Walk N' Roll

My next robot project is officially known as the Walk 'N Roll Project. :smiley: I was calling it the Octapod Project, or something like that, but now the robot has an official name.

If you are wondering why I have named it Walk 'N Roll, and you have not read my previous robot threads here, it is because this robot will have eight legs and four wheels on supports that can be individually steered and fully retracted up under the robot’s body. The wheels can also be raised and lowered to various heights for different clearance situations.

I have a complete section devoted to Walk 'N Roll on my website, which you can find a direct link to in my signature. Until then, here are a few pictures that are real close to what Walk 'N Roll will look like when it is finished.

Walking (Front):
http://www.thedynaplex.org/robotics/walkroll/img/walknroll-walking-front.jpg

Walking (Side):
http://www.thedynaplex.org/robotics/walkroll/img/walknroll-walking-side.jpg

Walking (Top)
http://www.thedynaplex.org/robotics/walkroll/img/walknroll-walking-top.jpg

Walking (Underside):
http://www.thedynaplex.org/robotics/walkroll/img/walknroll-walking-underside.jpg

Here you can see both decks. The lighter deck is the top deck and the darker one is the bottom deck whee the wheel towers and legs all attach. The wheels retract up into part of the lower deck, so I had to make cutouts in that deck to accomodate them. I wanted the wheels to be as far out of the way as possible for walking. With my CAD software, Alibre Design Expert, I can actually edit subassemblies right within a higher level assembly. In this case, I edited the bottom deck, which is part of the body subassembly, while I had the main Walk 'N Roll assembly open. This made it very easy to create the cutouts in the bottom deck to accomodate the retracted wheels. Using configurations for assemblies, I can also show the robot in different poses now. :smiley: I have four different configurations for the legs and two configurations for the wheels at present.

Rolling (Front):
http://www.thedynaplex.org/robotics/walkroll/img/walknroll-rolling-front.jpg

For leg retraction, I can just as easily have the legs form a sort of tee pee when they are retracted. I may create another retraction configuration to show this.

Rolling (Side):
http://www.thedynaplex.org/robotics/walkroll/img/walknroll-rolling-side.jpg

Rolling (Top):
http://www.thedynaplex.org/robotics/walkroll/img/walknroll-rolling-top.jpg

Rolling (Underside):
http://www.thedynaplex.org/robotics/walkroll/img/walknroll-rolling-underside.jpg

There arer really just two differences between what I show here and the way Walk 'N Roll will actually be when it is complete. Its body will be about three inches wider and an inch or so longer, and it will have 6 inch legs instead of the 4 1/2 inch legs this version has. This body is not wide enough to accomodate retracting 6 inch legs. Walk 'N Roll will definitely have a custom body.

8-Dale

Thats awsome, its amazing how flexible the SES brackets are, you can do almost anything with them! :open_mouth:

Hi,
Just wondering: Why 8 legs instead of six?
8 is certainly cool, but it adds a lot of weight, complexity, and $$cost.
How about an arrangement where each side is leg-wheel-leg-wheel-leg.

Pete

It won’t always be walking as an Octapod. It will have grippers on the front pair of legs and will walk as a hexapod when using them for gripping.

I don’t see where this would be a benefit other than to decrease the cost and weight of the robot. Walk 'N Roll will also serve as a prototype and experimental platform.

It will only use wheels or legs for motion. The only time it will use legs with wheels is if it is using its front legs as arms and is using its grippers.

This robot will also hopefully be a crawler, so will need the added pair of legs for stability, etc. I want to experiment with hybrid robots.

Another reason is an octapod is different from what most people are doing with walkers now. I want to explore some different territory. Walk 'N Roll will also be able to walk as a hexapod or even a quadrapod, so it can still fall back.

Lastly, it’s just what I want to build. It’s interesting to me. :smiley: I expect to be working on this project for a long time, so I will no doubt take detours to do other stuff along the way. :slight_smile: I may not even be able to build Walk 'N Roll with just servos for motors and am exploring other means of locomotion. I can also put Walk 'N Roll on a pure SES body, but will be going with a custom body for the final version of the robot.

8-Dale

Wow.
Someone’s got crazy CAD skills and a lot of time.
:laughing:

Just wondering…
Have you thought of making it possible to have the hex still be able to drive when the legs are fully retracted?
I’m asking this because, if you have a recession that supports the wheels properly, you’ll be able to turn off all the servos and just run the drive servos.

Methinks that would be helpful in situations that need power conservation.
Getting the robot to walk around is great for about 30 minutes, but it’s just too much stress on the servos to have them on continuously.
Of course, you could switch between leg walking and the wheels, to give each set of servos rest, but that doesn’t give your batts any rest.
:stuck_out_tongue:

I do OK with CAD, and I am learning pretty fast. :slight_smile: Yes, I have lots of time for stuff due being disabled and not working at a job.

Of course the robot will be able to drive when the legs are retracted. This robot is a hybrid. That is the idea of the Walk 'N Roll bot. It can roll or walk as required by whatever terrain it is in. This bot will be alble to walk as a full octapod, a hexapod, and maybe even a quadrapod.

This is what I am going to try and do. I want to be able to control power to the 6 back pair of legs, the 2 front legs, and the wheels all separately so I can have just the servos I actually need turned on. The two front legs will also carry grippers, so they need to be usable in either locomotion mode.

Since the units not being used will be fully retracted, I am hoping this will be pretty easy to accomplish.

Yes, indeed, and it will be one of my power management strategies for Walk 'N Roll - power just what it absolutely needs at any given time. I just have to see how the servos act when they are powered off.

8-Dale

Add another srevo as a locking mechinism. Have it push a pin out to lock a position of the wheels either up or down, then it won’t matter if the servos are off. Just an idea.

Walk 'N Roll already needs a total of 40 servos just to operate legs, wheels, and grippers. I really don’t want to add any more. :smiley: If I go with motors for actual wheel movement, then that would take away 4 servos, which would be good. However, I stil am not thrilled with adding more servos.

Of course, only one locomotion system will be in use at a given time, with or without the front pair of legs for gripper usage. The servos still add weight though, so I am hesitant to add any more.

I will investigate to see if there is a way I can trigger a locking mechanism when the wheels are retracted past a certain point. This scheme might work and won’t require adding another servo.

You do have a good idea here. :slight_smile:

8-Dale

Good idea, but you don’t need a servo - just a small solenoid.

Pete

In that case, MOSFETs are your friend.
If you go with logic MOSFETs, you can fully turn them on or off with a 5V signal from your micro.
I’d recommend turning on and off the ground going to each set of servos.
If you try turning off the Vdd from the battery, you’ll have to send a signal that’s 5V greater than that to turn it fully off (5V + 7.2V = 12.2V!).
I’d also recommend putting a 270 ohm resistor in series with the MOSFET gate that you’ll be driving.
That’ll limit you to about 20mA of current, which will prevent the micro from surging too much current all at once to turn the bugger on.
After it’s fully on, it takes little or no current to keep it on, so the 20mA limit isn’t bothersome.

Eeeeeek, no way I would want to deal with that! So turning off the ground it will be, if I do this.

So, I would need one logic MOSFET and a 270 ohm resistor, per servo connection, right? If I understand this correct, I would insert this between the microcontroller’s pin going to the servo, and the ground line of the cable going to the servo. Is that correct?

If I am getting this right, I might be better off to have a small board with the MOSFETs and resistors and two connections per servo - one going to the servo and one going to the microcontroller pin.

Then it would take one I/O pin per group of servos to turn them on and off.

How does that sound?

8-Dale

Arg! I just spent 15 minutes explaining why it may be a problem to do what Nick suggested, then the forum site dropped my connection…

The short version is:
Switching out the ground-side of a 3-wire device such as a servo is a potential problem, because the Vcc and Signal wires are still being driven.
I don’t recommend it. There is a “chance” of damaging your servos.

Instead, switch the PWM signals going to each servo, using (for example) a 74LS02 AND gate (4 gates per package).

Pete

I used to hate when that happened to me, when I had dial-up internet.

I was thinking about the whole idea of turning servos on and off to reduce power requirements and drain on battery. If analog servos are powered down, they can do erratic things when powered back up.

If servos remain powered up and they don’t get a regular PWM signal, they won’t hold their position.

With this in mind, I don’t see any good way to turn servos off when they are not needed, especially with retracted legs. I don’t know what the legs would do when their servos are powered off or they are powered on and don’t get a PWM signal regularly.

Powering down a servo or stopping the PWM signal to it might be a bad thing for retracted legs. It might be workable to keep the wheels in retracted position by having some sort of locking mechanism that kicks in when the wheel retract, but this is not possible for the legs as far as I know.

8-Dale

That sounds no worse than powering them off and then having them possibly jump around when powered back up…

I’ve had pretty good luck with powering off the servos and the SSC-32 that drives them. There is sometimes a little jerk when I turn them back on, but it’s not bad.

Pete

That is true. I could just make sure the servos were in a good position before cutting the PWM signal off. Would this still provide enough power savings to be worth doing though?

This sounds good, and I appreciate the feedback. Maybe I will pursue this power saving method then and see what happens. I can experiment some with this on Walter.

8-Dale

Maybe not. If you have a ‘resting position’ that does not put a significant load on the servo, then leave it on. Averaged over time, it won’t use much current compared to things like using the legs for walking.

The real point is how long does your battery last – I think that is mostly based on how much walking you do (or other high-load tasks).

Pete

Very true. I am currently planning to use one battery to power each side of an SSC-32 for servos, and will have two SSC-32s on Walk 'N Roll to divide up the servo power load.

I am thinking of using 1600 mAH 7.2V packs for everything. This would mean at least 3 battery packs just for servo power and one for electronics.

8-Dale

Hi,

I had the same idea but it was for a humanoid robot. Like the BRAT with arms and wheels (bigger would be too heavy for the servos so like that).

It’s a nice project and you’re good at CAD :laughing:

nice job and I cant wait for pics of the real thing

Sam

nice job and I cant wait for pics of the real thing

I hope to start working on Walk 'N Roll in February, but it may have to wait until March or April. There are still a few things I want to do with W.A.L.T.E.R. :slight_smile:

Thanks. :smiley:

8-Dale

Nice drawings. I love the design.

I had actually thought of doing the exact same thing before starting my Giant Biped. I was just concerned about the payload of the motors, batteries and drive train associated with having wheels and legs. I was only thinking of a hexapod though and no retracts. Motors weight alot and so do the batteries to power them. It would add alot of weight to the payload. With a light design, and some monster servos, it could probably be done. I can’t wait to see you get started. It’s cheaper for me to watch someone else build it than to do it myself. :slight_smile: