PBP projects for R/C models

[Kenjones1935]

Yes, the PBP is not freeware. But from the point of view of a middle school club or group project it is a 'one of'.

I did not anticipate the complexity of upping the speed of this car and only asking it to follow the wall. This has been a good lesson for me.

Ken

[mackrackit]

Yes, the PBP is not freeware. But from the point of view of a middle school club or group project it is a 'one of'.

You may want to read the back cover of your manual.

Something like this came up before, no group license.
That is why I get kids started with some basic ASM and/or what they can do with the demo version.

[Kenjones1935]

I just read the above mentioned Software License.

I'd rather use LOGO. But that does not seem forthcoming. Hmmm...

Ken

[mackrackit]

I would contact MeLabs, maybe they will work a deal with you.

[Kenjones1935]

Of course you are correct. Classic robotics is error signal driven.

My problem with PID is my lack of confidence. I do not know the capabilities and limitations of PBP and the 16F887. I have not figured out how to debug the stuff in real time on the bench etc. I have been lazy regarding the 300+ page PIC Owners Manual. I never mastered MPLAB.

Can you point me to some simple PID code in a familiar language - C or C++ (complete with libraries) for example.

Ken

[Kenjones1935]

My wall following code is state machine code. I am going forward. The right facing sonar tells me a distance. I compare that to constants and to the previous measurement. Given where I am relative to a line parallel to the wall and whether I am getting closer to or farther from the wall, I steer right, left or straight.

Here's the code:

keepgoingforward: '----Compare to right wall
IF oldrangeright = rangeright THEN
HPWM 1,Straight,50
ENDIF

IF rangeright > outertrack THEN
HPWM 1,HalfRight,50
ENDIF

IF rangeright < desiredtrack AND oldrangeright < rangeright THEN
HPWM 1,Straight,50
ENDIF 
  
IF rangeright > desiredtrack AND oldrangeright > rangeright THEN
HPWM 1,Straight,50
ENDIF

IF rangeright > desiredtrack AND oldrangeright < rangeright THEN
HPWM 1,HalfRight,50
ENDIF

IF rangeright < desiredtrack AND oldrangeright > rangeright THEN
HPWM 1,HalfLeft,50
ENDIF

GOTO main

Given my limitations this seemed the easiest path to take.

Ken

[Ioannis]

No need to go far away from your PBP lanquage.

Have a look at Henrik's routines here http://www.picbasic.co.uk/forum/showthread.php?p=54846

It is not easy as A-B-C but also not too hard to follow. May the difficult part would be to tune the PID parameters.

Ioannis

[Kenjones1935]

I sort of understand using PID to set a POT value via a DC motor. The feedback is stable and static.

My robocar needs to stay in a path while moving. The feed back - change in the echo response from the wall - is dynamic and is a function of the car's speed and direction. The PIC is plenty fast. The sonar response is not "dynamic" in this sense. My response delay problem is in the physical reality of the steering mechanism and the physics of car motion.

Another issue may be that the HPWM signal is not "really" to radio control spec. It is not a 50Hz pulse of width varying between 1msec and 2msec with 256 gradations. The frequency is more than 50 per second. The arguments I use for HPWM I arrived at empirically. I do not have an appropriate oscilloscope to accurately measure what I am producing.

I need a better understanding of what you are suggesting.

Ken

[Ioannis]

Hmm, have a look at the videos at YouTube about line followers with and without PID control of the robot.

Behaviour of non PID is similar to your nervous car. The same robot with PID control slides smooth on the line (or in your case follow the wall in a steady distance).

Why not give it a try?

Here is the same robot with and without.

http://members.jcom.home.ne.jp/felm/ltc_p.mpg
http://members.jcom.home.ne.jp/felm/ltc_pd.mpg

Ioannis