Trossen Robotics Community Blogs

How does a .Net developer build a robot?

nbdeveloper — Thu, 11 Jun 2009 21:48:25 GMT

�having never done it before it is a challenge. I recall the first time I ran across one of the Trossen YouTube videos. I was pretty interested but after assembling items in the shopping cart I abandoned the idea. This can be an expensive hobby. A year later my curiosity got the better of me and I bought a Phidgets interface kit.

I had been looking for a solution to a �toy� light tree for a race track, not for racing but for tuning my brother in laws 4-wheeler (ATV). I was impressed with how accessible the Phidgets hardware was with the .Net library. So much so, that I am mystified why more developers don�t already know about it. It is sooo easy to get into.

So, a few projects down the road and I�d decided I wanted to build a robot that I can interact with. I guess it was from watching Jimmy Neutron with my kids and thinking �How hard could it be to create a Godard?�, Jimmy�s robot dog. Of course nothing I could make would even come close to that but since being a software developer in a small town (that works for a company in a state far away); I guess I needed a buddy.

Being a high level developer, I get PC�s. I don�t get PWM signals, I don�t get loading Amtel boards with code and those wheelie things that some people say you need to monitor motor positions. There are a lot of developers that won't either.

Having said all that, I chose to buy a Mini-ITX board and build something up from there. Pico ITX is a bit more expensive and Micro ITX is way too big. Nano-ITX is the odd ball of the bunch and still more costly than a Mini. I wanted to order a case for it too but those were too expensive. I found a good one but the site�s stock was out�no telling when they�d have more in. So� I built it out of an electrical box at the hardware store. I have no idea what its original intended purpose was but I found modifying it was easy. Drilling and cutting resulted in smooth clean edges every time.

I also bought a pico power supply but found I needed to also buy an ATX power extension to plug it in. Damn capacitors do it every time!!!

Since the board had a single PCI slot I used a wireless card I had lying around. Buying stuff was getting on my nerves at this point because I had also purchased a web cam and a pan and tilt kit with the servo controller board too. Did I mention this seems like an expensive hobby?

I was barely able to fit in an 80mm fan. I was a bit concerned about heat initially so I left the case IO panel off until I could be sure things were running fine.

I used a SATA drive that was lying around. I was pretty happy with the mounting position once I got it in. I had to zip tie the back of the drive because the mounting holes only lined up with the front of the drive. This worked out in my favor as I was able to rotate the drive upwards giving me room to access the PS beneath it.

So, here it sits on a file cabinet next to my desk, just waiting for my really bad coding skills to give it some kind of life.

If I get the bugs just right I might make CNN as the first man to die at the hands of a robot. J

Webcam capture and image processing - first look

nbdeveloper — Wed, 10 Jun 2009 15:56:16 GMT

I'll say upfront, I've never captured an image from a web cam before. I went looking for a .Net library that is easy to use and full featured. I've looked over these articles:

http://www.codeproject.com/KB/cs/Intel_OpenCV.aspx

http://www.codeproject.com/KB/cs/alip.aspx

http://www.codeproject.com/KB/audio-...detection.aspx

http://www.codeproject.com/KB/graphi...detection.aspx

Two things came to mind after a bit of Googling for .Net capable imaging libraries. One: Using OpenCV as a base is the obvious choice. Two: Chesnokov Yuriy is brilliant.

I found OpenCVdotNet and the code sample seemed pretty easy to do a simple image capture. I installed opencv 1.0 (there's a 1.1 but opencvdotnet doesn't work with it) and tried it out. Sure enough, pretty simple and easy to use. Several lines of code later and I had a streaming (refreshing) image in a winform app. Nice. I also enjoyed reading thier paper on the Tracking of Humans Using Masked Histograms and Mean Shift.

After a bit of looking around in the API I noticed OpenCV had a machine learning area that had not been implimented in OpenCVdotNet. Bummer. I have no idea what that entails but I don't want to be left out. So I went looking around at other libraries. I found a thread on StackOverflow that suggested a few. The one that seemed commented on the most was Emgu.CV. After installing and reading some of the wiki documentation I can tell this is the one to use for a .Net developer.

I created a winform app:

I dropped a picture box control in it and docked it to the form and set the dock property to fill. I also dropped in a timer control and set the interval property to 40 (24 frames/sec ?). I also double clicked the interval event for the timer i the properties dialog to stub out the elapsed method. I love that...no code so far...

After adding in the references to the Emgu libraries and adding the using statements, I created a Capture object in the class and created an image from the capture in the timer interval method then set the image to the pictrebox image and wha-la! "Streaming" video to my winform app.

I decided, without knowing anything about the library (but having a little experience in photography and image editing) that I would try out the methods available in my image. The one I liked the best was Dilate, I think it made my hand look like a robot:

I tried out different methods and added some comments at the end. Now I'll have to go read some documentation to really understand what I'm doing. Pretty sweet though! Here's the code (only 3 lines needed to get a streaming video):

using System;
using System.Collections.Generic;
using System.ComponentModel;
using System.Data;
using System.Drawing;
using System.Linq;
using System.Text;
using System.Windows.Forms;
using Emgu.CV;
using Emgu.CV.UI;
using Emgu.CV.Structure;
namespace OpenCVdotNet_Test
{
publicpartialclassForm1 : Form
{
Capture capture = newCapture();
public Form1()
{
InitializeComponent();
}
privatevoid timer1_Tick(object sender, EventArgs e)
{
Image<Bgr, Byte> image = capture.QueryFrame();
//image._Dilate(10); //pixelize outlines, reinforce shape outlines
//image._EqualizeHist(); //error, needs 8uC1 images?
//image._Erode(10); //shape pixelization - lower contrast, lose shape outlines
//image._Flip(Emgu.CV.CvEnum.FLIP.HORIZONTAL); //duh, it flips the image
//image._SmoothGaussian(10, 10, 10, 10); //doesn't work, something about mask height and width needing to be >= 1
//Bgr bgr1 = new Bgr(1, 1, 1);
//Bgr bgr2 = new Bgr(2, 2, 2);
//image.Canny(bgr1, bgr2); //works, but nothing shows up, probably needs correct values to mark the edges...
//image.Convolution(new ConvolutionKernelF(100, 100)); //works, but nothing shows up, probably needs correct values to convolute the image...
//image.Dilate(10); //no change to image
//image.Erode(10); //no change to image
//image.Flip(Emgu.CV.CvEnum.FLIP.HORIZONTAL); //no change to image
//image.GoodFeaturesToTrack(2, 10, 10, 10); //error, averaging window size must be an odd number >= 3
//image.PyrDown(); //no change to image
//image.PyrUp(); //4 times larger image, no change, probably because the picture box is a set HxW
//image.Resize(240, 320); //um, just resizes
//image.ROI = new Rectangle(100, 100, 100, 100); //zoom, very cool!
//image.Rotate(45, new Bgr(Color.Blue)); //no change to image
//image.SmoothBilatral(10, 10); //error, only square windows of odd size?
//image.SmoothBlur(10, 10); //error, something about mask height and width needing to be >= 1
//image.SmoothGaussian( //same error as above with _SmoothGausian
//image.SmoothMedian(10);//error, something about mask height and width needing to be >= 1
//image.Split(); //splits but no change to image, this must allow manipulating of the RGB layers
//image = image.Sub(new Bgr(Color.Blue)); //subtract color - cool!
pictureBox1.Image = image.ToBitmap(pictureBox1.Width, pictureBox1.Height);
}
}
}

Shameless plug for Trossen

nbdeveloper — Mon, 08 Jun 2009 22:32:31 GMT

http://stackoverflow.com/questions/5...al-life-robots

As of 6/8/2009 I'm last on the page.

StackOverflow

nbdeveloper — Mon, 08 Jun 2009 22:31:04 GMT

http://stackoverflow.com/questions/9...sql-data-types

If you want milk, go to the cow.

Geeks there?

thomasgrantnice — Thu, 28 May 2009 06:56:57 GMT

Was wondering if I could teach some geeks here the uses of Dahn Yoga, and in return gain some knowledge about different technologies. Give and Take, i always an interesting concept.

Wall-e update

mallster — Mon, 25 May 2009 02:29:13 GMT

Might be slow but at least I'm ruff!

Wasn't sure wether to make my Wall-e a brand new or bashed-up unit. But after my rough ass go at painting the thing- made the decision a bit easier.

Problems and wins so far:
Pico P700 stopped working
5A sabertooth too small for the banebots motors and 3" track
The firgelli actuators working well straight from ssc32.
Surveyor svs working well and driving 10A sabertooth. (highly recomend the SVS, support is excellent and continually being improved)
Wall-e's legs working well, independantly moving each side with approx 20mm travel.
Case was laser cut 1mm m/steel and folded for $70 Aus.

I also looked at the bits and pieces left over from this one and found there was enough to make a smaller version (easy to get side-tracked). This one will be a lot more solid and easy to experiment with and will use distance, pir, light sensors with a axon and would like to add vision later on. (see 3d model)

Attached Thumbnails

MomPC: add video and audio outputs to Combo VCR/DCD/TV (progress 001)

kgw — Sun, 19 Apr 2009 08:55:51 GMT

Significant progress!

Found a scope on Ebay from someone very nearby. Today I took readings from many logical locations on the Magnavox combo TV�s motherboard and came up with the following details:

Please check this web page:
http://www.kevinwaite.com/MomPC/

Before I start connecting other video equipment to jumper J1298 � if anyone out there knows about the following measured voltage increases.

J1298 (a) appears to be a composite video signal � but it also appears to be 2 times greater voltage than the input composite found on Ext. V-IN J1297. J1298 (a) carries the selected signal of DVD/VCR and TV.

The RED J1831 (and it shows more trace activity was more red is shown) (same for Green and for Blue) � these signals are about 6 times greater than the input composite found on Ext. V-IN J1297. Test points RED GREEN BLUE test points carry the selected signal of DVD/VCR and TV.

Any input or guidance � is greatly appreciated.

Thanks.

Sincerely,
Kevin Waite

The main Tuner / VCR board found inside is a:
BTD850F01012 x4 Triple Main PCB -- the VCR box is found
above this board. I have found this board for sale as:
VHS PLAYER ASSY BTD850F01012 X4 from MAGNAVOX 27MDTR20/17
Part Type: VHS PLAYER ASSY

Tricks of the trade, usage conventions, and reverse voltage regulators...

RabidCicada — Thu, 16 Apr 2009 12:42:50 GMT

Alrighty,
I've been watching blog activity on trossen and it's been a little slow so I'm back to spice it up with my progress. Work has kept me busy so I just got to sit down last night and finish trouble shooting my servo driver.

Being that my background is in computer science and I am unfamiliar with usage conventions of the atmel atmega168 or typical tricks of the trade for robotics I've learned a lot with my little project.

Current setup has a potentiometer input controlling a servo through it's full range of motion. Full left on the potentiometer puts the servo at full left and same for the right.

Code is located here for those looking for a working example:
http://pastebin.com/f7d5fb77e

Note that if you are going to use that code you need to have the same circuit I have or you'll break something probably. It's just meant to be an example so someone can adapt pieces of it to their codebase(how to use the ADC or timer etc).

Alright...So I encountered quite a few issues.

Reverse Voltage Regulators. Don't know what they do(reversed)...but it definitely doesn't regulate when put in backwards. I've apparently had 9v running through my circuit intended for 5v the entire time. Nice. I found out when I was troubleshooting my ADC which was not doing conversions after I rotated the potentiometer enough. I put my multi-meter on the circuit and found the output from the potentiometer was higher then 5v. Now I'm not an electric engineer but that seemed impossible if it was getting only 5v. So I looked at my power control circuitry and saw I had the regulator in backwards:(. Now it's fixed. That leads me to a trick of the robot trade.

Hardware emulating a Potentiometer at a fixed value:
So...Going along I realised I needed another potentiometer. I use one to calibrate the intensity of my little lcd screen and I use one to drive the servo. I also only have on potentiometer. Knowing it's basically just an adjustable resistor I did a quick google (and I had tons of regular resistors sitting around). I found a guitar audio website where some guy talked about putting two resistors in series and tapping the power output at the meeting point of the resistors. In this case each resistor emulates one leg of resistance in the 3 leg potentiometer(input-output,ouput-ground). And the point at which they meet emulates the 3rd leg(output). Nice trick. Probably would have known that one if were an electrical eng. I positioned the potentiometer where I wanted it then used my multimeter to measure resistance of input to output and output to ground.
I used that trick to pick a fixed value for my lcd screen and used the actual potentiometer to control the servo. When I fixed the voltage regulator issue I also had to recalibrate my potentiometer emulator since I have a smaller input voltage.

Free Running Analog to Digital converter(ADC):
You can configure the ADC to run in Free Running mode. This means the ADC does conversions back to back without waiting for a request. All you have to do to make sure you don't read an invalid result is to read ADCL first with locks the result register then read ADCH and you are good:).

Project Outcome:
Servo control works with the Potentiometer Control:)

Existing Problems:
I think my interrupt handler is taking to long to execute.

Using my handling paradigm for driving the servo I have two timers. Timer0 is throwing an interrupt every 15ms. It causes an interrupt handler to execute which reads the ADC then takes the read result and converts it to a 1.1-1.9ms pulse on pin C4 by setting timer1 with a certain prescaler and top value. I set the pin c4 high then start the timer, wait till it's done, then set pin c4 low. That pin is connected to the hobby rc servo I bought and it responds by moving the a corresponding position. I was also running an infinite loop in main() to print out the ADC value and the top value of timer1. The problem is that the servo is working but I not seeing any prints on the lcd. I figured that my interrupts are taking so long to execute that the lcd print doesn't get to run. I went looking for ways to speed up execution in the datasheet and found the free running ADC trick that works but not quite enough to get me my lcd prints.

If anyone has suggestions on how to improve I would appreciate it:). I've got my code up on pastebin. I should really move the servo control code out of the interrupt handler:). Probably my biggest issue.

MomPC: add video and audio outputs to Combo VCR/DCD/TV

kgw — Mon, 13 Apr 2009 19:12:29 GMT

Hello, my name is Kevin Waite, I am working on a complex project for a stroke patient. One part/phase of this project is to add a 2nd TV flat screen to their environment. Very low cost is a driving factor. The patient has a MAGNAVOX 27MDTR20/17 combo TV/DVD/VCR -- and it does not have video/audio outputs -- this sucks -- because solving this -- conventionally would mean adding a cable turner DVD recorder -- and that leaves out the VCR -- this would greatly increase complexity of the IR computer controlled efforts for MomPC.

Creating (hacking) an audio outputs is trival -- because I can tap the speaker left or right or both or tap the headphone jack..... Audio output is not my concern. . .

The video ouput is. How does one test for a video signal?

The main Tuner / VCR board found inside is a:
BTD850F01012 x4 Triple Main PCB -- the VCR box is found
above this board. I have found this board for sale as:
VHS PLAYER ASSY BTD850F01012 X4 from MAGNAVOX 27MDTR20/17
Part Type: VHS PLAYER ASSY

This board is connected to the DVD player board (or at least this 2nd "DVD" board is found under the DVD player box, see (b) below. (I have photos I can post.) These two boards are bridged (wire connected) via two 19 wired connectors labeled CN2804 CN2803

I can describe in great detail the various connections between the 4 primary circuit boards found within this 27MDTR20.

a) BTD850F01012 x4 Triple Main PCB w/ VCR drive box above
b) BTD930F01031-A DVD / power board with DVD drive box above
c) high voltage TV tube board mounted vert. on the left upper part of the
TV (looking at TV from the backside)
d) small circuit board attached to the business end of the TV tubes cathode.

This TV allows the user to select between AUX, TV, DVD, VCR -- if a DVD is started or if a tape in the VCR is started -- then the scope of TV output jumps automatically to the corresponding play beginning device either DVD or VCR.

Bottom line -- if I could find the TV signal in composite / or / s-video / or / componet form -- at any pathway within the inside of this TV -- that would allow me to hack and get my desired Video output.

There are 8 wire ribbon cables connecting these boards, and there is
a 5 wire ribbon cable going from board (a) to the business end of the cathode of the TV tube. There is a logical flow of 8 wire ribbon connectors -- I mean -- it appears that there are both DVD and VCR 8 wire ribbon connectors going up to the vert. mounted circuit board that includes the high voltage circuits for the picture tube.

Please send me any help you know about to carry out or understand if this is impossible or not. These 8 wire ribbon cables are most likely carrying the video and possibly audio signals -- does anyone have spec sheet for these or any of these boards? If these are carying a video signal what type of video signal would it be?

Thank you in advance.

You can learn more abour MomPC at:
http://forums.trossenrobotics.com/blog.php?b=165
http://forums.trossenrobotics.com/blog.php?b=168
http://forums.trossenrobotics.com/blog.php?b=169
http://forums.trossenrobotics.com/blog.php?b=172

Sincerely,
Kevin Waite

The Mech that Matt Built

MSpurk — Sat, 04 Apr 2009 02:59:00 GMT

I am considering a unique change in leg design. Rather than have a knee that bends, I've been thinking about having two concentric tubes that slide inside each other. The lower tube would be the larger tube probably made of steel to allow me to weld toes to it. The inner tube would most likely be a solid piece of UHMW with two holes drilled in it to allow cables to run through it. So technically it isn't a tube. I'll try to draft it up tonight and post some pics.

-Matt-

The Mech that Matt built

MSpurk — Thu, 26 Mar 2009 02:10:33 GMT

I decided to start a blog outlining the construction of my Mech. I'm on a very limited budget, so hopefully I can utilize intelligent design to achieve a smooth walking gait. Let's see what I come up with.

From what I've seen of some other Mech's looking through the onboard POV (point-of-view) camera will make you sick. I think the key to winning will be to have a bot that has limited bounce and/or sway, and possibly have the bot auto compensate the aiming for any bouncing/swaying that it does encounter. I plan to counteract this by A.) using an Auto-Aiming system and B.) shifting the bot's COM (center-of-mass) without tipping the bot on it's ankles/hips. I will go into details on this later in the development.

When I started on my Mech I intended to utilize a quad for the stability and payload capacity. I've currently put my quad mech to bed and started work on my biped. I just think Bipeds add a level of coolness that I can't pass up. Progress is slow, but I'm nearing completion of the lower right leg. There are some members that are missing from the below image. This is my fit and function test components. I wanted to make certain as the leg moved through its motion there wouldn't be any binding or interference. Edit: I'm having some trouble uploading photos, so I'll have to come back to this at a later date.

That's probably an adequate first post. More to come.

Attached Thumbnails

Inaugural Post :)

RabidCicada — Tue, 24 Mar 2009 19:05:43 GMT

Hey Everybody. First post to my blog. Gonna tune colors etc using this post.

Figure I'll give you all a look at my initial setup. I started with a Nerdkit from Nerdkits.com (it was recommended on some site I visited).

I already feel like I'm outgrowing it. I need a bigger breadboard...

My first goal is to make a hexapod from scratch. That way I know what goes in to the lowest level work. After that I'll be more open to buying modular components.

Initial Workbench :) :

Whoops!...A little product placement in there for Caribuo Coffee!

Initial Setup is using a hitec hs-55 servo along with basic Nerdkit components. I'm using the potentiometer to control the lcd contrast ratio. Using a voltage regulator to drop the 9v down to 5v. Using one switch to control chip programming mode and another switch to control power to the circuit. Also using a 14mhz oscillator.

I plan to calculate the resistance needed for the lcd and slam in resistors so that I can free up my potentiometer :) .

Setup:

Once I get a good working Servo Driver written I will move on to constructing the body of the hexapod with 17 more servos (ack!).

Actuation of "5" Button Test

kgw — Wed, 18 Mar 2009 06:28:05 GMT

Actuation of "5" Button Test

It works!

http://www.kevinwaite.com/MomPC/pic001

The Sanyo 7200 is unharmed, it is seated in its plastic belt holster, snap in and
out -- but for the solenoid rods.... solenoid rods are being held in place with a plastic
collar and small spring (from pen) and eraser as a cork in the top, these (there
will be 5 to 6) have a completely loose push rod -- the pen spring gives a little more
push and also dampens the recoil from when the button pushing actuation is
ended.

These solenoids -- were less than $3 each they barely had the
strength to click the cell phone button -- requires tuning them in terms of being
the exact amount of distance. These solenoids only really have some force to
their push in their last 1/16" of travel.

The aluminum rails are bolted to the holster, 3 1/2 inch 1/8 bolts hold double
Plexiglas window over the cell phone. The solenoids have 3/8 threaded end,
so I tapped the threads for it into the Plexiglas --- may add epoxy later . . . .

The solenoid in this configuration in being trigger by a push test button.

HULC Exoskeleton Video

devilDroid — Tue, 10 Mar 2009 15:51:01 GMT

I need one of these for roller derby. It would be cheating but using technology for evil is kinda my gig...

YouTube - HULC Exoskeleton Promo Vid

Global Brain

devilDroid — Tue, 10 Mar 2009 15:48:44 GMT

Wolfram Alpha, a computational knowledge engine for the web, coming in May...
Twine article
Wolfram blog
Wolfram Alpha

Selecting PhidgetInterfaceKit 0/16/16 USB to 16 digital in/out ports

kgw — Thu, 05 Mar 2009 17:32:59 GMT

I went with:

The PhidgetInterfaceKit 0/16/16 provides:

16 Digital Inputs sensing up to 30VDC
16 Digital Outputs controlling up to 30VDC
LED indicators on all I/O channels

For MomPC -- this is a very good choice -- can drive the solenoids directly -- and also have digital inputs which is very cool -- I plan to install a motion detector -- to monitor when someone comes into her room -- this will allow me to know if she has not been turned for more than 2 hours -- as well as give a light detect acknowledgement of when the cell phone is active -- as well as others..

Thanks again.

btw: I found another very similar device USB digital output interface -- $10 higher cost -- but no inputs -- but slightly higher driver 9amps at 12VDC.
http://www.bibaja.com/cart/index.php...products_id=15

Kevin

Seeking help to pick USB controller for 6 solenoids

kgw — Wed, 04 Mar 2009 07:46:35 GMT

I have done some research about using the RC servos for MomPC, they are very interesting gadgets. However their constant �on� characteristic makes them ill suited from MomPC where she may use her cell phone via MomPC a few times per week. These means a handful cell phone actuated key presses per week. So I began looking at small push type DC solenoids � they seem perfect and are much cheaper a pathway after doing some hunting online.

What I would like to know/learn is what would be the best USB controller for PC to 6 to 8 solenoid programmatic control?

USB controller for 6 solenoids

To define �best� for MomPC criteria the following are considerations:

Powering the pulse actuation of 6 solenoids
12 VDC Coil resistance is 48 ohms and it requires about 250mA to operate

The actual powering of the solenoids via a single MOSFET and/or relay � please offer input on �best� choices to power the solenoids as well.

Simplistic programming control interface � open USB com port XX and send command strings to select which solenoid to actuate and its push duration/pulse.

A pre-assembled controller board would be nice � I can build � but to save time having it come built would be better.

Maybe allow up to 12 solenoids.

Maybe allow a second controller.

Any suggestions in getting MomPC to programmatically control the push actuation of 6 solenoids would be fantastic.

Thanks in advance.

Kevin Waite

It's Alive! It's Alive!!!!!! Mwa-hahahahahahaaaaaaaa!

JadeKnight — Wed, 04 Mar 2009 06:52:54 GMT

The main body mechanics of my J-5 are finally finished! Yay :D!

I hooked up the main torso servos and one arm to the SSC-32 for testing, and everything appears good. The only issue I have is the extremely high pitch whine that seems to come from the 5990TG servos. They are not binding (which I found makes a very different whining sound :p)... They just seem to make this sound when they are holding their position against the weight of the torso/arm.

The bottom torso servo (5990TG) also gets quite hot. I have shocks on the torso, much like the other J-5's around here, but they do not seem to help keep it cool. Looking at it, neither shock seems to lend much compression assistance to the bottom servo, mostly just to the mid-torso servo. The shocks do keep him upright, though, when the power is turned off. Which helps a lot! :)

I also got to test the first pass of my SSC-32 C++ code I wrote a long time ago, and it works quite well (only 1 minor bug, not too shabby for 'blind' coding). Now I just need to get some Axon code up and running. Then I will have the PC talk to the Axon through USB, and the Axon talk to the SSC-32 through one of its UART's, probably just passing along commands sent from the PC. The Axon has some pretty awesome features ;). It will basically become the 'brain-stem' of my J-5, linking senses (sensors), muscles (SSC-32/motors/servos), and other functionality to the brain (Pico-ITX).

Next up on the agenda is servo wire routing, Pico-ITX/Axon/SabreTooth placement, and the prototype of my new head assembly. I'll post up some pics and vids of the finished J-5 once I get the new head built, and all the wires cleaned up.

Green Machine Reloaded, Part 2B

lnxfergy — Mon, 23 Feb 2009 01:10:24 GMT

There's always gonna be some setbacks... such as the robot falls over when it comes to a stop. Or that you mounted the IR sensors at a 45 degree angle rather than 60 degrees. And occasionally, you find out your bot likes to climb the wall.. I really wish a video camera was running right then.

Anyways, I'm quickly changing the lower frame to move the battery backwards 1/2", move the motors up about 1", and put smaller wheels on. I'm also not entirely sold on this TPA-81 thermopile nonsense... its really kinda slow compared to some good old IR photodiodes... or a pyroelectric.

Re: seeking help with a stroke patient // MomPC

kgw — Sat, 21 Feb 2009 20:59:51 GMT

02-14-09 -- intial post

My name is Kevin Waite. I am a new member of chibots. I am looking
for feedback and help on the following design, prototype, and
programming of what I am calling MomPC.

The following is an explanation of the design and prototype of a PC
based controlled interface for a stroke patient, whom happens to be
my mother.

In the most simple form, I need a computer to translate large icon
touchscreen touches into mechanical button presses, one or several
button presses onto a specific TV remote control and onto a specific
cell phone. This would be phase 1.

Phase 2 would be to allow incoming calls to translate into automatic
button presses such as muting the TV. As well as allow the incoming
caller to accomplish things within the patient's room by using touch-tones.
Like triggering the call light or changing the TV channel. Software
that listens and decodes touch-tones is something I am also seeking
and I'm very interested in learning more about. Btw this sort of
thing in found in a home automation box for X10 PRO Telephone
Responder XPPHC06. [[ see later posts below, but the idea to use
an IR blaster for TV control is a vast improvement, (thanks Tony), I
have bought the IR Tira from Home-Electro.com]]

Phase 3 would be to allow data transfers over a cell phone to cell
phone voice call. This could allow retrieving real time status of the
patient and activities of the nursing staff such as keeping with the
patients bed turning schedule.

My preliminary research has identified several possible solutions
both hardware and some software. My goal is to complete the design,
prototype construction and software programming as soon as possible.
With in a month.

I am seeking your help in terms of saving me any foreseeable missteps
and to reduce my trial and error efforts in all ways possible. Thanks
in advance.

Unlike most robotic programming and design efforts what I am seeking
is actually, I think, fairly simplistic. What I need here doesn't
really require much in terms of environmental awareness. Furthermore,
phase 1 comes down to being able to programmatically control the
pressing of possibly six buttons on the cell phone and maybe five
buttons on the TV control.

Reasonable costs is a consideration.

My first thought is to use the types of servos found in the radio
controlled airplanes. Such as the following:

HD-1440A analog servo from Power HD
http://www.pololu.com/catalog/product/1040

I am thinking that all I really need to do, on a per button basis, is
to control a pushrod (push wire) type solution. Were pushing the
button is possibly a 1 mm, 2 mm or 3 mm push movement (via servo
rotation). I suppose a single servo could control two buttons and
what this means is that there would be the zero position and rotating
the 3� (or so) left to push one button and rotating 3� right would
push another button and then returning to dead center would mean
neither button would be pushed. I'm not sure if an analog servo would
be good enough for this, however these are used in RC model airplanes
and if you don't have true positioning than I presume flight control
would be pretty bad. This is why I think that these servos should
work as I am describing. I am very curios to know the pitfalls that
await me in this design. I suppose it might be a good idea to be able
to read back from the servo what position the server is in � but that
likely is not found in this sub $10 servos. So maybe a simple
feedback of positive acknowledgement when the servo is back to it
centered home state. Having this would allow automatic removal of any
usage drift. I am very open to suggestions here, more ideas, other
ideas.

This servo (HD-1440A) and the others I have looked are all using a
Futaba or Futaba-compatible connection. Is this good? I am looking
for a white paper about Futaba connections and a SDK for Futaba
connected and controlled servos.

Using this servo (HD-1440A) from POLOLU leads me to considering
their controller. The Pololu USB 16-servo controller is the most full-
featured member of our line of R/C hobby servo controllers. Seems to
be a lot for the price and seems to be compatible with my
requirements above � I would be off the charts to be get in touch
with anyone with working knowledge with this controller.

http://www.pololu.com/catalog/product/390

The above described quite a bit and these the areas I believe I need
the most help with.

Currently, the following is the 7" touchscreen I have found in
quantities of 2 to 20 and have a bid from its Chinese manufacturer of
$62 per unit (without shipping and misc). I have no previous
knowledge of this manufacturer but I believe I will be buying two or
three of these touchscreens in the near future if anyone out there
would be interested in adding on to my order. This is probably is
outside the size designs for most of you but this is a really good
price. I have drivers for this screen and some with paper .pdfs as
well. Kgw.bot@gmail.com

This touchscreen is the exact same as found inside the: Lilliput
629GL-70NP/C/T retail is about $260 to $300. The LCD panel has the
following model number: AT070TN83.

Thanks in advance.
Sincerely,
Kevin Waite
Kgw.bot@gmail.com

02-04-09 (response to Tony HA (home autmation ideas/suggestions)

RE: HA Systems (HA Home Automation?) � please post good links
to such HA system providing specific solutions the design
requirements outlined in initial post � I'm very curious.

Some more background information.
Using the patient's existing TV and cell phone � has the following advantages:

Very low costs, patient and family familiarity, prefect fit to current
patient needs and deficits.

cell phone (Samsung 7200) is already owned with 2 backup
units bought on eBay � this model has built-in speaker
mode (very loud and clear) and built-in auto answer �
both are prefect for MomPC

cell service is only $10 per month and unlimited
minutes between our plan cell phones

TV is already owned and is a larger screen format with cabinet �
patient needs a larger TV screen � supports CATV viewing of the
nursing home's cable channels, has integrated DVD and VCR players

My limited research thus far for touch screen solutions for stroke patients
are posting with very large retail prices $4k to $10k � and generally are
not addressing these basic environmental self control.

Within this nursing home setting there is currently NO internet access.
any 3g solutions to this problem would be in the $60 per month with data limitations.

Subsequently --- the requirement remains to actually actuate the buttons
on the cell phone and TV remote � however � an IR Blaster solution
for the TV part may be just as good and cut the numbers of servos in
half approximately.

So � that is a great idea � so I am now also very interested in IR Blasters
for the PC � I actually have an installed IR Blaster both on my Media
Center XP PC and an IR Blaster on the SlingBox. I have always had
problems with IR blasters being 100% accurate, with the SlingBox
showing as the most robust example.

Thanks.
Sincerely,
Kevin Waite

02-17-09 (progress report)

Thanks Kevin -- this is some great information and ideas. I
appreciate your feedback. I have no idea what better online
forum with the group also having local meetings I might carry
on my MomPC design brainstorming -- I'm open. I have yet
to check your links -- I will post dinner. Thanks again.

Regarding using an -- I have ordered and will
be developing around the following IR blaster: Tira-2.1 (IR Receiver/
transmitter, USB) http://www.home-electro.com/

It supports receiving and transmitting and appears to have a good
amount of software available. It is costing $49.00 which is about the
going price for this level of blaster. It allows addressing so more
than one of these blasters can used from one PC. It also has a
3.5m jack for an addition emitter connection. Has been around
for several years and has a fair size install base.

The text messaging idea regarding the cell phone is very
interesting -- the Samsung 7200 has text messaging. And
I am somewhat familiar with cell phones being able to act as
a modem for a laptop -- I even had the cable and cell phone
for this in the summer 2000. Cell phones are hard to deal with
in terms of ZERO SDK like documentation on their internal
workings and ports, to prevent hacking but also to prevent folks
from making there own games and ringtones thus killing
a margin rich income stream. For MomPC the key will
be voice calls to mom -- thus the cell phone needs to
function first as an auto-answer auto-speaker phone for
mom -- I am pretty sure that if you connect this phones
by their serial connections vs. the charging connections --
there is a fair likelihood the Samsung 7200 might think its in a
modem mode -- there is a lot of trail and error -- I have
no documentation regarding the 7200 port pins other than
seeing what its charger cares about. So DTMF would
be too shaky -- I would still love to try out some software
that can listen for touch tones via PC mic and see how this
fairs -- any one know of such software or plugin please post
the link. I have used the x10 touch tone decoder with my
cell phone with 100% prefect tone identification -- PRO Telephone
Responder XPPHC06 http://www.homecontrols.com/cgi-
bin/main/co_disp/displ/prrfnbr/2365/sesent/00/X10-PRO-Telephone-
Responder

If hearing and decoding 3 or 4 tones of the 12 would give me
all the hook needed for phase 2 of MomPC.

For outbound cell phone usage I am still seeing a requirement
to physically actuate the cell phone buttons at least 4 to 5 buttons
-- planning to use speed dial so [1] [2] [3] [talk][end] buttons
maybe [ok] maybe [back].

Has anyone worked with the folks at http://www.pololu.com
Their prices and site info appear good.

Has anyone worked with this little servos and their controller?
http://www.pololu.com/catalog/product/1040
and
http://www.pololu.com/catalog/product/390

Thanks.
Kevin Waite

02-21-09 (seeking sheathed flexible push pull cables for small RC cables)

Yo Chiboters,

I'm looking for some form of sheathed cable push pull type cabling,
akin to what you found on the curved and sheathed cables for hand
brakes on a bike. I can always use exactly that � but was wondering
about if there are better types and where I might find them? This is
related to MomPC (see my other posts) with using small RC type servos
to actuate 5 to 6 buttons on my mother's Samsung 7200 cell phone. The
brake cables on the bike are twisted cable wiring � if I were to use
this � are there small crimped on hooks and tackle type hard ware for
this type of wire? Any thoughts?

Please let me know.

Thanks in advance.
Sincerely,
Kevin Waite

More Progress pics

mallster — Sat, 21 Feb 2009 00:38:10 GMT

Have just recieved 2 x Firgelli L12's for the arms. They are a little small compared to the lynxmotion hands but hopefully they will do the job.... Also, still have problems with the pico. It has just stopped working altogether after going well for a number of weeks. The problem may be that it is running straight from 12v nimh which is 14 something volts when fully charged. Specs did say integrated power supply with this model?

Attached Thumbnails

Green Machine Reloaded Part Duex

lnxfergy — Mon, 16 Feb 2009 02:56:13 GMT

I've nearly finished the mechanical & electrical assembly of GMR. I have to redo the front plate of the head as the mounting holes aren't quite correct for the TPA-81. I also have to rig up the UVTron in the back.

So far the navigation software is pretty solid, since its a small revision of the code that won the CNRG in the fall. The room entry code needs some work, as well as the fire fighting itself -- mainly to address the move from a pyroelectric to a TPA-81. I also plan to rework the cornering code, so that the robot makes a smooth turn through the corners, rather than stopping and turning 90 degrees. This was where LGM lost the most time, his run at CNRG would have been good for about a 64 second final score at Trinity. That time would have beat the 2009 senior division winner, but would have been about 3x slower than the winner of the HS division. I'm hoping that with these improvements I can cut the final score to about 20 seconds...

-Fergs

Let there be light...

JadeKnight — Wed, 28 Jan 2009 17:19:37 GMT

Or at least enough to make my cheap webcam work... More on that later...

Work continues on my C++ Robotics Library.

I created some classes to communicate with an SSC-32. I don't actually have an SSC-32 yet, so testing it is pretty much impossible :p.

Last night I integrated OpenCV, to use for the vision. I know about RoboRealm, but I wanted to try OpenCV first...

After getting OpenCV to compile in my project, came the task of testing it with my brand new $15 webcam. The results were disappointing to say the least. It seems like I can either have an unusable dark image @ a nice 30fps, or a nice bright image @ a barely usable 6-7fps (as long as nothing moves, it's usable). I guess the camera's sensor doesn't work well in normal indoor lighting. At the moment, I'm still writing low level code, so the camera performance isn't going to hold anything up.

I made a post on the forums asking about better webcams, so hopefully when I actually begin higher level programming and construction, I can get a better camera.

Where to go from here... I think I'll probably start on multi-threading/distributed tasks. First task will be seeing if I can separate the camera image acquisition into a separate thread, so the main thread is not slowed down to 6 fps. After that, it'll probably either be voice recognition or custom servo animation support (depending on if I get an SSC-32 by then ;))

It begins...

JadeKnight — Wed, 21 Jan 2009 18:17:32 GMT

I started work on my C++ Robotics library last night. The framework is shaping up, and I have a testbed up and running (no actual hardware yet, just an EXE).

As a first step, I quickly integrated the Microsoft Speech SDK. There is now preliminary support for Text-to-Speech. I don't like the way the 'Sam' voice sounds, but the other 2 included voices ('Mike' and 'Mary') are a bit better. I'll finalize the Voice/TTS interface soon, but I think the speech recognition side of it can wait a while.

After that, I'll probably look at setting up a basic SSC-32 interface.

Wall-e Project

mallster — Sun, 18 Jan 2009 13:20:48 GMT

Wall-e Photos

Attached Thumbnails