Let's Make Robots!

key requirements for Servos for walker robots? What can I sacrifice?

Hello all.


I am in the midst of building a six leg walker bot. I am following a set of designs I found posted on here. Initially, it will be a 2 DOF setup, but the plan is to use it as a test bed, and expand out to 3DOF, add sensors, add increasing complex algorithms etc.


I would like to keep the costs as low as I can initially, but the flip side is that I don't want to cost cut to the point of introducing significant issues into the build, which then takes a lot of additional coding to manage.


The thing I am reseaching at the moment is servo selection for the legs. I was wondering if I could ask your advice?


In order to keep costs down, it would be good to purchase lower cost servos. However, these come with some issues:

1) Possibly higher current draw - some super cheap servos look like they are pulling 1amp in normal operation

2) Slow movement - 0.23ms makes for a cheap servo, but a slow one. I'm not sure whether this will fit in the with concept of a walker - maybe it'll end up more of a geriatric?

3) reduced load capacity - for example cheaper servos seem to sit around the 3kg/cm range, where as higher cost sit aroun the 6kg/cm range.

4) wider dead ranges


Like anything to do with an engineering project, it's all going to be about comprimise. I was wondering which of the above I can get away with, and which will lead to major issues with my build?

Are there commonly used servo models used in walkers?


Some model specifics:

Model size - approx 600mm long x 450mm wide

Weight - as yet unknown until built - which is a bit of an issue in terms of this question.

          In terms of construction, I will be using 4mm carbon/ply/carbon composite for all structural components in order to give good stiffness at a low weight. Lets me have the structural strength better than Aluminium at less than 1/2 the weight. Batteries will initially be 4.3mA/H Ni-Mh's weighting about 400g, but if I need to I can shift to LiPo to bring down weight, but that means canabalising my planes.


The servos will all be based in the body itself, not on the legs - so maybe that means I can play with mechanical advantage, and get away with a reduced Kg/cm rating?


My thinking at the moment for the walker is that, in terms of priority, it gos something like:

1) As small a dead zone as possible

2) current draw

3) speed

4) torque.


Am I on the right lines?




Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

There's two things that I came up with on my quest for a hexapod (which is frozen for a while):

1: Use the same servo's for the same joints. So for the hipjoints, use 6 identical servo's.
2: Don't cheap out! You don't need the top of the bill ultra-fast digital extreme torque servo's, but don't work with the ultra cheapo's. From RC racing it was already apparent to me that you don't need a 90 euro servo for steering, but a 20 euro servo will not do the job it is supposed to do. Or it will, for a short while and then burn out (I had a Towerpro MG-945 set my car on fire in the middle of a race, not nice with a full tank of nitro-methanol, I can tell you that!)

I haven't built one myself so I have no idea about the deadzones and everything, that is to be seen for myself as well.

Remember that most walking gaits will have the robots weight supported by only half of the legs while the other half are moving. As an example. If you have a six legged robot that weighs 1.2Kg then your legs must be able to support at least 0.4Kg. If the leg is 10cm long then the thigh servo needs a minimum 4Kg/cm torque rating. In Practise you should use at least 6Kg/cm servos. A knee servo can have a lower torque rating to help reduce cost. My experience has been that it pays to use servos with metal gears. A cheap servo isn't cheap if you need to replace it every week because the gears stripped.

Thanks OddBot


Can you comment on the deadzone of the servos? Is it an issue for walkers?


What I am not sure about is general accuracy. Things like deadzones, servo calibration between different servos etc kep making me think that there is going to be a compunding error building up as the walker walkers along. I.e - on short walks it's fine, but there is "control drift" over time, as it walks further.

Am I over thinking this?