I have a Toddler I bought some time ago and I recently bought 8 of the SX-52 Demo Boards.

Now, I am thinking that I really would like to use one on my Parallax Toddler to replace the BS-2 board.

I have always wanted to integrate ultrasound and IR to make the little guy able to wander without disasterous collision.

I have the twinkle toes, the ultrasound, serveral Ir options [one pair are on the toes looking downward for a drop off, others are looking forward for an obstacle.

Has anyone tried such a change over? Has anyone suceeded?

I had previously looked at PWM on the SX-28, but found that the examples were a bit hard to comprehend.  It seemed that they were mostly about mutiple outputs, not a complete concept.

Any suggestions about seting up the PWM output for at least two servos?

Actually, I am looking forward to this as I really need the challenge to get me fully into the SX-52.

Update.

If may be more appropriate to move this whole thread to Robotics.  I cannot really decide where it belongs.

I got started.  I drilled two additional holes in the SX48/52 Proto Board to mount to the Toddler.
 
One is between the capacitors and the voltage regulator whilst the other is across the board has demolished the P in Proto. Handily, nothing is in the way.

So the SX-52 is now in place atop the Toddler.

I located code for PWM in Gunther's book, but it is for rotating servos.  I have been pondering backing off a bit and using the original Toddler board for purely locomotion with the Twinkle Toes for immediate collision and the two downward IR sensors for immediate dropoff.  
Then the SX48/52 Protoboard could really be dedicated to mapping the world about it and some form of navigation.  It could redirect the Toddlers walking routines at anytime or it could just observe surroundings.

Eventually, it might even be give tasks or objectives [like go hide and rest, seek out a recharge, greet intruders..]
The reasoning behind this is that it seems similar to biological brain and locomotion relationships.  We keep moving along, sensing, and thinking at the same time.  Locomotion is more reflexive and not really cognitive, so having the Toddler do its thing independent of higher functions will allow 'cognitive' portion to avoid interuptions from the locomotion task.

And yet the cognitive could interrupt the locomotion at anytime.

While it some ways, it seems an 'easy way out', I really feel that having one centralized processor handle everything is a computerized ideal.  There seems to be a natural division of tasks that robotics should recognize.  BEAM robotics seems to contribute a lot of support to this as they have pretty well demonstrated walking to be a relatively reflexive task that can been done blindly.