Acceleration deacceleration on straight(), Wheel delay, and adjusting built in PID values

[techcatgato]

Hi all! I have encountered a strange problem. I've made a custom PID loop using robot.drive() with my custom acceleration deacceleration, but when I use it I see that one wheel always starts a noticable while after the other, which I think is hindering the performance of it. However when using robot.straight() it doesn't happen so i've a few questions:

1. Is there any fix to the problem with robot.drive()?
2. how can i customize the acc deacc curve for robot.straight()?
3. Can I tweak the robot.straight PID values if I use the gyro drive base?
   Code:
   acc deacc (x is percentage of the route done):

        if x < 0.5:
            speed = vmax * (2 * x**2)
        else:
            speed = vmax * (1 - 2 * (1 - x)**3)
        error = gyro.heading() - start_angle

PID using robot.drive()

def pid(dist, kp=1, ki = 0,  kd=0):
    print("pid started")
    finish_dist = robot.distance() + dist
    start_dist = robot.distance()
    vmax = 500
    start_angle = gyro.heading()
    integral = 0
    last_error = 0
    while robot.distance() < finish_dist:
        x = (robot.distance() - start_dist) / dist
        if x < 0.5:
            speed = vmax * (2 * x**2)
        else:
            speed = vmax * (1 - 2 * (1 - x)**3)
        error = gyro.heading() - start_angle
        if abs(error) > 10:
            kd = 2
        else:
            kd = 0
        p = 0 - error
        pfix = p * kp
        if error <= 0.01:
            pass
        else:
            integral = integral + error
        ifix = integral * ki
        derivative = error - last_error
        dfix = derivative * kd
        correction = pfix + ifix + dfix
        speed += 0.1
        speed = speed * 200
        robot.drive(speed = speed, turn_rate=correction) 
        last_error=error
    robot.stop()

[laurensvalk]

The drive base functions all use their own PID controller, so this code builds another on top of that. Two nested or conflicting controllers probably don't give you the desired result. If you'd like to build your own controllers, it is better to power the motors with motor.dc(value) so only your own controller if active.

Alternatively, you can change the P/I/D for the existing controllers to achieve a similar effect.

---

Then something to think about 💡

error = gyro.heading() - start_angle

...

 derivative = error - last_error

Numerical derivatives like these can be quite inaccurate. What is the mathematical expression for the time derivative of error in your case? Can you compute it exactly (hint: yes 😄)?

[techcatgato]

Oh alright. Are the constants from distance_control.pid() the ones used in robot.straight()?
And how can I change the speed for robot.straight() mid action (if possible) ?
Thanks in advance!

[techcatgato]

Oh alright. Are the constants from distance_control.pid() the ones used in robot.straight()?
And how can I change the speed for robot.straight() mid action (if possible) ?
Thanks in advance!