I have some s-curve implementations on my channel, all done with just basic additition/substraction. My implementation goes all the way to Jounce which controls Jerk and Jerk controls acceleration etc... Gives nice s-curve motion profile with simple math. When the robot tries to move, inertia keeps the top of the robot standstill and the robot tilts to the opposite direction, make the tilt angle dependant on the set speed and increase/decrease Jounce accordingly. I'd like to know if my "PID:less" control system could even work before making my custom CNC controller based on it... :)
Using 2 nested PIDs is a perfectly ok way to control the thing and it's one of the ways how sway less cranes work. The only negative thing you may have is tuning the PIDs both loops effect each other and so you have to balance how fast one of the PIDs react. Also how fast the attitude PID react will effect the acceleration of the car do to the physics of the system there is 2 ways to accelerate the car one is rapid the other is more gradual.
Yeah, now we're talking. Are you only going to use one BLDC motor or do you plan on adding another so it can steer? These kinds of projects always seem somewhat straight forward until you actually dig into it, then it becomes so much more complicated. Can you get into some details next video about how you control the ODrive (I'm assuming it takes its input in Steps) and how to drive it using an arduino? Waiting patiently for part 3. Great job.
@@LittleRainGames yeah, but driving at 50mph would require extremely fast adjustments and a balancing robot can't make adjustments without having updated and accurate position/tilt data. For comparison, a balancing robot is far more sensitive than for instance multirotors with regards to this. I honestly don't think the mpu6050 and an 8bit 16Mhz microcontroller is up to this task. It will be awesome if it works though.
Specifically regarding the tall, floppy chassis, are you certain that your assumption about keeping the concentration of mass near the top of the chassis is correct? In your previous video, you used the example of balancing a tall object on your palm, and you said it was easier if the object was more massive at the top, because it gives you more time to adjust the support point in response to movement at the top of the shaft. Having most of the mass at the top seems to be what is causing your vibration problem. It seems to me that instead of keeping the mass as high as possible, you'd want to keep it as low as possible, to keep the center of mass as close to the support point as possible so as to minimize the distance that the support point has to move in order to get back under the center of mass. Ideally, if the majority of the mass (and therefore the center of mass) was located below the support point (axle(s), in this case), the machine would be self-balancing. So wouldn't it make the most sense to have the accelerometer located as high on the chassis as possible, the gyroscope located as close to the center of mass, and the majority of the mass located as low as possible in order to maximize the sensitivity of the relative systems, and minimize the required reaction distances? Just a thought …
The thing whit having the mass on top is about the center of moment of inertia. Having the mass on top gives you more time to correct the angle and attitude of the board . It's true that having the mass in the centre of the wheels the board would center perfectly but that's not possible physically as you will always have some weight from the board. Just think about it this way the large mass on top of the stick will make the top of the board react slower and that will give the controller more time to correct the tilt
@@masha_kasha5061 Having the mass at the top of the stick may make the top of the board react more slowly, but it will increase the distance that the wheels have to move and will increase the speed at which they will have to move in order to compensate for changes in position of the top of the board. Rapid changes in position are what causes the oscillation due to the flexing of the stick. Stiffening the stick is one way of overcoming this (for instance by providing a cross member up the middle, or up both sides, creating an X- or I- beam shape), as is shortening the length of the stick. If the center of moment of inertia is lower, however, the stick becomes less "whippy". The top might move out of alignment more easily, but the stick would also have less resistance to bringing the top back into alignment. This is why the original Segway had its battery packs and most of its electronics below the level of the axles.
@@grantcampbell6026you are not wrong and all said is legitimate. My idea wasn't to comment on the oscillation specifically cos its pretty obvious that the design isn't the most rigid of things especially when you trying to move the thing normal to the board the cross section isn't the best of that thing. As for the center of mass it all comes down to what do you want to do if you prefer better reaction times you should have it lower if you prefer a bit more time of reaction for balancing make it a bit top heavy. But in all the idea of having all the weight in the axle isn't that great. That's all said if the length of the board isn't changed . Also about the Segway I'm not to familiar whit them but I assume having the weight under the axle would make whole system stable .
@@LittleRainGames the accelerometers role is to provide angle adjustments to the gyro based on the gravity vector. By having it anywhere else but on the axis of the axle you introduce additional errors. It doesn't matter for the gyro though.
I think if you wanna drive it with an rc transmitter you can just change the PID target angle or setpoint based on the value of the RC transmitter. This way the PID just adjusts to your desired angle... Not sure if you will mention it later in the video haven't completely watched it yet.
Open loop for balancing isn't a option especially if you trying to make the thing stand up right. Open loop controll is more suited in places like for systems that are more stable or edge cases like pendilums where you can use input modifications to nullify the sway of the pendulum
If you get it to work, can you please affix the tentacle machine to the top of it? Just because.
I have some s-curve implementations on my channel, all done with just basic additition/substraction. My implementation goes all the way to Jounce which controls Jerk and Jerk controls acceleration etc... Gives nice s-curve motion profile with simple math.
When the robot tries to move, inertia keeps the top of the robot standstill and the robot tilts to the opposite direction, make the tilt angle dependant on the set speed and increase/decrease Jounce accordingly. I'd like to know if my "PID:less" control system could even work before making my custom CNC controller based on it... :)
Using 2 nested PIDs is a perfectly ok way to control the thing and it's one of the ways how sway less cranes work. The only negative thing you may have is tuning the PIDs both loops effect each other and so you have to balance how fast one of the PIDs react. Also how fast the attitude PID react will effect the acceleration of the car do to the physics of the system there is 2 ways to accelerate the car one is rapid the other is more gradual.
So.. I feel like you're saying, it's not so simple...
Pretty epic project man! Great stuff!
Gave a like after 30sec. Great mindset
Just have the base balance change with your input then allow it to correct it will cause it to move if I'm right
Is he okay? Hasn't made a video in a long time.
Yeah, now we're talking. Are you only going to use one BLDC motor or do you plan on adding another so it can steer?
These kinds of projects always seem somewhat straight forward until you actually dig into it, then it becomes so much more complicated. Can you get into some details next video about how you control the ODrive (I'm assuming it takes its input in Steps) and how to drive it using an arduino?
Waiting patiently for part 3.
Great job.
There is going to be 2 Odrive motors. The Odrive can be controlled by a serial bus, USB or step direction. I think ill use the serial bus if it works
I don't think the mpu6050 is able to provide the sampling rate and accuracy needed for it to balance successfully in 50mph, but it's worth a try.
do you have an idea of what you use for this application, I really need it...
The mpu6050 can only find acceleration up to 1G, i dont think it matters for speed.
@@LittleRainGames yeah, but driving at 50mph would require extremely fast adjustments and a balancing robot can't make adjustments without having updated and accurate position/tilt data. For comparison, a balancing robot is far more sensitive than for instance multirotors with regards to this. I honestly don't think the mpu6050 and an 8bit 16Mhz microcontroller is up to this task. It will be awesome if it works though.
Specifically regarding the tall, floppy chassis, are you certain that your assumption about keeping the concentration of mass near the top of the chassis is correct? In your previous video, you used the example of balancing a tall object on your palm, and you said it was easier if the object was more massive at the top, because it gives you more time to adjust the support point in response to movement at the top of the shaft. Having most of the mass at the top seems to be what is causing your vibration problem.
It seems to me that instead of keeping the mass as high as possible, you'd want to keep it as low as possible, to keep the center of mass as close to the support point as possible so as to minimize the distance that the support point has to move in order to get back under the center of mass. Ideally, if the majority of the mass (and therefore the center of mass) was located below the support point (axle(s), in this case), the machine would be self-balancing.
So wouldn't it make the most sense to have the accelerometer located as high on the chassis as possible, the gyroscope located as close to the center of mass, and the majority of the mass located as low as possible in order to maximize the sensitivity of the relative systems, and minimize the required reaction distances? Just a thought …
The thing whit having the mass on top is about the center of moment of inertia. Having the mass on top gives you more time to correct the angle and attitude of the board . It's true that having the mass in the centre of the wheels the board would center perfectly but that's not possible physically as you will always have some weight from the board. Just think about it this way the large mass on top of the stick will make the top of the board react slower and that will give the controller more time to correct the tilt
@@masha_kasha5061 Having the mass at the top of the stick may make the top of the board react more slowly, but it will increase the distance that the wheels have to move and will increase the speed at which they will have to move in order to compensate for changes in position of the top of the board. Rapid changes in position are what causes the oscillation due to the flexing of the stick. Stiffening the stick is one way of overcoming this (for instance by providing a cross member up the middle, or up both sides, creating an X- or I- beam shape), as is shortening the length of the stick.
If the center of moment of inertia is lower, however, the stick becomes less "whippy". The top might move out of alignment more easily, but the stick would also have less resistance to bringing the top back into alignment.
This is why the original Segway had its battery packs and most of its electronics below the level of the axles.
@@grantcampbell6026you are not wrong and all said is legitimate. My idea wasn't to comment on the oscillation specifically cos its pretty obvious that the design isn't the most rigid of things especially when you trying to move the thing normal to the board the cross section isn't the best of that thing. As for the center of mass it all comes down to what do you want to do if you prefer better reaction times you should have it lower if you prefer a bit more time of reaction for balancing make it a bit top heavy. But in all the idea of having all the weight in the axle isn't that great. That's all said if the length of the board isn't changed . Also about the Segway I'm not to familiar whit them but I assume having the weight under the axle would make whole system stable .
Subscribed very cool
put the sensor down by the wheels to remove the flex oscillations.
But then the position of the top of the car is unknown. It’s not that simple
@@GearDownForWhat you can derive that from the angle, no?
@@SirArghPirate yes, but acceleration would be different. Although you could calculate as long is it isnt directly on the axle.
@@LittleRainGames the accelerometers role is to provide angle adjustments to the gyro based on the gravity vector. By having it anywhere else but on the axis of the axle you introduce additional errors. It doesn't matter for the gyro though.
I think if you wanna drive it with an rc transmitter you can just change the PID target angle or setpoint based on the value of the RC transmitter. This way the PID just adjusts to your desired angle... Not sure if you will mention it later in the video haven't completely watched it yet.
But what kind of gears did you print to drive the wheels?
**dab-sneeze**
The parrot sumo does this..
Dude great job keep it up, it's awesome even if murf?
Open loop, versus , closed loop
Frame is not rigid , that's the biggest problem
Open loop for balancing isn't a option especially if you trying to make the thing stand up right. Open loop controll is more suited in places like for systems that are more stable or edge cases like pendilums where you can use input modifications to nullify the sway of the pendulum