Lynxmotion Tech Support

Marcham,

I really like your latest post with the tank.png picture. This thing is really looking good.

If I may, I would like to offer a few thoughts, not as criticism because I don't know enough to be critical:
- I notice that this configuration places a great deal of tread on the ground where it could create a lot of friction to overcome and possibly some tread wear.
- If your "ankle" servos offer enough rotation, you may be able relieve most of the friction and stress most of the time by rotating the "toe" ends of the treads further upward so that ground contact is concentrated on the road wheels currently pointed to by the ends of the femurs. This would still allow you to use full tread contact in dusty, soft, or slippery surfaces.

I certainly am not suggesting any sort of redesign because I love what you've done. But since you are working with a tracked vehicle, I thought you might be interested in an exchange I had in March 2009 on the Lets Make Robots Forum with a former tank driver signed Cwignell. On that forum I was TechnoBuff until my brother envisioned me doing technology in the buff.

My very best wishes to you, RoboTed

//////////////////////////////////////////////////////////////////////////////////////////////////////////////

Everything that follows is from the LMR Forum. Snailkeeper correct me if I'm doing this wrong.

Track design for tracked robots
By TechnoBuff

March 11, 2009
I am brand new to the robot hobby with questions about track profiles.

The high sided triangular track profile of Wall-E or Johnny 5 looks really good, but seems to make inefficient use of track links. Other than good looks, is there a good design reason for this profile?

Of the other general profiles available (parallelogram - WWI tanks), (rounded rectangle - Christie type WWII tanks), (inverted trapezoid - short base down), what are the relative merits in terms of efficient use of links vs. object climbing ability?

Of these, which tend to cause the most stress on the links, and which cause the least?

Thank you for your assistance.



By cwignell
@ Wed, 2009-03-11 10:04
Hi TechnoBuf,

You will see the Wall-E arrangement on BullDozers, it has the advantage the it puts the final drives closer to the middle of the vehicle and having the drive sprocket higher means the track tend to self clear rocks and crap which you can pick up when turning; this reduces thrown tracks.

The rombod tanks from WW1 were designed around the minimum radius wheel which would cross the trenches then in use.

The convential approach with the drive at one end at a similar level to the idlers puts the powerplant and transmission at one end of the vehicle (usually the rear) and leaves a big crew space (well it's not, it's very snug because there is a lot of stuff in there with you.). This also serves to lower the profile, handy went people are shooting at you. Picking up rocks is less of an issue with the modern tank but if you look closely at some of them they have plow like arrangements to clear the tracks before the sprockets.

As for climbing, the convential approach is the best normally you can get up verticle steps a little bigger than the height from the ground to the front idler/sprocket. The max gradent depends on a lot of things, Hp, transmission, traction, surface and so on.

The Wall-E arrangement will have a poor step performance, and step performance is not a big issue for bulldozers.

Oh if your wondering, I drove tanks for a living once.

By TechnoBuff
@ Fri, 2009-03-13 23:43

Thanks, Cwignell,

Excellent answers concisely stated. Wall-E design now makes sense where CG considerations require driven sproket near center and climbing is minor design factor. Your comment on climbing is very encouraging. I had expected top sprocket would have to loom over step height rather than possibly being slightly below. A clarification question: When you described "conventional approach", I envision a Christie-like design but with the front and rear end wheels raised somewhat above the level of the ground contact base wheels and idlers guiding the tread on the return direction. This would give a little more climbing ability than an absolutely flat arrangement like some cranes and road paving equipment. Did I understand you correctly?

Re you last sentence, since you answered my inquiry, I assume "living" was the operative word. Good!

TechnoBuff

By cwignell
@ Sat, 2009-03-14 02:20
Hello TechnoBulff

Yes when if refer to to convertional approach I mean where the leading wheel is raised and the the rear wheel is raised and all of the Roadwheels sit on the ground, note, the leading and read wheel can be either the driven (sprocket) or an Idler. The trend with moden tanks is the rears are the drives as it allows all of the engine and transmission to be incapsulated at the rear, but there are some front drive tanks around. Speed and mobility on the modern battle field are important, climbing ramparts (steps) and trenches is not so important any more.

To be strictly accurate Christie like design refers to a convential approach as above, but uses large diameter roadwheels on a swing arm suspension, the Russion T-34 and the English Crusaider were early examples of this. The robot I am building at the moment is basically a Christie design. Christies were optimised for speed and some of them could be driven with out tracks. Track wear is still an expensive issue with tracked machines.

Regarding getting up steps, the actual height you can climb depends on the geometry of the step, the type of surface before, available traction, torque available and if the driver is willing to break the vechicle! One upper limit to step type obsticles is the Height of the Center of gravity vs wheelbase vs step height. Tracked vehicles can assum alarming attitudes.

Hey, Roboted. Thanks for the thoughts and info.

You're right. The amount of tread on the ground and their distances from the center of the bot would make skid steering very difficult. Honestly, with the motors that I've chosen to do this with it probably wouldn't even be possible. The way I've planned to steer should eliminate most of the lateral forces on the treads. This picture shows a top view of the robot mid turn.

So, as you can see, the treads actually turn, making considerations very different from the typical tracked setup.

I've thought about rotating the tracks a bit in some cases to reduce the amount of ground contact. However, in the case of skid steering I think it would make matters worse. Regardless of the amount of ground contact the friction force would remain the same. Friction force depends only on weight and the coefficient of friction, surface area isn't a factor. So in this case, it seems that rotating the tracks towards the driving wheel would localize the friction forces farther from the robot center making the moment from friction larger.

As for wearing, reducing the amount of contact might make that worse, too. Putting greater strains on a smaller section of track could eat them away faster.

I really do appreciate the input. I knew starting this out that there would probably be things that I would overlook. So, I'm happy that you pointed this out, it's good to actually make me think and justify the decisions I made.

It will be interesting to see the video when you get it working!

I like the tracks.

Alan KM6VV

_________________
Visit:
http://groups.yahoo.com/group/SherlineCNC/
http://tech.groups.yahoo.com/group/HexapodRobotIK/

Awesome project when it is done could you be so kind as to let us normal roboters see the technical designs and coding to be able to make one our selfs

Thaks

I'm Back!

I can't believe my last post was almost a year ago. I may have been gone, but I've been busy.

I realized a while ago that part making process I was using wasn't going to give me the robot I wanted. I revised my design to try and use G10 but that wasn't really cutting it. Then I stumbled on this guide on resin casting: http://lcamtuf.coredump.cx/gcnc. From then I figured resin casting was the way to go.

I've spent the last 4 or 5 months slowly stocking my closet machine shop with the materials and tools I needed. I learned some new CAM software and did quite a bit of trial and error machining runs to hone my programs. I finally got to a point where I could hold a tolerance within about 0.01mm. I've completely redesigned my quad to be produced with casting and have recently been making the molds to cast my parts. Today, I finished my first leg prototype. Here's a couple washed out photos of the leg and another of one of the mold positives and its resulting negative.






I think it turned out decent. The parts fit well with only a few modifications and different screws needed. I haven't quite perfected the actual casting process so some of the parts came out a little rough around the edges. The pinkish parts were really supposed to be a darker red color but I apparently bought the wrong pigment. I've got a few others ordered to play around with. Also, these aren't just pure polyurethane, they're filled with glass fiber and glass microspheres to increase rigidity and decrease weight. I treated the fillers with a silane coupler to promote adhesion with the resin and further enhance it's mechanical properties. I still need to play around with the proportions to strike a good balance of weight, rigidity and impact strength.

In the coming weeks I'm planning on making the necessary modifications to my molds and refining my casting process. Hopefully I'll have this all worked out soon, then I'll start machining molds for the body.

Nice. And welcome back.
The polyurethane casts look nice. What did you use, EasyFlo casting resin?
Nice clean mold too. Silicone based?

Look forward to seeing more..

_________________
Jonny Poole,
Freelance Concept Art & Animatronics, 3D Art and 3D Modelling, as well as Physical Modelling, Mechanical design, and illustration.
http://www.innerbreed.co.uk/
Create, Innovate, Inspire.

very cool look forward to see is roll/walk what gear ratio did you use for the motors?

kind regards

Thanks. The resin is IE-3075 from Innovative polymers, more viscous than water but has great mechanical properties. And yup, the mold was made of QM262 from quantum silicones.



Thanks, Codemaster. I really hope to have my bot as complete as yours soon. It looks awesome.

I used the 10:1 HP motors. I was aiming to get a little speed out of them. Not a very high ratio but I should still be able to get around 32 oz of force pushing the bot.

Cool I chose 100:1 its alittie slow but very powerful I was scared of it not moving more than the speed of it
How did and for what did you make the molds from are they heat resistant or not?
I will be working on my code in the coming week or two just shout if you what a test program or two.
I see your leg has a foot instead of just track, how do you plan to get from the stand to ride positions and back with that foot in the way? I found it very difficult but in sure you have got a plan.

Kind regards,
Codemaster

P.S. Awesome as always

I made the molds by machining a mold positive into prototyping board then pouring in silicone to make the negative that I could cast parts in. The silicone I used would be ok up to 200C but I don't think the resin gets higher than 30 or 35C while curing.



I think I could just do a sort of lift and tuck movement with the leg and it'll be alright; not quite there yet. And thanks, I'll give you a shout if I need a program. Are you using a propeller?

Hello,

I am using the propeller with the ps2 controller and the same motor controllers as you.
I'm trying to to make my programs easier to convert from I robot to another and improving the servo offsets.
But I need stronger servos before I can continue making my walking sequences.

Kind Regards,
Codemaster

More progress.

I've nearly finished the legs. I'm still waiting for the the stock to cut the other axles and some photointerputers to use as enoders.

Here's an 'exploded' view of the main parts for the leg.


Another pic of parts for the legs... and my ironing board.


And the pic of the four legs; 3 sans tracks.

I've just got a few more parts to recast then I can move on to the body. I'm gaining excitement again and regret falling away from this project for so long. It's been so long that I have to reverse engineer the board that I designed, but that shouldn't take long. I hope to have another update soon.

looking good
cant wait to see it walk

Awesome work.. really..
That make me want to try moulding

_________________
Eric Nantel
Qc Canada

Always liked Robot... Now i am learning to make some...! Wonderfull...

I've made some progress. I machined a couple molds for the body. Still need to do 2 more. I'd like to pour the silicone but the molds are too big for my vacuum chamber and I don't want to risk wasting the material. It will be worth the investment to get a bigger one.

Quick pic.


I've also been working on my algorithms for foot path production. Here's an excel screenshot of the algorithm I made to generate footpaths for turning. The box in the middle is the quad, each leg at the corner. The large blue dot is the center of rotation. The big axes is the global coordinate system, the small sets of axes are the local coordinate systems for each foot. In this spreadsheet I can change the step length and steering input and it gives me a good visualization of where the bot is allowed to step.


Here's the file if you want to play around, try changing the turn input to numbers between -500 and 500.