Cheap balancing robot can now right itself!

In this video I explain some of the challenges in getting a balancing robot to right itself. Besides having a centre of gravity that is sufficiently low, the motors and electronics driving them should be able to deliver a high peak torque. It took some time to get the electronics being able to do so; in this video I explain you why.
If you have any questions or suggestions: comments and feedback are appreciated!

[@yoursfithfully047]

Nice thinking man , cool project

[@YaseenExperience]

❤❤❤❤❤❤

[@AdityaMehendale]

Interesting "debugging explanation" Thanks!
For the bypassing of the current-sensing - Does the resistor mean that the "protection" features of the circuit 4:35 are no longer active? Aren't you concerned about over-discharge, reverse-polarity (unlikely) or over-charging?
On a related topic - would it help (instead of adding a bypassing jumper) to add a second pair of MOSFETs in parallel to the existing pair? This would reduce the drop withouting defeating protections.

[@kloppertje]

Thanks for the comment :). My idea was that the resistor works as a voltage divider. I have *assumed* the protection features are still active but have not tested this (shame). This is bad design; however at the moment I'm using a protected 18650 cell so this doesn't worry me. For later on, when I'll be using an unprotected 18650, this'll be definitely relevant.
A second pair of MOSFETs would definitely be a solution; would even be possible on a small footprint :).

[@Blacksmith52]

dollar tree boston dynamics
:)

[@zoeyzhang9866]

Nice one! Wondering if any custom PCBs and 3D printed case may help to realize the final robot? If so, we'd love to supply free services and witness the result! (PCBWay zoey)

[@tolkienfan1972]

Maybe a capacitor to provide low resistance higher current supply during the large load spike?

[@kloppertje]

Thanks for your reply! I did a quick experiment with this; a quite large capacitor would be needed. A solution could be to have a diode followed by a buffer capacitor to power just the ESP32. However, this would be a "patch" instead of solving the underlying problem (voltage drop of the source); I'd rather solve the underlying problem than patching it. Experience has learned me that stacking patches instead of solving root causes leads to unmanageable complexity.

[@tolkienfan1972]

You might want to supply the microcontroller via a separate power supply with a filter for conditioning. Inductive loads, like motors, can cause big voltage spikes, positive and negative, which can play havoc with digital electronics.
Cool project. Good luck!

[@kloppertje]

@@tolkienfan1972 I'll keep it in mind and test for voltage spikes when making a PCB for this project. Thank you :)

[@abdulwaheedouabihi5903]

have you uploaded the code for this, would be interesting to see what the robot is doing during the self righting, i assume you turn the motors at max speed for some short period of time to get the robot close to vertical and then "turn on" the pid balancing again.

[@kloppertje]

Thanks for mentioning; I'll try to push the latest code to the git repository this weekend. The code isn't spectacular (on purpose); I simply turn on the PD controller while the robot is lying on the ground. The code is here: gitlab.com/kloppertje/balancingrobotdc/-/blob/devel/src/main.cpp?ref_type=heads#L159 but since this video I made some changes to it regarding self righting. I'll notify you if I push the code

[@abdulwaheedouabihi5903]

@@kloppertje just another question, so if all you are doing is turning on the pid controller while the robot is down, why does it not automatically self right after it falls if you don't turn off the pid controller

[@kloppertje]

@@abdulwaheedouabihi5903 To be exact: I simply disable the motors, the PID controller keeps running in software while the robot is lying down :). There is a state machine in the code. If the angle is below a certain value, the robot is probably lying flat. The robot is enabled either when it is near vertical (so, someone put it upright), or when a self-right command is received

[@kloppertje]

@@abdulwaheedouabihi5903 The version of the code used for this video is now pushed, see gitlab.com/kloppertje/balancingrobotdc/-/blob/devel/src/main.cpp?ref_type=heads#L190

[@abdulwaheedouabihi5903]

@@kloppertje thanks