My Wine-Glass Carrying, Self-Balancing Inverted Pendulum Robot 

Thanks!

😂 exactly!

Thanks Dylan! All has been well thanks, I wish the same for you!

Thanks for checking it out!

Thank you!

Thank you!! Appreciate it!

Thanks! Appreciate it!

There is! www.shaysackett.com/inverted-pendulum-robot/ I have also linked to the github for this project in the video description that contains all the 3D models and code.

Not sure, never tried it!

Thanks I appreciate that! Good question, I'm using the Arduino Mega's DC jack which has a built-in voltage regulator and can handle a 7-12V input. The LiPo battery I'm using is a 1S 3 cell LiPo, with a max output voltage of 12.6 volts at full charge. 12.6V is a bit higher than the 12V jack the Arduino is rated for, but I took the chance that there would be some wiggle room left in that 12V spec, and so far it's worked out ok for me!

I have been interested in robotics since I was a kid, but I first started building stuff after joining my high school's robotics team. I would recommend choosing a project you're interested and curious about, and start building something! When I began this project, I had little programming experience, and I wasn't sure I would be able to make it balance, but I took it one step at a time and focused on solving one small problem at a time, and eventually got to the point you see in this video. Stay persistent, don't be afraid to fail or switch approaches, don't rush yourself, and try to focus on "the journey", as cliche as that sounds, not the end product. It's easy to get caught up in fantasizing how cool the final product will be, but at least for me I've found most of the enjoyment of these projects comes from what I learn and trying to solve tough technical problems. Hope that helps! Good luck!

Thanks! The parameters for the control loop were tuned through testing. I made a wired remote with 3 potentiometers that allowed me to tune the P, I, and D terms while the robot was balancing which made things very easy. I elaborate more on this remote and my tuning procedure in my blog post I linked to in the comments if you are still curious.

@@ShaySackett excellent idea! I might try that myself!

@@utetwo9709 Thanks! Yes it's well worth the time and effort, I guarantee it will save you SO much trouble trying to tune the PID algorithms! What I found is if I couldn't tune the PID algorithms to have sufficient performance in about five minutes, something was wrong with my code. I think that's a good rule of thumb to have. It's not worth spending hours or days trying to get a PID algorithm tuned only to discover your code is the real underlying issue.

Never weighed it deliberately, but it can balance a water bottle off-center of the robot pretty well.

Thanks I appreciate that! No I did not have to change the PID tuning when the robot was loaded! This was actually a direct result of using 2 PID loops. The first PID loop controls the speed of the motors directly, and commands the motors to speed up or slow down to achieve a set angle (in a perfect world this would keep the robot perfectly stable and balanced). However, we don't live in a perfect world, and I found my robot had a tendency to drift and "run away" over time, and build up momentum in one direction or the other; I discuss this more in the blog post I linked to in the comment section if you want more details on why that happens. Regardless, I added a second PID loop whose job it is to command the robot to always have a linear (forward or backwards) velocity of zero. The second PID loop does this by commanding the angle of the robot to change to counter and motion forwards or backwards. This second PID loop sets the desired angle of the robot, which is the input to the first PID loop, so in effect the first PID loop is "following" the commands of the second PID loop. What this means is that this robot is not 100% reliant on having a good PID tuning to achieve balance; if there is a load that is off center or drags the robot one way or another, the robot realizes it is speeding up, as an example lets say the robot is leaning and moving to the left, and as a result that "second" PID algorithm I mentioned commands the robot to tilt to the right to slow the falling/lean to the left. In this way, the robot can achieve equilibrium when loaded with objects it was not expressly programmed to carry. I never programmed it specifically to carry any of the objects I showed in this video; that was all a happy by-product of fixing the problem of my robot's tendency to "run away". I hope that answers your question! I got long winded there, but that's a pretty thorough explanation of "why".

@@ShaySackett Great explanation. Thanks. I think that approach is called cascade control.

@@basicmachines I did not know that, thank you for that information! I will have to look into that. I think you also left a comment on my blog the other day recommending a controls bootcamp playlist; those looked very good I'll definitely be watching those videos later!

Do you have encoders on your robot? I also ran into this "run away" problem. Check out my blog post in the description, I go into more detail there under the "MPU-6050 Software and Complimentary Filters" section. Likely what is happening is caused by the fact the robot has a sense of orientation from the IMU, but no sense of speed. So pretend the robot leans to the right a bit, the motors will speed up to the right to correct for the lean, but now the motors are going faster to the right but the robot is not aware of this and this can create a feedback loop where the motors just speed up and eventually can't keep up and the robot falls over. No amount of tuning the PID loop will help here. This is why I added encoders to my robot. Add encoders, then add a second PID loop that controls the speed of the robot. So the goal of this robot is not only to balance, but also to maintain a set speed, say of zero. The speed PID controller controls the angle of the robot to bring the robot speed to zero and prevent runaway. As a side-effect, this is also the first step to getting the robot to drive forwards and backwards. Hope that helps!

@@ShaySackett nice. I'll check it out

If I was willing to spend more money, I would have bought motors with better encoders, or bought dedicated encoders that had more resolution. I think what was is limiting the quality of the balancing right now is largely the fact the the encoders are cheap and the DC motors with their gearboxes have a fair amount of backlash.

@@ShaySackett Thanks for the advice!

Yup! Better late than never I hope, I added all the files related to this project in a link in the description. I haven't looked at it in a while so I hope its ok. Have fun! Good luck with your project I learned a lot with mine.

If you want more detail I would recommend checking out my blog post I linked to in the description. It's too much to explain in a single comment.

Shoot me an email, I have it listed both on my about page on my youtube channel, and on my contact page on my website. I'll see if I can help you out. I apologize for the late reply

Mechanical Engineering!

In this project I used two PID controllers.

🤣HA thank you for the comment and the cultural lesson! I'll keep that in mind when I visit France in the future, wouldn't want to have a faux pas.