Circuits

[steamboat]

Hello Group,

I am playing with the pololu motor driver and want to build a closed loop system for the leg motors. Therefore I want to integrate a rotation sensor for each motor. This will need 7 pins for the pololu and at least 2 pins with interrupt for the motion sensors. Not much pins left for other fun things to play with. Additional I am running out of memory!

What do you think about using an autonomous motor closed loop unit/PCB with an individual 32u4 which supports the motor driver, the motion sensors and maybe a collision sensing system build from some bumper switches or US-sensors.

This "spinal cord" motor unit could communicate through a two wire serial communication with the brain. The commands could be like go ahead till you hit and obstacle - go ahead two steps - turn 90° to the left (for this we also could integrate a compass into this system) - walk ahead with synchronized legs - walk ahead with 180° offset for left and right legs and so on.

As usual for spinal cord systems this would relieve the master brain a lot from the control work, free some memory and save connetors in the same way.

[jmccartney]

(09-11-2014 06:45 PM)steamboat Wrote:  Hello Group,

I am playing with the pololu motor driver and want to build a closed loop system for the leg motors. Therefore I want to integrate a rotation sensor for each motor. This will need 7 pins for the pololu and at least 2 pins with interrupt for the motion sensors. Not much pins left for other fun things to play with. Additional I am running out of memory!

What do you think about using an autonomous motor closed loop unit/PCB with an individual 32u4 which supports the motor driver, the motion sensors and maybe a collision sensing system build from some bumper switches or US-sensors.

This "spinal cord" motor unit could communicate through a two wire serial communication with the brain. The commands could be like go ahead till you hit and obstacle - go ahead two steps - turn 90° to the left (for this we also could integrate a compass into this system) - walk ahead with synchronized legs - walk ahead with 180° offset for left and right legs and so on.

As usual for spinal cord systems this would relieve the master brain a lot from the control work, free some memory and save connectors in the same way.

There is a lot to think about here. I ran into some of this when I was designing the first iteration of the minimal shellmo board.

Splitting this functionality off into a spinal cord module is an interesting/inspiring idea. I think we would need to think of an alternative to the 32u4 because not only is cost a major factor, we wouldn't need the USB transceiver. Maybe an ATtiny. Another thing you did not mention but is also an issue is the number of PWM pins we are using. Splitting this off to another module will free up PWM pins as well.

What about cost? Another PCB doesn't really add much, but another MCU would add some..

[steamboat]

(09-16-2014 12:42 AM)jmccartney Wrote:  ...Maybe an ATtiny...

We should be able to write "Arduino inside" on the PCB. Only with this label we will get some friends/customers!

I will set up my Wood Walker with the 32U4 and have a look how many I/O's are needed/missing. Maybe a simple board in the near future would be better than none ever...

[jmccartney]

(09-16-2014 03:21 AM)steamboat Wrote:  

(09-16-2014 12:42 AM)jmccartney Wrote:  ...Maybe an ATtiny...

We should be able to write "Arduino inside" on the PCB. Only with this label we will get some friends/customers!

You misunderstand me. I am saying your idea is good, but the 32u4 is too expensive/overkill for this application. ATtiny is cheap, you can use arduino on it, and the part that we care to advertise having "arduino inside" would be the 32u4 "brain" anyways...

I am working on a simple power board with some safety features while you guys continue to hatch out what we "really want"