View Full Version : [Project] New mechwarfare build concept

Carl Gustav
07-14-2015, 12:51 PM
Hello everyone, I'm starting to design a new mechwarfare competitor. Check out my idea's and give me advice on how I can make it better and more viable.

I know very very little about robotics but I'm learning more everyday. At this point I want to get my idea's out into the community and see what the responses are to see if I'm going in the right direction. I don't want to invest in a micro controller and servo's until I'm sure that they are the right ones for the job.

I wanted to make a mechwarfare capable version of Stanford's isprawl. I thought it was hydraulically driven, but it turns out its not. Its hard to describe what it does use so here's a youtube url of a very similar robot, and it describes how it works. /watch?v=SMlIoYvWUOU

Now I know I'l need a microcontroller, I know I'l need a custom drive train cnc'ed, I know I'l need an electric motor to run the drivetrain, I know I'l need 6 servo's to position the angle of the legs , And I know I'l need somekind of wireless video and wifi control. ( wireless video and wifi control I know verrrrrry little about, I've been searching the threads for info on the basics but I've had no luck, please help) I'l also need to come up with some type of springy material to use as the hip joints. I know where to get the metal wires for the leg actuation.
I know of a 3d printing service near me so I'l use that for the frame.

I plan on making the exterior torso of the mech look like this, only with 6 legs (the legs will have to look completely different), and it will have to have the metal actuating wires run out of the body, around, and then down into the legs. I could carve the exterior out of balsa wood, or use 3d printed parts if its light weight enough. The metal actuating cables would block the gun from having more than maybe 50 degrees of motion, so the left and right motion of the gun and camera will be limited, but the up and down motion would be unaffected. I would like to use a mouse for control of the gun/camera turret. I might use a custom plastic version 2 (m16/m4) or version 7 (m14) aeg gear box with a dual sector gear for high rate of fire/low fps, and might use feeding system like one out of a aeg ak drum mag (auto winding) so that I can store the bb's in the body and not in an exterior hopper like I see some of your mechs use. Would love to sculpt a shroud for the barrel that looks like a mini gun.
If anyone has any ideas, any ideas at all, on which microcontroller, motor, servo's I could or should use, I'd really appreciate the tips.
Also, how can I get started learning wifi control and video transmission. I don't even know what to call it? Is it called ip/tcp control? What kind of hardware and software do most mechs use for that?
Is there anything else I should learn up on that I've missed touching on to really make this project top notch?

-Carl Gustav

Carl Gustav
07-14-2015, 01:20 PM
I know that iSprawl uses cirrus CS-5.4 servo's.
isprawl weighs 315grams or .69 pounds.
I'm guessing my mech will be atleast 3 times that weight, so I'l need servo's with more torque and holding power.
I plan on sparing no expense so maybe one of the dynamixel offerings will work well. I wonder if there are any robot servos that are smaller, lighter, faster, and torque-y-er than dynamixels.

07-14-2015, 03:42 PM
You may get slightly cheaper servos from Dongbu instead of Robotis. For robotics, you're unlikely to find anything "better" from the RC hobby crowd, and from the motion control / industrial automation crowd, you'll likely to find much heavier and more expensive stuff.

The best "bang for buck" in robotics servos is likely the MX-64. For less money, and less capability, the AX-12A and the DRS-0101 are probably good alternatives.

Btw: the Japanese insect walker you described may violate the Mech Warfare rules in two ways:
1) It has more than four legs
2) It uses a cam-driven locomotion system (there's a cam that drives the actuation wires that drives the pistons that bounce it forward.)

07-14-2015, 05:43 PM
While there is a class that permits 6 legs ('open'? league), I don't think it has ever actually been supported in any of the official competitions due to lack of entrants. The only league guaranteed to have actual battles at any official event would be the standard/regular (no more than 4 legs) league. There might be support for 'hardcore' league battles if the venue has a sufficiently safe arena/enclosure and any entrants actually show up.

Yep, the isprawl design almost certainly violates the 'no cam' drive system by powering more than one leg with a single actuator. Might be able to get away with one driving motor plus cam (converting rotation to reciprocating motion) per leg extension 'joint' and using cheap servos for steering the legs, but I think it is very unlikely that a single driving motor with one cam to control all the leg extension 'joints' using bowden cables would be legal. So instead of one continuous rotation motor and four/six servos to control leg vectoring, you would have to have four legs with each leg consisting of one continuous rotation motor with its own cam (or just use a normal servo) to drive one bowden cable leg extension and one servo to control leg vectoring for a minimum of eight standard servos (or four standard servos and four individually driven continuous rotation motors/servos each with its own cam).

iSprawl's predecessor, aquasprawl, used single action hydraulic cylinders for each leg, but that limited the retraction speed. Better version would have used double action cylinders for each leg and two of the 'drive'/'master cylinders' 180 degrees out of phase (or just swap input and output ports).

Was thinking about constructing a prototype/minimalist 'Jamie' (GITS:SAC Jameson chassis) from my 'spares'/'why did I buy this again?' supply of eight HXT900 servos to control pitch and roll of each leg then using pololu/sanyo micro gearmotors to control the lead screw that turns each leg into a linear actuator. It would operate quite a bit like iSprawl except each leg extension would be individually controlled by a single gear motor, there would be two degrees of freedom in each leg vectoring for higher maneuverability/dexterity, and it would be much slower because of the lead screw mechanism.

Carl Gustav
07-15-2015, 12:17 PM
I forgot that the hexapod class almost never has any entries. I guess I didn't really think that that design would be considered cam driven, but it totally is.
Good thing you guys told me that before I spent a ton of money of cnc'ing.
I guess its back to a hydraulic quadruped then.

^What do you guys think about this robot? Have you ever seen it before?

07-15-2015, 01:04 PM
If each extension/cable is controlled by a single, independent motor/servo, then it would very likely be acceptable. The 'no cam' rule is intended to restrict walkers like the strandbeest and attacknid that use a single, continuously rotating motor to drive multiple legs through a cam/linkage system instead of independent control over each leg. You can have linked control over certain joints of multiple legs (e.g. one servo that controls yaw/roll of two legs), but it is the 'driving multiple legs with one continuous rotation motor' that is verboten. There are designs like 'twitch (http://forums.trossenrobotics.com/showthread.php?5637-Twitch-My-3-Servo-Hexapod)' that use very few servos to control multiple legs/feet, but their statically stable design (three feet per 'leg') prevents them qualifying as true bipeds.

Had not seen that little hydraulic walker, but quite nice. Almost certainly uses custom machined manifold, valves, and cylinders, and probably the pump as well (cheap pumps can be found, but not sure how long they would withstand exposure to non-water liquids).

07-16-2015, 01:12 PM
THAT is a cute walker :-)