Scioly.org

Bazinga+ wrote:

windu34 wrote:

NilaiVemula wrote:What are some effective ways to reduce skidding? We have tried balloons around the wheels, friction tape, and electrical tape, all with no avail.

Make your vehicle heavier

Making the vehicle heavier will not help because your maximum acceleration and decelleration depends only on the traction between the wheels and the floor, the mass of the car cancels out (a(max)= F/m=mgu/m=gu, where u is the friction coefficient) The only way to reduce skidding is to get high traction wheels (most everybody uses 'banebots' wheels, they work great) or to accelerate more gradually.

The force of friction will scale linearly only in a very ideal situation. With soft rubber treads that will deform as more weight as applied, the traction will go up faster than a linear rate when more mass is added.

You can also double up your wheels or increase their width, which doesn't increase the normal force, but still will increase your traction, which again goes against the standard classroom model for friction.

DoctaDave wrote:

Bazinga+ wrote:

windu34 wrote: Make your vehicle heavier

Making the vehicle heavier will not help because your maximum acceleration and decelleration depends only on the traction between the wheels and the floor, the mass of the car cancels out (a(max)= F/m=mgu/m=gu, where u is the friction coefficient) The only way to reduce skidding is to get high traction wheels (most everybody uses 'banebots' wheels, they work great) or to accelerate more gradually.

The force of friction will scale linearly only in a very ideal situation. With soft rubber treads that will deform as more weight as applied, the traction will go up faster than a linear rate when more mass is added.

You can also double up your wheels or increase their width, which doesn't increase the normal force, but still will increase your traction, which again goes against the standard classroom model for friction.

While this is true, the effect of adding more weight and getting wider wheels is rather slight, especially of your wheels are clean. Another thing that helped with traction is cleaning the floor before testing/running the vehicle and cleaning the wheels with alcohol wipes every few runs.

Bazinga+ wrote:

DoctaDave wrote:

Bazinga+ wrote: Making the vehicle heavier will not help because your maximum acceleration and decelleration depends only on the traction between the wheels and the floor, the mass of the car cancels out (a(max)= F/m=mgu/m=gu, where u is the friction coefficient) The only way to reduce skidding is to get high traction wheels (most everybody uses 'banebots' wheels, they work great) or to accelerate more gradually.

The force of friction will scale linearly only in a very ideal situation. With soft rubber treads that will deform as more weight as applied, the traction will go up faster than a linear rate when more mass is added.

You can also double up your wheels or increase their width, which doesn't increase the normal force, but still will increase your traction, which again goes against the standard classroom model for friction.

While this is true, the effect of adding more weight and getting wider wheels is rather slight, especially of your wheels are clean. Another thing that helped with traction is cleaning the floor before testing/running the vehicle and cleaning the wheels with alcohol wipes every few runs.

Like Bazinga said you could clean the floor before running the vehicle. I did this by using a swiffer and you would not believe how much dirt I picked up on the floor during competition and even testing. Another suggestion, if you don't want to get your wheels wet, is to use the swiffer to clean your wheels too.

dcrxcode wrote:

NilaiVemula wrote:Has anyone tried using a PID loop to correct for any error in their Arduino distance programming? If so, has it helped significantly decrease your distance score?

I wrote some PID code but didn't implement it because there's not much of a point when you can get centimeter accuracy using an encoder. From what I've read, tuning would also be a pain that wouldn't be worth the time.

What would your input to the PID controller be then? It would seem as though you'd use an encoder. In which case PID is a valid control scheme, but may not be optimal or advantageous given the parameters of the event.

andrewwski wrote:

dcrxcode wrote:

NilaiVemula wrote:Has anyone tried using a PID loop to correct for any error in their Arduino distance programming? If so, has it helped significantly decrease your distance score?

I wrote some PID code but didn't implement it because there's not much of a point when you can get centimeter accuracy using an encoder. From what I've read, tuning would also be a pain that wouldn't be worth the time.

What would your input to the PID controller be then? It would seem as though you'd use an encoder. In which case PID is a valid control scheme, but may not be optimal or advantageous given the parameters of the event.

Yeah, the input was the encoder - to clarify, you can get centimeter accuracy using an encoder without PID. PID is super cool, but wasn't advantageous for me.

Would it be better to use a brushed or brushless motor? I accidentally forgot to turn off the esc and it burned out . Now I'm just considering if it's worth the upgrade.

cubes wrote:Would it be better to use a brushed or brushless motor? I accidentally forgot to turn off the esc and it burned out . Now I'm just considering if it's worth the upgrade.

Honestly doesn't matter a whole lot unless your vehicle is really heavy (over 3-4 kg)

NilaiVemula wrote:Has anyone tried using a PID loop to correct for any error in their Arduino distance programming? If so, has it helped significantly decrease your distance score?

Usually skid is your main enemy in this event, so accelerating and braking consistently is key. I used a PID algorithm that used the formula (v-u)*(v+u) = 2ad to compute the braking acceleration, given current speed v, ending speed u, and the distance to stop, d. Basically, I establish some rate of deceleration, and from there determine the amount of braking required by the motor. I start with guessing the amount of braking I need, then comparing the actual distance I went to the calculated distance I should go within a given time period using the encoders (basically comparing actual speed to calculated speed). Then I dynamically adjust the variable that controls braking depending on whether the EV is going higher or lower than the desired speed.

I also recommend that you brake to a full stop about 10 cm before your final target. From here you should create a loop that cycles the motor between on and off cycles of about 50 ms until your EV reaches the final distance. This allows the motor to make minute increments of a mm (or less) before reaching your target, improving accuracy greatly.

Using this method I was able to reduce my 5 cm error to about only 1 cm or less.

cubes wrote:Would it be better to use a brushed or brushless motor? I accidentally forgot to turn off the esc and it burned out . Now I'm just considering if it's worth the upgrade.

I think brushed is the way to go. Brushed is far cheaper and has more than enough power to propel your vehicle 8 meters in 2 seconds or less (my vehicle was like 1.5 kg and it was fine). One of my team members also struggled with controlling brushless motors at slow speeds, and had trouble getting his vehicle to stop accurately. On top of that it's a pain reading output from an ESC.

Bazinga+ wrote:Here is my Electric Vehicle as well as the steering mechanism I used:

https://www.youtube.com/watch?v=pKC8MYR ... e=youtu.be

I highly recommend using the caliper steering method as it has worked very well for me and others that have used it, with the top 3 vehicles at MIT invitationals using this mechanism.

An alternative for the caliper is using a digital micrometer which may make mounting a 3D-printed part with a hole for the micrometer shaft and slot for a ball bearing much more convenient. It is also far more sturdy for mounting to the side of the vehicle and easier to adjust to the .01 mm accuracy that you will need for this event.