Snake robots

[Zypkin]

I was just waiting a piece of hardware worth the money to buy a bunch of servos and start working to a big, fat snake
The dynamixels are simply perfect.

After a week spent doing some tests with the default robot configurations, I now started to develop the snake project.
Here are some shots: http://picasaweb.google.com/zypkin/TheSnakeSFactory

As you can see so far I put together 12 servo and did some test... 6 DOF on X axis and 6 on Z axis... but it is quite easy to drive the all thing with some custom functions.
In about 6 hours I build the snake and implemented some simple gaits like linear progression (forward), sideslide and roll.
I am still messing up with the cable which is a bit tricky to handle for a thing which is moving so much... just think that would be better to switch to wireless... ZigBee??? ...any suggestion

I will upload some video soon...


Cheers,
Zyp

[cdraptor]

The snake design is pretty cool. The CM-5 is pretty big. I was looking at the gumstix products. The robostix mainly - Jon Hylands has done some work with that and could probably provide more info. I was considering the robostix with the wifi - leaving the brains at the computer level and develop a TCP/IP application to communicate between the robot and computer. Takes away from the autonomous nature, but you can do alot more with a full PC brain. The biggest problem I think would be finding a good powersource. The robostix is pretty slender so if you found a powersource you could integrate it into the length of the bot

[Zypkin]

Yep, the CM-5 is the first thing I need to get rid of.
I definitely want to develop the control suite on PC since the gaits are too many to be handled by the robot in a stand alone configuration... meaning that it cant adapt to environment choosing itself the proper gait... but it can be driven from a laptop using any combination of gaits to face any situation.
In other words, if it's easy how to make a snake-like robot move, it's less obvious (and easy) to send it where you want... and the amount of sensors you'd need to make it fully auto... well... it's simply a dream

I think something like the Arduino Mini board could suite fine, since it has just to make some simple math and drive the Dynamixel bus. I was more worried about the wireless connectivity... wondering if there is anything more efficient than the ZibBee for the range.

About the power source... of course it must be onboard... I am currently trying to fit one battery together each motor... I should be able to get from 4.4 to 6.6 Ah... enough to run it for abt one hour... I hope

[Zypkin]

Considering the space constrains I finally decided to go for the Arduino stamp as main controller for the Snake.
For wireless connectivity I will add Xbee Pro module from Maxstream which is rated for up to 1mile of outdoor coverage... my friend tested it successfully up to 1400m... impressive
You can see the datasheet here: http://www.maxstream.net/products/xbee/ ... 2.15.4.pdf

Using these modules the robot will be remotely driven from PC where the operator can choose the proper gait and behavior.
For power all the thing up I'll use at first 3 Lipo cells, 2200mAh. I can eventually increase to 6 or 9 in future.

While I'm waiting to get all the parts I need I'm working on the gaits... so much fun to drive things using math functions
...but to carry around a tether is definitely not the best way to make experiments

Write again when I have something more ready
Zyp

[limor]

Considering the space constrains I finally decided to go for the Arduino stamp as main controller for the Snake.
For wireless connectivity I will add Xbee Pro module from Maxstream which is rated for up to 1mile of outdoor coverage... my friend tested it successfully up to 1400m... impressive
You can see the datasheet here: http://www.maxstream.net/products/xbee/ ... 2.15.4.pdf

Using these modules the robot will be remotely driven from PC where the operator can choose the proper gait and behavior.
For power all the thing up I'll use at first 3 Lipo cells, 2200mAh. I can eventually increase to 6 or 9 in future.

While I'm waiting to get all the parts I need I'm working on the gaits... so much fun to drive things using math functions
...but to carry around a tether is definitely not the best way to make experiments

Write again when I have something more ready
Zyp

1400m!! ? thats amazing. please document the process and share your experience. Bluesmirf is tested with Bioloid and i've personally had great experience with it but it is designed to work at 10-20m or so.

Also, I dont recall someone talking about LiPo'ing their Bioloid on the forum.
Can you please specify which Lipos you use, voltage regulator, low-voltage cutoff, etc.? do you connect them in parallel? do they fit into the CM5 cavity? thanks!

[cdraptor]

Zypkin

I look forward to updates. Hopefully you document the process so you can share some details with the community

[Zypkin]

Sure, I'll give you a detailed description as soon as I manage to have all the components on the table

In the meantime have a look at the snake upgrades to 16DOF, in static pose: http://picasaweb.google.com/zypkin/TheSnakeSFactory

Just one note: it cannot be driven using the motion editor cause you need to send sine waves along it's length... this means you have track realtime the position of each of the 16 servo and reposition them following the function which generates the gait... so the configuration it's changing in time and cannot be predefined.
Also, the CM5 is too big and heavy to be used as it is... you have minimum to take out the board and reposition the batteries onto the snake... or use a long tether... which is what I'm doing now while I wait for the other components.

Zyp

[Biodog]

That's funny, you're realizing what robotics reseachers are on since a few years, and I'm not sure they have your results
Your "sine wave" stuff is very interesting, and I hope to see a video of your movements soon!

@Limor: not sure about it but I think JonHylands uses Lipo packs in his bots. Personnally I use a 11.1V 4800mAh 3s1p Lipo Pack in the RobuDOG, and it can be directly connected to servos or pass through a charge/discharge card.

[Zypkin]

That's funny, you're realizing what robotics reseachers are on since a few years, and I'm not sure they have your results

What if I am one of those researchers? ...LOL...
...I didnt start yesterday to make robots, so I guess I know pretty much what I'm talking about
I didnt write it is an easy thing to do... I'm just working on it... and so far I've done good, meaning that the thing is moving, strafing and rolling. I am kind of excited for the results... and that's one of the reasons for which I'm here sharing with you.

Anyway... if U run a search on Google or Youtube you will find several examples of the very same thing I'm working on. Sure, there is much less than on any other kind of robot... just because driving an hyper redundant multilink robot is much more complex than driving 2 wheels and the hardware is much more expensive
Progression using sine functions on snake-like robots is something known since looooong time ago, I didnt invent anything.
Just have a look at this: http://voronoi.sbp.ri.cmu.edu/projects/ ... gaits.html

If you go on Citeseer (Robotics search engine for so called researchers) you can easily find papers like this one: http://citeseer.ist.psu.edu/cache/paper ... 97path.pdf
...which is quite complicated but tells you a lot on how far you can go with planning the motion of your robot

Finally... for all the non believers:
http://www.youtube.com/watch?v=JdoPXVQbidw
This is actually the only video I have and it shows the very first successful sequence of movement... therefore it's quite creepy


Enjoy,
Zyp

[BillB]

Very good. I like it alot.

After recently applying the Bluesmirf hack to the CM5 which involved taking the CM5 to pieces it was clear that the bulk of the CM5 is the battery housing and that the CM5 electronics actually is quite a thin lightweight unit.

The CM5 battery appears to consist of off the shelf rechargeable batteries that are soldered in series.

It may be worth getting some rechargeable batteries, attaching them to every other bracket and making a compact housing for the CM5 electronics. That was you will be able to keep the CM5 but distribute its size and weight along the whole snake. And probably be able to use the CM5 to charge the batteries.

Bill

[JonHylands]

@Limor: not sure about it but I think JonHylands uses Lipo packs in his bots. Personnally I use a 11.1V 4800mAh 3s1p Lipo Pack in the RobuDOG, and it can be directly connected to servos or pass through a charge/discharge card.

I don't, yet, but we are planning on going there. I know the servos can handle the 11.1 nominal voltage (I think max voltage would be around 12.6 volts, 4.2 per cell). The default max voltage for an AX-12 is 14 volts, so as long as you're always well under that, you should be fine...

- Jon

[Zypkin]

you will be able to keep the CM5 but distribute its size and weight along the whole snake. And probably be able to use the CM5 to charge the batteries.

Yep, that's the idea... there are 8 AA cells in the CM5, 2300mAh... one could easily use the very same type of battery and attach one of them 8 different segments of the snake.
As I wrote before I need to write a custom firmware for the contoller so I'm gonna lose the charging option on the CM5
Not a tragedy... it's just another good reason to move to LiPo
LiPo batteries offer a more convenient ratio of energy/volume so with only 3 cells (aprox same size) you can have the same power (actually more). This means that using 9 cells (9 segments occupied) you can have 3 more time autonomy.

Zyp

[limor]

I am looking for a proven LiPo solution to offer on RoboSavvy for Bioloid and other robots. Including high rated mAh, charger, low-voltage-cutoff and possibly current balancer.

[billyzelsnack]

If you are going to sell off the shelf I highly recommend this charger..

http://www.fmadirect.com/products.htm?cat=45&nid=4

While you are at it you should also offer http://www.liposack.com to charge your batteries in.