17 degrees of freedom with 17 servos
William Heath
I read an excellent introduction to some of the most advanced biped robots. In this tutorial it described using 17 servos to make the robot. They said the robot would cost 1000.00 dollars. I am a member of the Tech Shop and have access to a laser to cut acrylic etc... I don't understand why it costs 1000.00 to make a robot like this. Here is my idea and the costs involved:
1. Tmobile g1 (used) 100.00
2. arduino bare bones (15.00)
3. 9 volt battery (2.00)
4. 17 micro servos (17x3.50 = 59.5) http://www.hobbypartz.com/topromisesg9.html
5. Acrylic (20.00)
Total = 196.5
My curiosity is how to interface with the 17 servos. If I am not mistaken they all take pwm and there are only 4 pwm's on the arduino bare bones. I have heard of a biped type robot that uses only 6 servos. Anyway, I finally understood the idea of making these advanced robots. I want to build on my current understand of circuits, arduinos etc... to understand more of how to create a cool biped. Can anyone enlighten me?
Walt Perko
Hi,
All these robots have 16-24 $60+ servos … that’s where the money goes … the brackets and controller are essentially FREE.
http://www.lynxmotion.com/Category.aspx?CategoryID=85
http://www.kondo-robot.com/html/Info_En.html
Things to do with your bipedal robot …
http://www.r2pv1.com/media/81-20080631-RoboNova-1-ShootsBigBadBobBandit2m.wmv
http://www.r2pv1.com/media/81-20080630-RoboNova-1-LaserGunShotMovingTarget.wmv
And the part CAD files are available for FREE downloading;
http://www.3dcontentcentral.com/Search.aspx?arg=robonova
=========================================================================================
C U L8r, °|° Walt Perko °|° "Kids ... teach
them the good stuff, and they still learn the bad stuff."
OPEC of the West California
"The World Needs a New Economic
Model"
--
You received
this message because you are subscribed to the Google Groups "HomeBrew
Robotics Club" group.
To post to this group, send email to hbrob...@googlegroups.com.
To unsubscribe from this group, send email to
hbrobotics+...@googlegroups.com.
For more options, visit this group at
http://groups.google.com/group/hbrobotics?hl=en.
Rupert Hart
Interfacing is easy. Buy a serial servo controller, SSC. This one board will control up to 6 servos from one data pin of your microcontroller. See www.parallax.com and other sites. I have several boards all linked up for a 4ft3 head with robotic creatures inside it. Best. Rupert
From: hbrob...@googlegroups.com [mailto:hbrob...@googlegroups.com] On Behalf Of William Heath
Sent: Monday, December 21, 2009
4:02 PM
To: hbrob...@googlegroups.com
Subject: [HBRobotics] 17 degrees
of freedom with 17 servos
Hi All,
--
Steven Nelson
--
Randy M. Dumse
> My curiosity is how to interface with the 17 servos.
A ServoPod(TM) made by New Micros, Inc., has 26 PWM outputs
capable of controlling RC Servos. (Note: I work for NMI). All
these PWM and Timers are implemented in hardware, so you set
them and forget them and they generate PWM continuously without
further processor intervention. The unburdened-by-PWM 40 MIPs
DSP processor is very good at doing IK in floating point as
well.
The ServoPod(TM) isn't as cheap as a Arduino, then again, the
ServoPod(TM) draws much more current, is about maybe 4 to 10
times more powerful than an Arduino, and has far more hardware,
far more pins, needs no SSC to do 26 RC Servos, needs no shield
to connect to 26 RC Servos.
You can see a ServoPod(TM) on a EH3 Lynxmotion robots with 18
servos here:
http://www.lynxmotion.com/images/html/build057.htm
Randy
Nathaniel Lewis
noticed in your description of the robot that you are powering it with a
9V battery. I don't think it will give enough power to run a 17 servo
robot for too long. You can look at Li-Po batteries that use use in
electric RC airplanes and helicopters and they will give you 7.4V at
1200 mA/H in a package twice as wide, twice as thin, and a third of the
weight. I use a big variant in my Stinger robot.
Nathaniel
> --
>
> You received this message because you are subscribed to the Google
> Groups "HomeBrew Robotics Club" group.
> To post to this group, send email to hbrob...@googlegroups.com.
> To unsubscribe from this group, send email to hbrobotics
Nathaniel Lewis
looked into one awhile ago, you had to buy a compiler for the DSP
onboard and it wasn't cheap. I would use a dedicated servo controller
to control all of the servos.
Nathaniel
> --
>
> You received this message because you are subscribed to the Google Groups "HomeBrew Robotics Club" group.
> To post to this group, send email to hbrob...@googlegroups.com.
> To unsubscribe from this group, send email to hbrobotics+...@googlegroups.com.
Randy M. Dumse
> Don't you need a $200 compiler to write programs for it
> though? When I looked into one awhile ago, you had to buy a
> compiler for the DSP onboard and it wasn't cheap.
Not that I know of. There has never been a need to buy a
compiler for the DSP. Don't really understand what you are
talking about.
The ServoPod(TM) costs $199. It comes with an interactive
IsoMax(TM) installed. That, and the free download NMITerm
terminal program from our web site, are all you need to develop,
download, test, flash, autostarted, etc., your programs.
You can use other languages if you like, but our IsoMax(TM) is
our state machine based multitasking software. It is unique to
us, but we include it in the price of the board.
Randy
Bob Smith
starting to define the Linux driver interfaces for our
FPGA-based quad and octal servo controllers. I would
appreciated your advice.
-- What is the best way to specify the position of the
servo? As a start and stop time? As a percent of
full travel? As an angle?
-- Do we need to specify the minimum and maximum times
for the servo if we use percent of full travel for
position?
-- Should we have slow start and slow stop?
-- To slowly move a servo you need to slowly change the
pulse width. How is that be specified? Nanoseconds
per second?
-- Any other thoughts or advice?
BTW: If anyone is interested in being an alpha customer
for this, the FPGA board cost $99 and an FPGA image for
servo control is free. If you don't need all 32 lines
for servo control you can add quadrature decoders and/or
H-Bridge controllers into the FPGA mix.
thanks
Bob Smith
Alan Marconett
position, and the time for the move.
I'd also provide a "coordinated move", such that a group of servos all
start and move at the same time, and complete their moves at the same time.
Having a set of limits for the servos would off load the program
controlling sending the commands.
The initial start up of the servos should be controlled, as their
position isn't known at the time of the startup. This could keep the
servos from an "initial jam".
Alan KM6VV
Rupert Hart
for each segment. And related, a way of ensuring smooth continuous motion,
not jerky motion from one position to another.
Best, R
-----Original Message-----
From: hbrob...@googlegroups.com [mailto:hbrob...@googlegroups.com] On
Behalf Of Bob Smith
Sent: Monday, December 21, 2009 9:25 PM
To: hbrob...@googlegroups.com
thanks
Bob Smith
giuliano carlini
servo library as a much less expensive servo controller. Or use a 644p
to get more pins.
Cost $5-$10.
Just feed it +5 and Gnd from an arduino, and use I2C (A4 and A5) to
communicate between the arduino and the offboard 168. That leaves 16
pins to control servos with a 168. Or something like 28 with a 644p.
giuliano
Rupert Hart
atom microcontroller. Servo control board is about $50, microcontroller
about $50 in the ready-made board I use. That can be programmed thru serial
for no cost. Total cost $100 plus servos.
R
-----Original Message-----
From: hbrob...@googlegroups.com [mailto:hbrob...@googlegroups.com] On
Ted Larson
called ISOMax which is FORTH based, that comes for free with it. So, if you
don't mind learning a new programming language, it is very good at what it
does.
-Ted
-----Original Message-----
From: hbrob...@googlegroups.com [mailto:hbrob...@googlegroups.com] On
Behalf Of Nathaniel Lewis
Sent: Monday, December 21, 2009 8:38 PM
To: hbrob...@googlegroups.com
Subject: RE: [HBRobotics] 17 degrees of freedom with 17 servos
William Heath
-Tim
Bob Smith
> I'd go for specifying the servo time (500-2500 uS) to indicate the
> position, and the time for the move.
we'll specify this in nanoseconds. Probably overkill for most
servos but someone might like 50 ns resolution.
> I'd also provide a "coordinated move", such that a group of servos all
> start and move at the same time, and complete their moves at the same time.
OK, yes, the SSC controller has this. We can give the Linux
driver a group step with a resolution of about 20 ms. That should
be precise enough I hope.
> Having a set of limits for the servos would off load the program
> controlling sending the commands.
But burden us with documenting them and the user with then
reading and understanding our documentation. Hmmmm...
> The initial start up of the servos should be controlled, as their
> position isn't known at the time of the startup. This could keep the
> servos from an "initial jam".
OK, yes. I was going to leave the servo pulses completely
off until the controlling program gives the first change command.
Your note 1(position and time) should be sufficient to prevent
an initial jam, right?
Rupert Hart wrote:
> Bob, one thing that I could use is a way of specifying angle/sec, preferably
> for each segment. And related, a way of ensuring smooth continuous motion,
> not jerky motion from one position to another.
OK, yes. We have angle/sec already on our quadrature decoder
so this is easy to do. It lets the "higher layer" robotic
software remain unchanged if you swap out one servo for another
with different specs. Since degrees/sec quickly becomes floating
point, we'll also offer "raw" access with is either 16 or 32 bit
integers. We assume 32 bit Linux but don't want to assume that
floating point hardware is available.
giuliano carlini wrote:
> Should be able to use an offboard atmega 168 (or 328) running the mega
> servo library as a much less expensive servo controller.
> Cost $5-$10.
Right. We are not trying to be a low cost leader. We are
trying to be the ease-of-use leader for medium size robots. To
us this means ARM or a PC running Linux. Using an FPGA gives us
50 ns resolution on all the pins all the time if we want it, but
the real advantage of an FPGA is that the you can select almost
any combination of 9 peripherals from a list of 25. Our intended
sales flow is something like:
- Buy a Baseboard
- Go to the web site and select 9 peripherals for it.
- We do an FPGA build with those 9 peripherals. (It's automated)
- We send you the FPGA image
- You download the image to the FPGA board using the 'cat' command
- You run our open-source user-space driver daemon
- All 9 peripherals now have /dev/ entries accessible to your software
(At start-up the FPGA tells the daemon which drivers to load)
Alan, Rupert, Guiliano: Thanks.
Bob Smith
Ted Larson
The $2 servos you posted, only have 10oz/in of torque, and have plastic gears inside that will strip or break easily when a large load is placed on them. Most bipedal robot designs, have tremendous forces placed on the servo, just to maintain a standing position, let alone actually walk. This is why most people try to use servos with metal gears in them, and why servo manufacturers have clued in, and starting making more robust servos specifically designed for robots. With that said….I am not saying you can’t design a robot that uses a bunch of the $2 servo’s you posted….it’s going to depend a lot on the design, and how the forces act on the servo’s.
-Ted
Walt Perko
=========================================================================================
C U L8r, °|° Walt Perko °|° "Kids ... teach
them the good stuff, and they still learn the bad stuff."
OPEC of the West California
"The World Needs a New Economic
Model"
-----Original
Message-----
From: hbrob...@googlegroups.com
[mailto:hbrob...@googlegroups.com]On Behalf
Of William Heath
Sent: Tuesday, December 22, 2009
10:36 AM
To: hbrob...@googlegroups.com
Steven Nelson
Walt Perko
Hi,
Check with Trossen Robotics online … I bet they can get you something in that range for much less …
Nathaniel Lewis
servos accept the same pulse range but have different abilities on position,
it would need tuning. Not that it would be too much of an issue, but the
tuning software would have to be built into the controller to accurately
send it units of degrees or radians per second.
Nathaniel
-----Original Message-----
From: hbrob...@googlegroups.com [mailto:hbrob...@googlegroups.com] On
Behalf Of Rupert Hart
Sent: Monday, December 21, 2009 10:55 PM
To: hbrob...@googlegroups.com
Nathaniel Lewis
Ouch. There is the robot fighting back from picking it up too much.
Nathaniel
Nathaniel Lewis
languages other than the one included.
Nathaniel
Nathaniel Lewis
Nathaniel
-----Original Message-----
From: hbrob...@googlegroups.com [mailto:hbrob...@googlegroups.com] On
Behalf Of Bob Smith
Sent: Tuesday, December 22, 2009 10:43 AM
To: hbrob...@googlegroups.com
Alan, Rupert, Guiliano: Thanks.
Bob Smith
--
Nathaniel Lewis
I lost the link to the $2.77 servos. Can anyone post it again?
Nathaniel
From: hbrob...@googlegroups.com [mailto:hbrob...@googlegroups.com] On Behalf Of William Heath
Sent: Tuesday, December 22, 2009 10:36 AM
To: hbrob...@googlegroups.com
Nathaniel Lewis
Nathaniel
-----Original Message-----
From: hbrob...@googlegroups.com [mailto:hbrob...@googlegroups.com] On
Behalf Of giuliano carlini
Sent: Monday, December 21, 2009 10:57 PM
To: hbrob...@googlegroups.com
Bob Smith
> If you don't mind me asking, what is the target price?
The FPGA board is $99 and "Pin Peripherals", as we
call them, are included in the price of the board.
Once you own a board you can request as many different
builds of Pin Peripherals as you want.
The logic and USB interfaces for the peripherals are
in the FPGA. You still need to build any drive circuits
needed for your application. A picture of a system with
some demo boards is available here:
http://www.demandperipherals.com/gallery/peripherals.jpg
Clockwise from the top are:
servo4 (with three servos attached)
hbridge2 (driving two gearhead motors)
ping4 (a quad ultrasonic distance sensor interface)
patget4 (driving an LCD display)
stepu (unipolar stepper motor)
stepb (bipolar stepper motor)
count4 (event/frequency counter with a light-to-frequency circuit)
quad2 (with just the one UI control knob)
bb4io (buttons and LEDS on the FPGA board itself)
The LCD is demo of a RAM-based pattern generator. The
pattern it generates drives a set of shift registers
with output latches tied to the LCD segments.
Not shown in the photo are:
dc2 (dual DC motor controller)
irrec (Consumer IR receiver
irxmit (Consumer IR transmitter)
maxbotix2 (dual chain Maxmotix range finder)
pwmin4 (quad PWM input measurements)
pwmout4/8 (quad/octal PWM output)
rcrec (RC radio decoder)
simple4/8 (basic 4/8 bit bidirectional I/O)
tonegen (tone generator)
watchdog2 (dual watchdog timer)
SPI (simple four wire SPI)
SPI16 (SPI bus with a 4 bit address)
FSRI ("Far Shift Register Interface")
FSRI16 (FSRI == SPI on a 10 foot ribbon cable)
The Verilog is done for the above and we are now doing
the Linux device drivers. My recent questions about
desired interfaces are so we can better design the Linux
interfaces to the Pin Peripherals. Looks like I'll be
asking a lot more questions in the next couple of months.
Please let me know if you want a Pin Peripheral that is
not on the list.
Bob Smith
Alan Marconett
You're calling it an FPGA board, but are apparently running Linux on it?
What core does it have, ARM7, ARM9 etc?
Alan KM6VV
> -----Original Message-----
> On Behalf Of Bob Smith
>
Bob Smith
> You're calling it an FPGA board, but are apparently running Linux on it?
> What core does it have, ARM7, ARM9 etc?
Oops, I'm sorry this was not clear. It has a only a
_USB_ interface. The USB needs to go to your laptop
or single board computer with Linux running. The
FPGA provides the pins and precise timing that a
PC/laptop/SBC can't offer.
Bob
Alan Marconett
tempted to suggest you develop for Windoz as well. With the critical timing
issues off of the PC, then latency should not be as much of a problem. In
fact, doing a CNC controller (servo motors or stepper motors) would then be
a very desirable project for your board. Think FlashCut and others.
Alan KM6VV
Bob Smith
> OK, I see where you're going with it. Sounds quite useful, although I'm
> tempted to suggest you develop for Windoz as well. With the critical timing
> issues off of the PC, then latency should not be as much of a problem. In
> fact, doing a CNC controller (servo motors or stepper motors) would then be
> a very desirable project for your board. Think FlashCut and others.
We would like to do Windows as well but I don't know
any Windows programmers available to work on this.
The protocol and peripheral register definitions will
be publicly available in a couple of weeks. (It is
in final edit and clean-up now). With the protocol
and register definition published anyone could write
a program for Pin Peripherals.
BTW: Clearly the USB cable carries the reads/writes
for all of the peripherals and it would be a real
pain if every robot builder had to write a single
huge program to deal all of the peripherals all at
once. This is where our user-space device driver
comes in. It provides separate device nodes for
each peripheral and does the mux/demux to make it
all work. Here is a diagram that might help.
http://www.linuxtoys.org/usd/robot_drv_arch.png
Alan, order a board and let's put 8 stepper motor
controllers on it. Maybe you could then help define
the Linux interface to your stepper motors!
Bob
Alan Marconett
have too many projects already demanding my attention!
The last time I did a Windoz program, it was for DOS! ;>)
Too bad USB wasn't a little easier to do under DOS. But nobody wants a DOS
program nowadays.
Is that what you're calling the FPGA board, "Pin Peripherals"? Or is that
the driver support name?
Actually, to do a CNC controller module the way I'm thinking, you'd really
want a core processor of some kind, I think. Maybe a few AFSM's would do
it, 'tho. The reason for that would be to off-load the PC of the chore of
generating the actual step/direction bits. Which would be a major off-load
on the USB traffic.
Alan KM6VV
> -----Original Message-----
> On Behalf Of Bob Smith
>
Bob Smith
> Is that what you're calling the FPGA board, "Pin Peripherals"? Or is that
> the driver support name?
The FPGA board is called a "BaseBoard4" and is what we
currently offer on the company web site. Pin Peripherals
refers to our selection of FPGA robotic peripherals that
are included in the price of the BaseBoard. We have not
released Pin Peripherals yet and so they are not on the
web site yet. Later we hope to offer Card Peripherals
which are circuit boards that will contain all the drive
electronics.
thanks
Bob
> http://www.demandperipherals.com/gallery/peripherals.jpg
> http://www.linuxtoys.org/usd/robot_drv_arch.png
George Warner
> OK, I see where you're going with it. Sounds quite useful, although I'm
> tempted to suggest you develop for Windoz as well. With the critical timing
> issues off of the PC, then latency should not be as much of a problem. In
> fact, doing a CNC controller (servo motors or stepper motors) would then be
> a very desirable project for your board. Think FlashCut and others.
Or if you make it HID compliant then it "just works" on Mac, Linux and Windoze…
Basically to be HID compliant you just need a detailed device descriptor…
I'd make it so each pin peripheral has it's own input & output reports and these
are collected into a single device descriptor for the single USB device. Changing
which pin peripherals are connected change the device descriptor.
I've lots of experience with USB/HID devices so if you need any help with this please ask…
--
Enjoy,
George Warner, (408)974-0668
Schizophrenic Optimization Scientist
Apple Developer Technical Support (DTS)
Bob Smith
> Or if you make it HID compliant then it "just works" on Mac, Linux and Windoze�
> Basically to be HID compliant you just need a detailed device descriptor�
We debated this when we did the design of the FPGA board
and decided to use the FTDI USB-to-serial interface. We
don't have a EEPROM tied to the FTDI part so there is no
way for us to get it to USB enumerate as anything other
than the FTDI USB-to-serial port it is.
The complexity of getting and dealing with the USB device
IDs and manufacturer IDs is a little overwhelming. We
wanted to keep it simple starting out.
> I've lots of experience with USB/HID devices so if you need any help with this please ask�
Thanks!! A USB/HID approach may be the right way to go on
our next board design.
thanks
Bob Smith