Re: Desensitizing Joysticks
 

hey, look at the post time on my reply ;)

Re: Desensitizing Joysticks
 

My most productive coding hours are between midnight and 3am, and I can't explain why....yawn....why I'm so tired today... :yikes:

Re: Desensitizing Joysticks
 

That is very helpful file.

What i was thinking is, have a max change function in real-time like you mentioned.

Basically you jus need a variable that holds the current joystick input. For example if your joystick was at 135 and you pushed it to 250 all of a sudden, You can check you current motor output which was 135 and then current joystick input which is 250, now you keep adding a certain number lets say 10 to the motor output untill it reaches 250.

This is one way to do it, the other way is to subtract 250-135, you get 115 devide it by a number lets say 10 so you get 11.5 , now ur max change is 11.5 , add that 10 times to your motor output or until it reaches desired position, this way you will have variable max change function.

Re: Desensitizing Joysticks
 

But, your robot doesn't move when you give it a small value, because of friction and other factors. In '99 we did some tests with our robot, and you had to give it something to the effect of 145 to get it to move from a stop. So, we programmed in a deadzone from 117-137, and then starting at 138, we scaled the values from 145 to 254.

Re: Desensitizing Joysticks
 

Ahhh, good point Joe. Note though, that the nonlinear function I gave essentially has that deadband built in (very low response to joystick inputs around neutral), but it's a smooth and continous transition to the "on" state, instead of being abrupt. I guess once the bot is built up and you're testing it to see how much motor output is needed to start moving, you could tailor the transfer function a little to get it to "turn on" at that motor output. The smooth nature of the function takes care of the scaling from that point up to the max.


This works too. But I think you should be careful about remembering values from loop to loop and using them to calculate motor output. Suppose you're in the middle of changing the motor output as you suggest, but then the driver suddenly moves the stick full the other way. Now your math might suggest that the max change to the motor is huge, and you might get a big drive current spike as the motor responds. If you're going to do this, just be careful about how you write the calculation so that you don't over-juice the motor.

Re: Desensitizing Joysticks
 

What we use is a exponenetial function. This simply shifts the power curve out some while still retaining full power at the edges. Instead of it being completely linear, it follows a curve (that looks suspiciously like a cubic function). At 50% joystick travel, you may only have 33% motor output, but at 100% joystick travel, you have 100% motor output.

Another technique that I learned from flying R/C Heli's is to grip the joystick at the bottom with just a few fingers....This helps to make movements less extreme....they only downside is that it is a little tuffer to get to the buttons...The advantage is that the movement comes from the fingers instead of the arm and wrists, and it affords you alot more accuracy.

Bill

Re: Desensitizing Joysticks
 

This is what I came up with during our team's meeting today after school:

/*
pwm is a jostick value, 0-254
pwm_in is the pwm_out of the default code
pwm_out is the new value going to the joysticks
null_zone_swing_fraction is the amount you'd want the joystick to move without the motors kicking in.
*/

pwm_out = (pwm_in / (1-null_zone_swing_fraction)) - null_zone_swing_fraction(1 - null_zone_swing_fraction)

So. Its quite simply a y = mx + b "transfer" function. enjoy.

Re: Desensitizing Joysticks
 

Exponential Function: simple sin(x) function:

amplitude = 2 * 127 (pwm max)
period = 4 * 127

make sure, as i forgot to in my previous post (above):

pwm = pwm - 127;

That way the origin is the middle of the null zone. makes for nice calculations.

Re: Desensitizing Joysticks
 

The motors have gained more power since '99.
The new drills introduced last year have almost no delay at low power values.

The deadstick area is still nice, but you also need to scale the output rather than start at 138.

Re: Desensitizing Joysticks
 

You could use low pass filter. I'm at school right now and don't have access to what I've written, but I'll try to post it when I get home.

Re: Desensitizing Joysticks
 

This is what I love about programming...there are so many different ways to solve the same problem. All of these are nest ideas. Good luck everyone!

Re: Desensitizing Joysticks
 

I've posted the low pass filter on the white pages. It's called low pass filter. (Duh! :))

An example usage:
Basically, you call this function with what your current speed is, where you want it to be, how fast you want it to get there (higher numbers are faster), and how large you want the dead zone to be (i.e. if the joystick is at 129, you probably want to be at 127. You can set how far from 127 it will be corrected with this param.). It returns what value you should set the PWM to next.

By calling this function multiple times you get a nice gradual increase in output values, so if you move the joystick as fast as you can in circles, very little of that will be picked up. How sensitive it is to input is set by you. The exact equation for finding what to set sensitivity to:
255
-------------- = t
(1000/26.2)*s

t is time in this equation. deterimine what time (in seconds) you think is best for it to be able to go from full reverse to full forward and plug that equation. Solve and use that as your input to the function, ignoring any decimal portion. For instance, if t = 1 sec, then s = 6.

Note that the equation assumes it is being called every 26.2 ms.

Post again if you have any questions about the code or if you'd like to comment on it.

PS I had to re-write it because somebody took my disk that I had that on, so it hasn't been complied. It should be right, though. :)

Re: Desensitizing Joysticks
 

Beware of low pass filters. We used an LPF on the joystick once and we decided that we didn't like using it. Under acceleration, it's fine, but under stopping, it's not so good. To make the story short, it will cause you to overshoot (because it ramps down your input just like it ramps it up). You can try some funky "if increasing do this, if decreasing do that" logic, but I really don't think it's worth it.

Also, the deadbands that have been described here are discontinuous deadbands, which are generally considered "hack" deadbands in the controls world (due to their abrupt change at the deadzone point). For a continuous deadzone (more commonly used), do this:

DeadZone = 15; // Calibration parameter
if (Joy > (127-DeadZone) && Joy < (127+DeadZone)) Joy = 127;
else if (Joy <= (127-DeadZone)) Joy += DeadZone;
else Joy -= DeadZone;

Re: Desensitizing Joysticks
 

Thanks for the warnings.

Re: Desensitizing Joysticks
 

This works great on our bot, everyone from SWVGS Maximus thanks you.
We used a switch statement and it did not slow down anything. It is a big improvement from what we had. :)