Laptop-Webcam Robot for the real world

A company I work for is planning on opening a branch location. We will be sending partially constructed welded structures, and a crew on site
will be responsible for finishing them. The structures will be up to 12'x12'x40' in size.
As it is, it is not unusual for the shop guys to ask our design department to come up to help resolve an error or omission. While we do plan on spending more time to do detailed drawings, there is sure to be times when having the designer on site will be helpful.
I am thinking about proposing that the company make (or purchase) a telepresence robot (OK, I know it is not really a robot). I did a little searching on the net, and didn't really see anything that looked viable. Low cost is an important thing if management is to bite, plus making it ourselves would be fun.
At its most basic, just a manual push-cart with a laptop, web cam, and wireless networking could be useful. At its most basic function, the remote shop guy can just push the cart to where they need it, and point the camera at the area in question. They can also use the same cart to view the 3-D models we create of the design. The designer can do some remote desktop sharing to point at areas of interest on both the camera image and the 3-d model display.
But, of course the real telepresence thing comes in if the designer of the structure can move the mobile device around to get a better look. Perhaps also use a laser pointer (mounted to the camera) to point to things on the actual structure as he is talking.
For locomotion, I picture omni-wheels as being the most straight-forward method. This is because I picture the user doing a lot of left-right and forward-backward motions while looking in the same direction with the camera. Another motivation for the omni-wheels is that I can see a secondary use for this robot ; to draw on the shop floor some complex curves we design and annotate it with a little text. A robot that had differential steering would seem to have a hard time doing the maneuvers to draw text.
The problem with the omniwheels, is that the ones I see for sale seem pretty small for my application. This will be on a welding shop floor. There are commonly 1/2" Diameter cables laying on the floor, and the commercially available wheels are certainly unable to transverse those cables. I've also seen a movie of a forklift using omniwheels... so obviously the concept has successfully been scaled up.
But before I start to try to design and build omni-wheels from scratch, I want to consider if I've shot down the differential steering option too quickly. If I make a rotating head for the robot, and make it so that it automatically keeps oriented in one direction while the base rotate, perhaps they will be just fine.
Any comments on what I've discussed so far?
Joe Dunfee
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
snipped-for-privacy@yahoo.com wrote:
They may wireless webcams that don't need a laptop at all. http://www.nextag.com/wireless-webcam/search-html
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
This might be what you need. I don't think the web cam pans up and down which might be a nice feature but it does have an arm. Maybe you could secure the eye to the arm?
http://www.evolution.com/er1 /
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
snipped-for-privacy@yahoo.com wrote:

This sounds like a good interim solution to see if the video link works. No need to spend dozens or hundreds of hours developing a robotic base if the video turns out to not be as useful as you'd like. Just place a camera with an pan/tilt head on a dolly, and have a person move the dolly as needed. The person controling the camera can operate the pan/tilt head to change the immediate view.

Or the designer can just talk on the phone and have someone on the site to do it.

Well, there are omni wheels of up to 4.7" from Omniwheel.com, which might be large enough. Still, I think you'll want to experiment using these on uneven surfaces, however. Given cables and other crud on the floor a tracked vehicle might be better. You can always provide additional degrees of freedom by using a pan/tilt turret for the camera.
-- Gordon
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
: Or the designer can just talk on the phone and have someone on the site : to do it.
I would guess that the 'fun' part of his original question has a lot to do with it. This could be a spirit-building exercise if people are into it. It could also be a great PR tool -- get you on the local news, This Old House, etc.
-Chris
--
==========================================================
Chris Candreva -- snipped-for-privacy@westnet.com -- (914) 967-7816
  Click to see the full signature.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Christopher X. Candreva wrote:

True, but in my experience remote video isn't very good for inspection purposes unless the setup is well designed for it (lighting, angle, high quality camera, etc.). Building teleoperated base may be fun, but management (the "Big M") tends to get upset when things don't work out as planned after money has been spent, and a simpler method would have proved out the plan. Always good to take things one step at a time.
Should a teleoperated video robot not work out, eventually some bean counter will ask why someone just didn't take some quickie digital pictures and e-mail them. PR is fine, but looking good to the boss is even better!
-- Gordon
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Don't know if i already posted this. This is a pretty sweet robot. I don't have one but I've read great things about it. I don't think you can pivot the web cam up or down but there is an arm, so you you may be able to attatch the eye to the arm to give you a wider range of vision.
http://www.evolution.com/er1 /
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Hi,
For the camera controller, have a look at my site I have a AVR that accepts commands from the network using UDP to drive several servos,LED's and a VFD.
--

Regards
Ian Dobson
  Click to see the full signature.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
snipped-for-privacy@yahoo.com wrote:

See my sig for one possible implementation.
I do think your cart idea is a good place to start, though. It can at least give you a feel for how useful telepresence will actually be.
Hope that helps - m
--
(Replies: cleanse my address of the Mark of the Beast!)

Teleoperate a roving mobile robot from the web:
  Click to see the full signature.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On 6 Jun 2005 22:40:22 -0700, snipped-for-privacy@yahoo.com wrote:

I'd get a camcorder with a flip out screen, built in light, and an analog video output. Camcorders have very good optics, zoom, and focus capabilitys (up to touching the lens for small objects). Just start with using a camera tripod for supporting the cam. You can send the video back to a computer with a capture card/device over a long piece of telephone wire (video and power on the four conductors. Talk to those on the other end via telephone (cordless or cell unit with a headset) and they can tell you where to point the cam. Use a free or commercial webcam program to send the video fames to those on the other end. Check out the basic video setup before spending $$$ on the robotic part.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Thanks everyone for the replies. I am replying to my own message so I can consolidate my response to everyone.
The MR1 was one I was considering for myself to play with, since I have a spare laptop. But, for this application it is just too short. It would require significant modification (add height, wider wheelbase, larger wheels, etc.).
Gordon McComb said;

The overall diameter of the wheel assembly is not the limitation. Their 4.7" diameter wheels have only about 1.75" diameter at the small wheel. The fork-lift version I mentioned before had a different configuration for the smaller wheels. The result was a larger diameter for the small wheels, and the need for only one wheel, instead of the pair the ones on Omniwheel.com require. Here is the link if you want to see; http://www.airtrax.com/ I've not seen these kinds of omniwheels available.
Today I spent more time thinking about dimensions and features that I would want. There is no question that I would want the keyboard and monitor at the remote site to be comfortable for a person standing to operate. This is because I want there to be 2-way sharing of the display so they can bring up and view 3-d models and drawings.
I also want the camera to be the viewpoint of a person standing, or or siting in a chair, because these are easy positions to perceive your view orientation.
I did find one commercial base, which I think would meet my requirements and allow me to build a sturdy elevated top shelf for the laptop and camera. I don't have a price back from there yet, but I suspect sticker shock will rule it out. http://www.aai.ca/robots/labo3.html
Regarding the video; Si Ballenger wrote;

I know Gordon said similar things about video quality. In our case, I think we will be in excellent lighting, so that is not an issue. But, I would think that 640x480 resolution would be minimum. The company will be setting up a very high speed internet connection, and I suspect will be able to get data pipelined so that standard internet delays would not be present. Thanks for those of you who have a web-controlled robot... it really showed the need to keep the update fast and as close to real-time as possible..
As I continue to think about it, the perception of the objects you are viewing through the camera is the biggest issue. I draw these structures in 3-d and have a good display with real-time rotation of the shaded object. But it is still easy to mis an interference between two parts, which is very plainly obvious on the shop floor. Even a photo-realistic rendering of the object isn't nearly as powerful as an actual walk around the physical object. This is my particular motivation for the robot base. However, maybe I can settle for a compromise....
It may be that the "prairie dog" motion of a viewer is more important than stereo vision. By this I am referring to the left-right head bobbing that a prairie dog does when he stands on his hid legs to observe something. If I don't move the base by remote control, perhaps I can move [translate] the camera left and right, in addition to the standard pan and tilt. In the past I tried out a program "cam-to-pan" which uses a normal computer camera looking at the operator. As you move your head a bit left/right or up/down, it moves the mouse. Perhaps I can take that same data, and move the camera left and right.
I guess I will have to try it out to see how well it works.
At the moment, I am leaning away from even suggesting a motion base. I will let you all know of any results if we try it out.
Joe Dunfee
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On 7 Jun 2005 21:39:00 -0700, snipped-for-privacy@yahoo.com wrote:

You may want to look at the Axis brand of cams. They are more aligned for your type of commercial application, but they will cost more too.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

Polytechforum.com is a website by engineers for engineers. It is not affiliated with any of manufacturers or vendors discussed here. All logos and trade names are the property of their respective owners.