What parts do I need to build the following hexapod robot?

[happy_hippo]

Hello, I recently decided to build a robot     And I want a fast -moving symmetric hexapod (small), like hexapodinno

利基應用六足機器人

Perhaps it's a long shot project but I will have all summer to do it, and hopefully it will cost a bit less than 1300 dollars (cost of those hexapodinnos).

As I understand, the hexapodinno has an onboard "Servo Commander 32", which is a miniature personal single-board computer + servo controller. So, commands are programmed into it. But I want to make a robot that is controlled wirelessly, so that I send commands to each leg from my computer and maybe some sensors send commands back to my computer. I previously used matlab to control 3D-stages via RS-232 protocol (serial port), so I want some similar setup if possible, if not then perhaps some other alternative.

Body
I also decided to built it all from scratch. I'm thinking about buying 18x Hitec HS-55 ,which are very cheap on e-bay and claim to have the following specifications:

• Motor Type : coreless motor
• Gear Type : All Nylon Gear
• Dead Band Width : 2 usec
• Connector Wire Length : 150mm
• Stall Torque : 1.5kg/cm at 4.8V
• Operation Voltage :  3.0 - 7.2Volts
• Dimension : 22mm x 12mm x 29mm
• Connector Wire Length : 150mm
• Operating Speed :  0.12sec / 60 degrees (4.8V no load)

In reality I will not be surprised if these have like 30% lower specifications though. Compare with Hexapodinno servos: torque: 1.8kg / cm, speed: 0.13sec / 60 degrees.

I will base my leg design loosely on either this:

Arduino Robot - Quadruped - prototype - body kinematics test 2

with wooden legs + hexagonal base.  Or http://letsmakerobots.com/node/26107 where polymorph+chopsticks were used.  Which do you think is better for faster moving hexapod?

I will try to write a program where the hexapod will be moving like in the hexapodinno video, i.e. 3 legs support, so I think the body has to be somewhere 0.5-0.8kg maximum.

Electronics

I haven't really looked into this in detail yet, but I will probably get two Li-Ion batteries, one for electronics, one for servo.

Bluetooth transiver or Wifi? not sure on this one.

And maybe Axon microcontroller or similar? Also haven't looked properly into this.

Any suggestions/ideas will be greatly appreciated!
Thanks!
(P.S. I'm doing a degree in Physics)

[blackbeard]

I'm certainly not one for kits and i usually wouldn't suggest this but in the case of a hexapod of that complexity i'd get a kit and use the axon or axon 2 as the "brain" and you could use either an RC controller of blue tooth which is very cheep. either way you are going to have to use a microcomputer of single board computer of SOME type. now assuming you're not going to listen to me about the kit thing here's what i would suggest, look at the design of these robots and design the parts using cad software. if you're in college or university i almost guarantee that they have a CNC mill or maybe even a 3d printer. this is NOT easy and you should seek help from those who know how to use these pieces of machinery but otherwise you can print your own pieces and it will be much better quality.

[happy_hippo]

thanks for your reply blackbeard, I think I will make a very basic body for now (although indeed we have a 3D printer at university and I may consider it at some point).

I have a couple of follow-up questions:

(1) What is the difference between AXON and AXON II? And will it make any difference to my proposed robot (16 servo hexapod + up to 5 sensors I guess).
And can I buy this anywhere in England/Europe?

(2) And bluetooth -wise, as I understand I can connect a bluetooth card to the AXON controller via UART, and then pair the bluetooth transiever with any bluetooth enabled computer, and then create a virtual com port on the computer, which will give access to the AXON via hyperterminal (and therefore programs like MATLAB). So that I can control each servo individually.
Will something like this http://cgi.ebay.co.uk/ws/eBayISAPI.dll?ViewItem&item=200554920545 do the job?

(3) I do not intend to store large routines in the controller, maybe some simple leg movement routines, which I will execute from matlab.  Am I correct to say that the controller works in this way? Or will there be a significant delay between sending a command from my computer and its execution?

Thanks!

[blackbeard]

thanks for your reply blackbeard, I think I will make a very basic body for now (although indeed we have a 3D printer at university and I may consider it at some point).

I have a couple of follow-up questions:

(1) What is the difference between AXON and AXON II? And will it make any difference to my proposed robot (16 servo hexapod + up to 5 sensors I guess).
And can I buy this anywhere in England/Europe?

What's the difference? i don't know but do i know that both have been used for hexipods of that type and i'm sure either can be suitable for your purpose. ask Admin if you want more details since he's the guy who designed them . in any case you can buy them here on SOR and i think they ship world wide but admin can surely fill you in more on that. i don't know if an arduino mega would be suitable but it is cheaper and has allot of i/o however the axon is far superior for robotics.

(2) And bluetooth -wise, as I understand I can connect a bluetooth card to the AXON controller via UART, and then pair the bluetooth transiever with any bluetooth enabled computer, and then create a virtual com port on the computer, which will give access to the AXON via hyperterminal (and therefore programs like MATLAB). So that I can control each servo individually.
Will something like this http://cgi.ebay.co.uk/ws/eBayISAPI.dll?ViewItem&item=200554920545 do the job?

Yes. if i understand it works like a serial port and you should be able to make a VB program that elegantly controls it. i haven't used it so i don't know what unit is best.

(3) I do not intend to store large routines in the controller, maybe some simple leg movement routines, which I will execute from matlab.  Am I correct to say that the controller works in this way? Or will there be a significant delay between sending a command from my computer and its execution?

Thanks!

Any microcontroler you get will have it's own programming environment. you controller will have the routines for things like moving, turning, stopping, reading sensors, killing humans, sending data (if desired) and basically doing what a robot does. the commands sent from your computer should be very simple such as sending a byte which the program on your microcontroler will recognize as the command to perform the pre programmed tasks.

[georgeecollins]

This is a very do-able, project, particularly since you are starting with a hexapod.  Hexapods are much easier to start with then quads or bipeds. 

An HS-55 is not a coreless motor, it is a three pole motor.   HS-55s do not have a lot of torque overall but they have very good torque for their small size and weight.  They should also be pretty inexpensive, I see they are ~$10 new.  They are really good for the size and price.  But if you are going to use small servos you should plan on building your hexapod very light.  Make the frame of light wood or thin aluminum.  You will need a small light battery.  Also, be sure to not make the legs too long, since the longer the lever, the less the force. 

An alternative could be to use standard size servos which also can be found relatively cheaply.  An HS-322 is about $8 and has twice the torque.  But it also weighs much more.  An HS-422 is $10 and has a bit more torque then that.   With either of these choices you will get a bigger, heavier bot, but perhaps more lift to carry your battery. 

Finally, digital servos will tend to "shiver" less and have better holding torque then regular hobby servos.  They do cost more, but you can see the improvement with a HS-5485HB.  That puts the cost into another area though, so I think I would try my first one HS-55 or HS-422 to save $. 

[happy_hippo]

thanks for your replies

georgeecollins, I actually just looked at that hexapodinno robot manual (http://custobots.com/store/images/Hexapodinno%20Instruction%20Manual_v117.pdf), and there it gives different specifications for the servos it's using (S03T servo) :
-standard size: 73g
- 0.33 sec/60 dgrees
- 7.4 kg/cm !!!

Maybe it's so fast because of these servos? In the video those servos look quite small. But I guess standard size is not so big either. I guess the speed of servos is not really relevant in hexapods, since a fraction of a second is quite fast, but shifting load quickly (i.e. high torque) is much more important. Is this correct? Should I start looking for much higher torque servos?

Also, maybe I just need 6 high torque servos (to save money), that rotate the legs horizontally, because other servos (joint) are not used as much when the robot walks, and I want a fast robot, otherwise it's boring.


Thanks

[georgeecollins]

Hexarduino uses standard size servos.  It's built to carry more weight.  Those legs are made with thick plastic, and it has a deck you can mount stuff on.  You can build a hexapod with HS-55s it just has to be small and light. 

Frankly, it just looks like a normal speed hexapod to me.  If you want to see a fast hexapod, try :
http://www.trossenrobotics.com/p/PhantomX-AX-18-Hexapod.aspx

(For a quad, my own Sterylite6000 was pretty fast)

As far as what you need:
- You need an appropriate controller board. 
- You need 12 or 18 servos (2 DoF or 3 DoF)  Don't discount starting with 2 DoF, it will save a lot in cost to start with and you probably can convert it. 
- An appropriate battery.  Probably LiPO or NiMH
- Something to make the frame out of.  Probably sheet aluminum or light wood.  You can get kits, but it will cost more. 
- Servo cables. 

[happy_hippo]

hello again

Apparently there are hexapodinno and mini hexapodinno robots, with different servos. The mini one has 1.6kg/mm it says in the manual, which I assume should be 1.6kg/CM. And speed 0.09 s/60, dimensions: 22.8x12.0x25.4 mm, 13.6g.  So, I guess HS-55s are not so different after all, just need to build a small robot.  I will use AXON microcontroller, and I guess it's quite small and light.

georgeecollins, 12 DoF is a good idea! I will do that first. And probably it will be less programming to do, since I want to use MATLAB, and there's not much out there on how to do it. Thanks.

I have exams coming up, so I will be hunting for parts on the internet for now, and hopefully start building something after I'm done with my exams and got all the parts.

Thanks for your help everyone!

[happy_hippo]

I have access to a 7.4 V, 1800mAh Li-ion battery and a 3.7V 1930mAh Li-ion battery, but Axon II controller requires 6-7V (but up to 16V) and the servos I have ordered are rated at 3 - 7.2V. So as I understand I have 4 options:

-connect the 7.4V battery to AXON2 and servos (12 of them) to unregulated ports and hope for the best  (0.2V over rated volatge)

or

-connect the 7.4V battery to AXON2 and servos to regulated 5V ports (there are 16 of them). Is this possible?

or

-connect 7.4V battery to AXON2 and the 3.7V battery to servos with the dual battery configuration of AXON2. Although this will add a bit more weight and 3.7V is quite low for those servos I think, I'd like them to be close to 6V.

or

-get some sort of device that could reduce 7.4V voltage to lower, e.g. a diode (as someone suggested on forums it can reduce voltage by ~1V or something)or power regulator , although I would prefer not to spend a huge amount of money on this or add more wieght.


Which of these is the best solution? or perhaps other better solutions?  

Thanks  

[rbtying]

-connect the 7.4V battery to AXON2 and servos (12 of them) to unregulated ports and hope for the best  (0.2V over rated volatge)

7.4v is a nominal voltage, they can hit almost 9v when fully charged.  Your servos will probably live, but it's risky, and will reduce their lifetime.

-connect the 7.4V battery to AXON2 and servos to regulated 5V ports (there are 16 of them). Is this possible?

You will burn out the regulator on the AXON2.  So no, it's not possible, at least without changing something.

-connect 7.4V battery to AXON2 and the 3.7V battery to servos with the dual battery configuration of AXON2. Although this will add a bit more weight and 3.7V is quite low for those servos I think, I'd like them to be close to 6V.

Possible, but not preferable, for the reason you mentioned - at 3v, the servo will be running at a minimal power state.

-get some sort of device that could reduce 7.4V voltage to lower, e.g. a diode (as someone suggested on forums it can reduce voltage by ~1V or something)or power regulator , although I would prefer not to spend a huge amount of money on this or add more wieght.
Thanks 

Diodes won't work well, since the current draw will vary depending on what action your hexapod is going through.  What you want is a switching power supply, rated for the servos you are using.  One of these is probably good enough, and it's about the cheapest you're going to find, as well.  (micro servos draw ~0.5A each, 16servos * 0.5A/servo = 8A)

[happy_hippo]

Hmm, if it's so tricky, I think I will just get a 4AA battery holder and solder a hitec lead to it 
Thanks

[Gertlex]

Hmm, if it's so tricky, I think I will just get a 4AA battery holder and solder a hitec lead to it 
Thanks

You can also do 5 AA (5 x 1.2 to 1.3 = 6 to 6.5 V) rechargeables in a 6 AA holder, with one empty battery spot jumpered.  Like this: http://www-personal.umich.edu/~erelson/files/BatteryShortedSpring.png

(The second spring is what happens when the battery pack shorts itself.... =-x )

Also, I recommend an assortment of battery holders.  They're good to have for testing stuff.

[georgeecollins]

Happy-

Not to make your life harder, but you may not want to use AA batteries for this robot.  They often have a long life, but they weigh pretty much for the voltage you are getting.  It's hard to generalize, but AA's can weigh 80g a pop, plus you have a holder.  That can get you to 400+ grams pretty quick.  And you are using minature servos as I recall. 

You could try it, but if you have poor ground clearance think about a different battery. 

George

[georgeecollins]

Here is an example of a battery that I think would be lighter than 5 AAs:
http://www3.towerhobbies.com/cgi-bin/wti0001p?&I=LXVRH2&P=7

I have never used that particular battery and you may just be fine with the AAs.   But that is an example of something you could do if you wanted it to be lighter. 

[calvino]

I think you should use Lipo batteries, they are small and have a good energy density (I think is the best at this point) http://www.hobbyking.com/hobbyking/store/uh_viewItem.asp?idProduct=8979 here you have a 7.4V with 2200maH, but you can find it with more life, for example this one
http://www.hobbyking.com/hobbyking/store/uh_viewItem.asp?idProduct=10277 7.4V with 5500maH, but I think this is too big, the point is that I think this will fit your needs.

[georgeecollins]

While LiPo batteries are much better, you may not want to run the servos you are purchasing at 7.4 volts.  Does the controller you are using have a voltage regulator

Also, LiPos are more expensive in general, particularly once you include the charger.  I am assuming you are on a budget and I think there is a middle ground between AAs and a LiPO pack. 

[happy_hippo]

I'm using AXON II as my controller. It arrived a couple of days ago
It took me maybe half an hour to install all software and flash led-demo program. Even on my Windows 7 64 bit. I connected it to 4 AA battery pack (I described my LiPo options in my previous post, but as I was told none of those options will work, unless I buy some other parts). And it seems that this 4 AA battery pack will fit nicely in my robot body.

So, I think it should be fine like that. I'm just waiting for my 12 servos and bluetooth card to arrive  
I will probably make a video of what I've been doing at some point. But that probably in the beginning of July, as I need to concentrate on my exams now  
Thanks

[happy_hippo]

I just got my 12 x HS-55 servos from China, they look a bit dodgy though. But anyway, I will try them for now.
The first problem I encountered is attaching a head to the servo (the circular one)

As I understand I have to put the white head on top and fix it with a screw. But there is no thread for the screw. So I have to push it quite hard with a screwdriver to make a thread, and I accidently twisted the servo.

Does it mean that my servo is misaligned now or even broken? I haven't tried attaching it to AXON yet because I want to attach those head to my 12 servos first.

Thanks

[happy_hippo]

Hello, I finally got my bluetooth card from China. And I decided to test everything a little , exactly like in Admin's bluetooth (UART) tutorial.

I connected my bluetooth to UART 2 on Axon, 1 servo to H3 terminal, and battery. Then I flashed "computer_controlled_example" from admin's program list.

http://dl.dropbox.com/u/26776513/computer_controlled_example.c
http://dl.dropbox.com/u/26776513/hardware.h

And when I connect to hyperterminal, bluetooth lights stops flashing and just becomes lit(so it's connected). But hyperterminal is not doing anything (no text) and nothing happens when I press E or R.

I attached a picture of my configuation:



I guess it may be the code, I've never seen this kinda code before maybe someone could help me to write some code for this configuration?

Servo: HS-55 (connected to H3)
Bluetooth: connected to UART port U2 on AXON II . On computer it's com5, baud rate at 38400, Data bit:8, Stop bit:1, No Parity and Flow control (bluetooth card settings: from manual)
Axon2 controller.
Battery 4x1.5V AA batteries.
I want to do what admin does in his bluetooth tutorial (opeining and closing robot hand):

ŪN¬

I got it working with USB, but it's not working with bluetooth
Thank you very much!
      

[happy_hippo]

oh man.. they incorrectly labelled RXD and TXD on their bluetooth card... took me all day to figure it out. But it's working now!

[happy_hippo]

Hello

I decided to make a video of how I was building my budget hexapod   (controlled via UART from MATLAB).

It has latest software installation and configuration guide: AVR studio 5 with Webbotlib 2.06 + Project Designer, WinAVR, McUber and it's tested to work on Windows 7 64 bit (32 bit probably too) and hardware description (although it doesn't look very professional with wood, but I will make proper parts on a 3D printer soon   )

http://read-stuff-here.blogspot.com/2011/08/making-hexapod-robot.html

Currently I'm trying to design a walking routine, but it's so complicated in 3D with all those angles and stuff

Anyway, I hope it can be useful for someone  

[georgeecollins]

You know before I said I didn't recommend AA batteries as a power source but now I think I was really wrong about that.  I tried paring AAAs with a micro servo bot and it works pretty well.  I don't think it is great for the environment to use disposable batteries but the power to weight ratio isn't bad.  

I don't know what I was thinking when I made this post.  I have been trying to make the cheapest walking robot I can (you can see it on my site www.backyardrobots.com, I just put up a picture) and so I thought with little servos HKS-939MGS (not too different then HS-55s) I could get away with 4xAAA.  I could work the the servo or the controller board with AAAs, but not both at the same time without brown outs.  So I switched 4xAA with off the shelf alkaline batteries, trying to keep it cheap.  That works, but the battery lifetime for a robot is not very good at all. 

Alkaline batteries have a relatively long life but they suffer relative to NiMH in situations with high current draw.  In high current draw situations the voltage drops really fast.  I have read pages where people have found alkaline batteries drop below 1v in less then an hour at 500mah current draw.  Guess what?  Servos draw a lot of current under load, especially 8-12 servos lifting a robot.  It wasn't that bad testing poses on my work table, but walking across the floor with obstacles the batteries really only lasted about fifteen minutes.

After fifteen or twenty minutes the servos would stall or the controller would brown out or reboot.  If you think of that voltage drop curve at 500mah, you are failing long before the batteries drop below 1v.  At 4x1.2v you are getting below the recommended level for
controller and the servos.  I am not using the same servos or the same controller, but I kind of doubt it will be that different when you start sending it out on extended runs. 

That's going to get worse when I add a Ping sensor.  They are really sensitive to low voltage conditions and I have had brown out problems with in robots with much larger batteries (aka fluffy).  They fail < 5v for sure. 

Two things seem possible to mitigate this.  One thing I am going to try 4xAA with Lithium batteries.  They are higher voltage and supposed to be able to sustain higher current drains much longer.  The problem is they aren't really that cheap-- it's like $2.50 a battery.  The other option is more tried and true, a 5x 1.2v NiMH pack.   A good flat pack is $25 and you need to have a charger.  But NiMH batteries hold up well under a lot of current drain and in the long run a rechargeable battery is probably a better bet then one you are going to throw away.