2012 Arduino DIY bots & line followers

Anonymous (not verified) Tue, 31/07/2012 - 18:59

In paralell with Pingubot developement race, was the arduino based bot. I liked better this one, since the Pingubot, except for the programming, had almost everything in its place mainly due to the integrated sensor bar and driver for motors.
For Arduino, I had to integrate QTR-8RC Reflectance Sensor Array with Arduino Uno, build myself an L293D driver and build my own arduino software (wich at the end was quite different from the pingubot one, though finally based also in PD control).
So getting it to work, at least as line follower, was a far greater challenge for me. In fact it was a three stage iteration, with lots of changes in the way.
First Arduino iteration. Trying to speed up things, one stage of the geared servo (working as dc motor) was eliminated thus losing torque for speed. As you can see, response frome motors was poor loosing the line and torque dropped at times so much the bot would stall.
The body was built from mecano like parts and regular bolts and screws, weighting a bit too much.


Second Arduino iteration. Body was totally new, with PMMA and a few screws to reduce weight. Distance between sensors and motors was increased and wheels were smaller to increase torque.
It worked fine at the beginning (it stops in the video due to a loos wheel), but after a while becouse of reduced gear stage and high voltage (12+) motors would just give up, so reduced gear config for this servos was discarded. Best way would have been to buy some fast and capable micro motors but budget was tight.




Third Arduino iteration. Body was adapted to new PMMA huge wheels. Two new tuned servos with the original gearing (lower speed, higher torque) were used.
It worked fine first with if routines. Voltage was 12+ volts.






Higher speed still with 'if' routines:






Even higher speed still with 'if' routines:





Finally bot was PD program controlled:





It might seem video is speeded up but it's not. Turns are softer even with speed.
It's not labeled final because bots were intended to compete at a two event contest: line follower and maze solver. For the latter bot wasn't ready ontime so it didn't compete and even if I don't ever get back to it I consider it a 'work in progress' bot. Still, it was a great satisfaction and learnt a lot.