For the feet you might try a round slightly dished foot to reduce the possibility of getting snagged on objects by the flat sharp corners.
Gads! What big feet you have grandma!
They look a little large to me. Mind if I suggest getting something like the rubber-coated disks used for appliances? They are about 1"- 1.5" in diameter, and generally have a threaded stem.
I see what you mean by 2 1/2 DOF! Quite innovative!
At lease you’re walking, I’m not walking yet!
Oh, I’d say you’re [size=150]way[/size] past crazy, like the 'rest of us!
Alan KM6VV
I think it really could work well as a 6 servo, 1 1/2 DOF system. All the feet servos are doing is moving in the same axis as the shoulders, at the same speed and direction, so they could be replaced with a mechanical linkage.
It could work much better then it looks because my robot has a terrible chassis-- who would have guessed that a plastic food container would be a bad chassis? It is also terribly balanced and uses no accelerometer. If it either was evenly balanced or I used an accelerometer to adjust the tilt on the fly at the shoulders it would be much more graceful. A stiff chassis would also help.
My theory was that using fewer servos can be cheaper (it wasn’t!) and by having two servos lifting you could focus on having two strong servos doing the hard work, and four smaller ones pushing you forward and steering. I think there is some validity to the approach.
You have any tools? such as a milling machine maybe? you could create your own chassis out of aluminum.
I don’t even have a table saw 
One of the reasons why I have been using things like plastic bins and broom handles is that they are easy to cut and drill with hand tools.
What I like about the servo erector set is that it allows you to do a lot of different things without having to manufacture your own parts. I think I could improve the chassis and come up with a mechanism to swivel the feet with off the shelf parts here. I think I have really learned the value of that.
I am more comfortable doing my own thing with software and electronics, it is closer to what I know. I like to experiment with new robot designs but I don’t expect I will ever mill my own parts.
One can make reasonably strong and light structures from the 5.2mm lauan sheet plywood, which is easily cut and drilled (and is only $2.88 for a 2’x2’ sheet at Lowes).
I know the feet look large, but they help keep the robot from tipping to one side. I agree that it would help if they had curved edges. But I tried small round feet and they were much worse for climbing and soft surfaces like grass. It may be better if the ankle servos were a little more powerful, because the feet are like levers and I am not sure they can really support the weight that they can sometimes bear.
I got to do a number of test runs today. My biggest problem seems to be that my robot is very wide for the course. It is around 70 cm wide and the course is 100 cm (1M) wide. On a flat surface the robot goes pretty straight, certainly straight enough for two meters. But what if I hit a diagonal step, or a step on one side. It’s pretty easy to get turned 30 degrees.
I have a digital compass for course correction. I have experimented with that a lot (you can see some at trailofdestruction.com) so I was able to get it working quickly. Here is the problem: gates that turn the robot often are not as good at getting all feet over an obstacle.
So there is a possible fail state where the robot gets one foot over a step, but not the other, and gets turned off course. In trying to turn back on course it fails to get over the object in such a way that it is stuck in a loop. It’s a really tricky problem because the robot has no sense of when its feet are rubbing against an obstacle. I think I need to make sure that all my turning is done in such a way that I am always going forward (no spinning in place) and always lifting the legs enough to get over a high obstacle.
I would love to add:
- Pressure sensors or switches to the feet. It could take months to adapt the gait to the information they gather.
- A GPS to do the robo-Magellan race. I don’t think a walker has ever done that race and I think I am close to having a robot that could. I have payload to spare and I have no trouble with outdoor surfaces.
- Lights, switches and doo-dads that are easy to do but make your robot look cooler. That’s probably all I will get to.