Author Topic: Walker Robot Legs (Read 11988 times)

cjwillms · « **on:** April 07, 2006, 10:52:02 PM »

I'm try to design my own bot, based on this picutre, and this is what I have so far. Its a really bad drawing, but it should work.
http://img128.imageshack.us/my.php?image=botleg2eg.jpg

And here's the design for the servo controller. It's a bit of a mess, but it should be functional.
http://img83.imageshack.us/my.php?image=servocontroller3ho.jpg

Admin · « **Reply #1 on:** April 08, 2006, 07:27:46 AM »

Just a few comments . . .

Dont use transistors, use mosfets
http://www.societyofrobots.com/electronics_advanced_components_tutorial.shtml#mosfet

Consider putting heat sinks on them . . . test to see if you need em.

As for balance, what happens when your robot lifts a leg?

And for the spring . . . they are more difficult to use then you think. I have tried using springs in a similar manner, and they just dont work right if you dont plan it really well. Look into air springs too . . .

cjwillms · « **Reply #2 on:** April 08, 2006, 11:14:15 AM »

I probably will use a mosfets. Based on your article, they do sound better, but they seems to waist a bit of power (excess heat).

Each leg will have a foot on the bottom. It'll kind of look like a upside down T . As long as I make the frame sturdy enough (carbon fiber), the foot should be able to stop the bot from tipping over. Here's a picture of a toy robot with pretty much the same idea
http://www.jeffbots.com/climbatron2.jpg

cjwillms · « **Reply #3 on:** April 09, 2006, 10:17:30 PM »

Ok, I used 3DSMax to model the legs. It’s a lot more functional then what I had previously drawn up, and it should theoretically work. The only thing I’m not really sure about is the feet. It should be able to walk on flat surfaces, but slopes will be a problem. I could add another servo to control the feet, and then use 2 IR ports to control it’s position.

I’ll probably work on an animation demonstrating how it’ll actually walk.

A quick note, the light brown boxes represent the servos :p

cjwillms · « **Reply #4 on:** April 11, 2006, 08:20:25 PM »

For the past few days, I've done some calculations on my design. Originally, I had wanted to use carbon fiber tubes to build the chassis, but I ran into a few problems. So, I think I'll use HDPE instead.

I'm currently planning to use 301.68in^2 at 3/18in thickness of HDPE, and 6 Hitec HS-311 servos. The servos will weight 9.12 oz together, but I still really don't have a clue about the HDPE's weight. As of right now, I'm just guessing that it'll weigh about 30oz for the servos & chassis, plus another 3.2oz for the battery.

If possible, could you provide me the weight for some HDPE weight at any dimensions? It'll really help in my calculations.

Admin · « **Reply #5 on:** April 11, 2006, 08:23:57 PM »

It is about .96 grams per cubic centimeter.

My cad software does the weight calculations for me =)

Dont forget to consider bending of your parts. Carbon Fiber doesnt bend, while HDPE and aluminum might.

cjwillms · « **Reply #6 on:** April 11, 2006, 08:26:59 PM »

Bending really shouldn't be a issue, since the legs will be made out of boxes.

Anyway, thanks for the weight value.

cjwillms · « **Reply #7 on:** April 11, 2006, 08:45:49 PM »

That stuff really is light. From my calculations, all the HDPE I need, should weight about 4.3246848 oz. That's only 16.64 oz all together. Now I don't have to worry about the servos :p (~54 oz/in torque).

Do you think I should use 1/4in (or more) thickness, instead of the 3/18in? I really don't think it'll make much of a differnce, but it wouldn't hurt.

Admin · « **Reply #8 on:** April 11, 2006, 08:51:56 PM »

You probably want to use 1/4in for the legs, and go thinner as you work your way up the robot.

A lot depends on your design of course.

Dont forget to include battery weights!

cjwillms · « **Reply #9 on:** April 11, 2006, 08:59:17 PM »

As of right now, I'll probably just use a 3.5oz 7.2v Li-on battery pack (2400mAh). I still need to run the calcuations for power usage, but if it last for more then 30 minuets, I probably won't use more then 1 battery for now.

the design is basically just a bunch of 2x2x6 (LxWxH)in squares, with the servos pretty much directly connected to the joints.

2Hektik · « **Reply #10 on:** April 12, 2006, 05:47:48 AM »

Arggghhh... the imperical measurements burn my eyes!

cjwillms · « **Reply #11 on:** April 12, 2006, 07:55:16 AM »

imperical?

Anyway, you probably mean the english system. I like the metric system myself, but I'm just use to the english system.

Afroman · « **Reply #12 on:** April 14, 2006, 02:33:01 AM »

Imperial (no C) is the correct term. If you said "english system" outside of the USA you would just get a weird confused look because England hasn't used the imperial system for many many decades.

2Hektik · « **Reply #13 on:** April 22, 2006, 06:48:58 AM »

Yes... that unruly beast

cjwillms · « **Reply #14 on:** May 01, 2006, 07:23:04 PM »

Ok, I've gotten almost everything I need done. I do have a quick question about torque though.

Using your static mechanics tutorial, I cosidered the lenght of the arm, the top part of the leg (since the knee joint will need to lift the rest of the bot). The weight of the beer can, is the weight of the other leg, along with the weight of the servos.

I'm a bit confused about the whole beer can part. Unfortunatly, while I was taking my beer out of the fridge, I dropped it. Now, some how, in the shape of a L . Does the forumal for torque change at all, or does it stay the same?

In other words, is it still
Length of upper leg * weight of other leg + Length of upper leg * 1/2 * upper leg weight

if so, then I only need a 41 oz/in servo for the knee to lift the rest of the bot.

Admin · « **Reply #15 on:** May 02, 2006, 10:28:15 AM »

Hmmm I dont quite understand your question. I assume you are trying to calculate the torque the knee joint servo would need.

The maximum force the knees would ever see is as you said, when the robot head to waist to knees position was L-shaped.

So the equation is close to what you had, torque required is:

length of upper leg * (weight of everything minus weight of lower leg on the ground) + length of upper leg * 1/2 * upper leg weight

Just know that this is worst case scenario, no one actually sticks their leg out 90 degrees when they walk.

And obviously required torque would be less than half if both legs are on the ground as when standing from a sitting position.

cjwillms · « **Reply #16 on:** May 15, 2006, 09:12:52 PM »

OK, I just though I'd update you as to how my project has been going.

I've gotten in all the supplies I need, and I've pretty much finished the legs. Each leg can support the rest of the bot, and I've pretty much eliminated all bending.

ServoCity did freaked me out, though. I don't know why, but they include a small bag of mixed candy... After that, I had a problem with my batteries (2x 2cell 3v battery's). I left them on the charger for the stated time, but both of them didn't hold a charge. I though it was kind a weird that 2 brand new batteries didn't work, so I opened up the charger (didn't want to wait until tomorrow, plus it would take a 1.5 hour drive to take it back) and fix the problem with some solder. It appears to be working now, but I still have to wait until tomorrow to find out.

Other then that, I now just have to refine the timing of my microcontroller and figure out the algorithms to make it walk.

CRAPPY CAMERA ALEART :/
The top picture is the bot standing on both legs
the bottom picture is it standing on one leg

Admin · « **Reply #17 on:** May 15, 2006, 09:55:14 PM »

Whoa, really neat!

You gotta post a video when you get it doing stuff. I am working on a place on this site for people to post stuff, not sure when it will be done though . . . Basically a place for people to share mechanical designs and schematics and code. Interested?

I am also interested in learning about any biped specific pitfalls and mistakes you might have made and how you fixed them. That way when I try my first biped, I could save myself some pain . . .

I can only assume the 'tweaking' phase of programming a biped to be long and painful. If you learn tricks to program walking and turning, feel free to share

cjwillms · « **Reply #18 on:** May 15, 2006, 10:08:36 PM »

I'll post the schematics and everything when I'm done.

Anyway the biggest mistake I made had to do with the joints. Originally, I had wanted to make the servos the axis of each joint, but that put way to much strain on the servos (a lot of flexing). Now, I'm using screws as the axis.

Other then that, the foot is a big problem. Theoretically, the foot has be at least half the width of the bot, otherwise the center of gravity will cause it to tip. To fix this, I made the feet kinda like a L, and then made it at a angle, to increase the force that holds the rest of the bot up.

Admin · « **Reply #19 on:** May 16, 2006, 05:00:38 AM »

What if you used a more expensive servo with a metal output shaft? Do you think it might still flex? I remember on an old robot I built, the plastic servo output shafts broke because I kept applying too much bending force . . . $:-\$

cjwillms · « **Reply #20 on:** May 16, 2006, 05:36:42 AM »

I've never worked with the more expensive servos, so I really couldn't tell you. If I had to guess, I would say no. Just look at this picture
http://www.communistrobot.com/articles.php?sorted=nameasc&limit=%&page=&articleid=73

From what I can tell, they use metal servos, but they still have a extra axis to support the robot.

cjwillms · « **Reply #21 on:** May 21, 2006, 04:10:38 PM »

Sadly, I ran into a wall yesterday. I've pretty much finished the programming, and I researched how to change the speed at which my microcontroller runs. I found out, and tired it for my self. Unfortunately, when I changed the value, it set my chip so I can't use it any more. Now I have to spend ~100USD to buy a programmer to set it where I can use it again, as well as set it at the speed I want :/. In other words, I have to wait about 2 weeks before I can start working on it again

.

Anyway, I attached the code. It has a lot of documentation, but I've found it hard to read assembly if you didn't write it your self. Especially since it has 2260 lines of code (not including spaces and documentation).

Admin · « **Reply #22 on:** May 22, 2006, 08:25:39 AM »

Haha yea Ive done that before too, really annoying . . .

You could just buy a new microcontroller. Should only cost a few bucks plus shipping - much cheaper than a $100 programmer.

cjwillms · « **Reply #23 on:** May 22, 2006, 10:09:59 AM »

I would, but the microcontrolller must be running at 4 MHz on up. Either way, I'm going to have to buy a programmer, so I can change the fuses.

News:

Author Topic: Walker Robot Legs (Read 11988 times)

Related Topics