Self-balancing robot from scratch

Рет қаралды 15,173

Күн бұрын

This was my attempt to make a self-balancing car all by myself, without using kit, or reading equation online, or downloading program codes. The purpose is to challenge myself on real-time embedded system hardware and firmware design, and my understanding of physics.. As I didn't look at others' solution (including KZbin), I posted first, then realized many amazing solution in KZbin too. While my result may not be that fantastic, bear in mind that i did it from scratch.

Пікірлер: 15

@64findtheway 6 жыл бұрын

You've made it from scratches - my respect and good luck with your next projects :)

@SiliconSoup 6 жыл бұрын

Thank you. It was much harder than I thought, without look at any reference at all. It was after posted this video I found out many robots in KZbin move much better than mine. But anyway, i had fun.

@allawhussein 4 жыл бұрын

highly impressed by your patience

@messierthirtyone777 2 жыл бұрын

Can you please do another tutorial of the derivations? Thank you

@nigeljohnson9820 6 жыл бұрын

This is very impressive work. We take for granted those two wheeled vehicles that allow someone to stand upright on them. It is interesting that the PID failed to control the balance. While your physics equation works there are clearly more complex equations that are used on the bipedal robots. Could it be the limitation in your solution was the sampling rate or the accuracy of the balance sensor(s). Where they level and accelerometers? A lack of feedback gain might explain the performance. Did you apply any corrections for the digital sampling effects in the feed back loop?

@SiliconSoup 6 жыл бұрын

I build the hardware and software from scratch (almost). When the balancing didn't work and both hardware and software were unproven, and the effect of CG, sensor location, sensor accuracy were all unknown to me at that time, it was extremely difficult to solve the problem. Most importantly I didn't even know whether PID algorithm would inherently stablise the car and I made it a point NOT to search for answers from the internet and spoil the fun. As PID was only my first guess, I would not hesitate to explore other methods if I got stuck for too long. Now that I have seen videos of this sorts on the internet, obviously I come to realize that PID works. Had i knew in advanced that PID worked i would simple stick to it instead of deriving balancing equation from Newton laws. When I used Newton equation to balance the car, I also increased the sampling rate, and it is probably the low sampling rate that failed the PID algorithm. Thanks for pointing out there are more complex equations out there. If I want to have continue this project at a more complicated level, I will definitely look up for these equations, download code and have fun in a very different way from this first attempt which is to enjoy building something totally from scratch. Thank you for your comment.