X-Drive directional drive control code release (kind of)

[Joey_24K]

I’ll keep this condensed into different dropdowns for different questions.

units

@mvas

Where do the -1 and 1 values come from? I’m trying to wrap my head around why there are no units involved here despite the speed of the motor being involved in the calculation. Speed is dependent on units. Are these values arbitrary and 1 is used for convenience?

This is correct. There are no real units like voltage or RPM, the range of -1 to 1 was used because you could just multiply the value by the max rpm, percentage voltage, or whatever you feel like using. You may have missed what I said in my last post, which I do not blame you for as it was rather long.

@Joey_24K

It outputs -1 to 1 for each motor, which you would then map to the minimum and maximum values for your motors.

interpolation

@mvas

Also, the purpose of the interpolation (in this case) is to find a balance for the motors to move at between turning speed and linear speed. This is because the robot physically can’t move straight and turn instantaneously at the same time. Otherwise, you’d just use a point turn and then drive forward, correct?. It sort of averages it out I guess (biased towards whatever R is).

This is correct. Instead of facing towards your target and then moving, you start moving towards it and rotating to your target rotation at the same time, similar to moves showcased here: (although with an x-drive and not mecanums ofc)
https://www.youtube.com/watch?v=S9eCk-3m7do&feature=youtu.be&t=31

turning ratios

@mvas

Related to the above questions, what would happen if you don’t use this ideal ratio? Would the chassis reach the ideal orientation before the endpoint and then just travel full speed linearly the rest of the way? Otherwise, it’d never reach the final orientation until the robot is at the target point?

There are plenty of ways to go about this. You could face the target point until close enough and then face your target rotation. This “golden ratio” so to speak ratio would look nice, but it it is not necessarily the fastest. (What if it is facing diagonally relative to the point and ends up driving there with just two motors?)

pid

@mvas

This makes a lot more sense. I think including this little description is the missing “key” (so to speak) for me to understand all of this. All the calculations shown above would be the only thing driving the motors. The PID would only be in control of the global variables. I always wanted to make a pure pursuit program with an X drive, but I never could wrap my head around the math needed for an X-drive to follow a curved path. This seems like a good way to go about that.

Glad you got it! Probably should have made the importance of Smore clear, in hindsight. Even odometry and all this fancy directional trig will be useless without well-tuned PID. (At least for precise movements.)

P1, P2 , and s

P1 and P2 are the values calculated for just movement. P1 is for front right and back left wheels, P2 is for the front left and back right wheels. These serve as the “V1” values when interpolating from movement to rotation.
s is set to the larger of the two values P1 and P2 so that both can be divided by s later. This ensures that the largest P value is equal to 1 (or to the global var S, since s gets divided by S)

@mvas

After having typed all this out, I think I have answered some of my own questions

Your final conclusion is correct.

---

@mvas

is there any external resource you used or would recommend for more of the theory behind the math (with visuals and such)?

The only things I can think of would be 5225A’s position tracking document, which I’m sure you’ve already read, or the YouTube channels 3Blue1Brown and Numberphile.