap of article I can get my hands on. Often times, I will read even research papers where I can't understand even 10% of what is there. But, the 10% is still something I can learn. Mostly I read DIY blogs. /23

I also upgraded my coding skills. Looking back, it was my laziness that stopped me from doing this a long way back. Should have done it in 2013. Anyway - better late than never. And I relearned 10+1 and 10+2 Maths and Physics. It is a treasure trove of knowledge, I tell ya. /24

All the story was to highlight one simple theme: it is incredibly easy to skill oneself in this world today. I would go as far to say it was never so easy to skill oneself as it is today. Unless you want to do cutting edge stuff, one doesn't have to pay for learning. /25

Today - it is incredibly easy to get our kids to learn very valuable skills, both engineering and research, for a pittance really. How do you set up an experiment? How do you control errors? How do you test? how do you reason it out? These used to be abstract questions in 90s /26

It is not just that instrumentation is more available and cheap - it is also that the programming has gotten remarkably easier! The combination of easy programming + instrumentation is a wonder that waits to be untapped. Kids will immensely benefit if they are interested. /27

So what do I think is needed? [pick and choose according to your interest]:

1) A single board computer like Pi - there are many choices today. Pi has the biggest user base. Between a awesome hardware with minimal user base, and a meh hardware with big user base, pick latter /28

2) A micro-controller board with all paraphernalia. In some cases one might find the Raspberry Pi / Jetson Nano to be sufficient as a controller. However, as the complexity increase, I find that a separate controller is a good thing to have. I pick Arduino for this purpose. /29

3) A BLDC motor + Electronic Speed Controller [ESC]. This is actually an interesting learning kit. Controlling a BLDC is fundamentally different from controlling a normal DC motor. Of course, the Arduino kits come with Stepper motors. Learn to work with them all. /30

4) learn to work with servos. This is actually trivial in Arduino - but controlling a fair number of them [like in a hexapod or humanoid robot] takes some ingenuity .. Arduino is a serial controller - no parallel processing. You have to think like a controller. Cool stuff. /31

5) if possible - ** if possible ** invest in these two things: a robotic arm. This is an amazing learning tool. I makes you look at the world with a different eye. You will start viewing everything in matrices :-)

Watching a series of numbers on a matrix convert themselves ++

++ into precise 3d movement in real world, is an experience. May be not a spiritual experience, but damn is it close?

The important thing with all of this, is that you start thinking of acquiring and processing information in realistic terms. You understand physical limits. /32

You understand what kind of info can be collected, and what can't be. You start thinking of what information becomes a proxy for something else [for instance, can sunlight levels in India be a proxy for temperature, given month?] Kids start asking interesting stuff. /33

Apart from all this seemingly geeky stuff, I also built a new found respect for those who can make stuff. Get your kids / treat yourself with a mini hand drill + a wood carving kit. A tool will beg to be used. If you can instill some discipline on using it, dividends follow. /34

Being able to make small but trick parts using wood / plastic sheets and then using them in your projects is a different level of zen. It teaches a lot of things - and forces the creator to think functionally. Aesthetics will follow - but functionality has to be conquered. /35

And it is all not costly. A typical middle class urban family spends Lakhs per year on their kid's schools. In semi-urban cities, it is to the tune of 1L per year. All of the things I have listed above don't come to more than 30K. And what's more? They can work for years.

For college students [engineers], my new found fad is a whole hearted recommendation - 3D printing.

Some of these technologies make me wonder - how did people do engineering without all these before? And the corollary: what else can be done now, w/ all this awesome stuff? /36

Veering back to the starting of the thread - I look back, and I think: perhaps there was no better time for someone to learn. Anything. Anything at all. If someone like me - with no formal math or physics background after 10+2 [25yrs ago] can do this, what can our kids do now?

More importantly - I think this is also the best time for people to kick themselves to action. Something as simple as making toys is made so much more cheaper and easier today. Cost of making is being driven low. Everyone gets to express themselves in solid material. /38

India has so much jugaad - and I have always been a detractor. But imagine: if jugaad can be standardized by modern tools like design + modeling, the scale up and adoption becomes so much faster. There is a lot to be done, and am honestly excited :-D //39 [or whatever] /Fin.

TL;DR version: read a lot, experiment, invest and learn for the joy of it. The world today is probably the best time and place to learn all sorts of interesting and fun stuff. No better time. Fail, and learn. You can afford to fail. Cost is minimal. Get kicking.

https://twitter.com/BrentBeshore/status/1353882450917797889?s=19

More succinctly, this ^^.