ThereminVision sensors

Just learnt about ThereminVision sensors (http://thereminvision.com ) in this news group. Very interesting and cheap solution compared with IR and Sonars.
Does anybody have any experiance with using this sensor and could compare it and others?
Also would be interesting to get feedback from OOPIC and OOBOT developers how to connect this nice sensor to the very nice boards (see more info http://groups.yahoo.com/group/thereminvision )
Thanks. Ek.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Theremin's sensing idea seems to be very simple but very interesting. I thin it's applicability to "real" robotics/engineering projects are limited because it has the following contraints; * Very limited sensing range (10-30cm) * Sensor size ; sensing range is a function of sensing antenna * Environment introduce sensing inaccuracy * . . .
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
xyz wrote:

Actually, for obstacle detection (think non-mechanical bumper) and avoidance range and accuracy constraints aren't much of an issue. Where more accuracy or range is required, obviously other sensors would need to be used.
In particular, the design of the unit should make for good, inexpensive coverage for robots with a large area. Those of you that have built larger robots know what an expense/hassle can be incurred getting adequate basic obstacle detection for a larger robot. Large mechanical bumpers can get to be a hassle to construct, and other non-contact methods can get to be expensive.
My concern was that different material types might cause odd, missing or false readings.
--
(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

You might want to go and check out the Yahoo Group "FrontRangeRobotics", they more or less turned me on to the TherimininVision sensors a while back. I beleive that several of those guys have used the sensors. The sensors are good for close in work, and work well on oddball things like table legs, couches (where the robot could get stuck under the couch), drapes or curtains, dogs and cats (if it is even possible to get a dog or a cat to be that close :) ). If you have a lot of IR interference, daylight, etc, the sensors work well as they aren't affected by IR interference. I've used the Freescale Semiconductor (Motorola MC33794) version of this sensor, but it only works on people sitting in car seats, thus it isn't very sensitive like the therimin sensor is. I am thinking about using the Therimin sensor on a sumo bot for closer in work, as the IR and sonar sensors on the two competing bots tend to interfere with each other, and some guys can have their sensors jam yours out.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Ek wrote:

It's basically a "capacitance operated relay" circuit.
If you want a commercial version, get the DEI 508D car alarm sensor. Same principle. This triggers when someone gets close to the sensor. Popular with convertable owners. About $38 each.
Simple devices of this class require too much manual adjustment.
            John Nagle
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

this
it
Ek,
I have used the ThereminVision sensor. It is an incredibly simple design. It comes with 4 "sensors" each requiring an antenna or plate of some kind. When I built it, I did not have any antenna handy, so I just used 4 lengths of a fairly stiff copper wire (about 6 inches each). I actually attached the thing to a Parallax mini-sumo and was able to use the basic-stamp to operate it. I also wrote a quick PC app that attaches to the BS-II via the serial cable and graphically shows 360 degree direction and range to a "target" within the detection range. With 3 or 4 sensors it's easy to triangulate direction to a single target. Multiple targets are difficult or impossible to triangulate, but you still get a solid "proximity to each sensor" type of reading. My experiments show about a 6-10 inch detection range (with 6 inch antenna). One strange thing is that the sensors and "processor" are laid out on a nice green PCB, but the connections are scattered all over each board making it impossible to use some nice connectors. All in all it's very cool and works great. Very sensitive (in it's detection range) and WAS very cheap, I just checked and it looks like the price has gone up. I have robot that is about 4 feet tall, and I'll be adding the theremin to it to aid in obstacle avoidance. I currently have plexy panels on each side, but I'm planning to replace them with electrically isolated metal panels (or maybe just cover the existing panels with foil) and use them as large theremin sensor plates.
Hope that helps.
--TE
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

compare
lengths
the
or
(in
be
panels
I doubt that this thread can be ended by TE :-)
Seriously, could you tell if it is possible to cover (see image on http://oap.sourceforge.net/prototype.php ) black plastic risers, which divide two chassis's boards, with foil to make antennas for the sensor? Or use metal bars inserted in the risers as antennas? I guess not. What would be right solution?
Ek
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
I like to use brass tubing from the local RC Hobby Shop myself. of course copper tubing works too. The brass flat strips would probably work good as you can mount them to the sides of a plastic cover or case.

Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Last question to be sure that I understand it clear. -In case of attaching brass to the plastic, will not plastic add noise to the sensor?
-Will it work if the plastic support pipes are replaced with brass ones?

in
attached
to
difficult
detection
I'll
have
make
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

Ek,
The "about 4 foot tall" robot I mentioned in my earlier post IS an OAP bot made from the Zagoros base. I just did not cut it down to be short. I made my own controller boards so I could add some functionality (like Theremin, balancing, etc) not included in the official OAP boards. I just measured, and from top of the eyes to floor is about 3 feet, not 4. Anyway, I cut a piece of right angle aluminum to make brackets so I could put 4 plexy sides on the bottom and 4 on top. Each plexy panel is about 12 inches square. I've already painted the PVC and "base" panels fun colors (red and yellow) and added some decorative corrugated tubing around the PVC supports. I'm planning to try to use the thread rods inside the PVC for the 4 antenna, but if the sensitivity is too poor, I will replace the lower and/or upper plexy panels with aluminum or maybe just cover the existing panels with foil. I've just been so busy. My bot has all of the parts that you can just buy, including the theremin. I have completed a power board and motor controller board. I can remotely drive it from a PC application via the 802.11 network and see the images from the dual cams on the screen. I do not have the head control or sonar control modules down yet, nor do I have the theremin interface yet. One of the features I added to the power board is a USB interface to the VIA motherboard. The power board is then the "master" I2C for all of the other slave boards like the motor controller. It works really well, however, in my next rev I'm gonna put the USB <-> I2C bridge on it's own module and remove it from the power board. In fact the only reason I'm going to redo the power board for a 2nd rev at all is because there is a slight problem with the motor driver from Magnevation. Specifically, the LMD18200 H-bridges have an undervoltage lockout between 9 and 11 volts where the outputs just shut off. My experience with the ones I have is that they just stop working when the battery drops to about 11 volts on one side and 10.9 volts on the other side. Everything else works just fine, but the bot suddenly starts turning in circles then halts altogether and refuses to move. So in my next power board rev, I'm going to use two 12V batteries and switch them in series to drive the motors, and back in isolated parallel for charging at the base. I'll probably re-do the motor controller and just incorporate the H-bridges and skip the magnevation board altogether. I may also incorporate the theremin circuit on a board with the I2C slave controller so I can run all 4 sensors at once and do some pre-processing (math) before sending the data to the VIA.
I also want to improve the head. It's just too flimsy. It seems ok when rolling around on the floor, but when riding around in the passenger seat of my car, it flops around and vibrates mercilessly. I've already had to replace the tilt servo once due to stripped gears from a bump in the road!
Also the suggested sonar layout for the OAP bot is great until you try to put side panels on the thing. Then the 45's are partially located "behind" the panels and get lousy performance. I have relocated them to be radially centered on the center of the platform instead of the center of the drive wheels. A side benefit is that all of the side panels are now identical and interchangeable. Changes, changes . . . I could go on all day. Nothing is ever perfect!
Meanwhile, I'm working on my version of a Robo-One robot and a few other projects as well. So many projects . . . so little time. Oh, and BTW my OAP bot is set up to dual boot Win/Linux. When I initially got my 802.11g wireless adapter there was no linux driver available and I had to install win. As they say "Linux and Open Source are only free if your time has no value.". (to all: don't bother to flame back at me, I've heard it all before: "win sux lin is the best" blah, blah, blah. I just needed it to work NOW and did not have time to write my own driver or go begging the open source community to do it for me. Bottom line: it works great on both platforms and I have more tools for the win platform so: win first, lin follows after.)
--TE
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Thanks all, personal gratitude to Earl and Kap'n!
Ender, please do not forget to post your results and progress on using this sensor. I hope to put together chassis and PCBs by that time and start using sensors. Would you also send a link to snapshots of your robot(s)?
Ek

probably
to
made
sides
but
plexy
buy,
controller
network
head
I2C
on
reason
a
where
they
bot
and
for
may
of
"behind"
radially
and
is
open
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

sensor.
Yeah, I'm spending all my spare time building things and that don't leave much time for documenting them. I do need to slow down and post some pics. It's on my list...
--TE
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
You can goto www.dprg.org and see a lot of photos of some of my robots and others.

Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Thanks, I love this WEB site.
Ek

on
foil
not.
cut
square.
I'm
antenna,
bridge
the
batteries
parallel
just
pre-processing
when
seat
to
drive
identical
Nothing
other
my
install
to
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Thread Ender wrote:

That's very helpful for me -- I'd like to use the system for obstacle detection -- basically a non-contact "bumper". Did you find that any particular materials failed to show up? Also, does the detection range vary much by marterial type?
Thanx -- 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
in message

Kap'n,
Yes. Somewhere on the theremin site (or maybe in the PDF user manual) there is a list of materials and their relative properties. In practice I found that you can readily detect something large like a chair, but not a tablecloth hanging down from the edge of the table. Curtains are not too bad, but that's probably because there is piece of wall or window sill behind it. Also small protrusions are missed: My 6 year old left a "fishing pole" dowel sticking out from the stair rail about 2 feet above the floor and my bot never saw it and got hung up on it. All in all it's still much better than trying to figure out how to make a "whole body bump sensor". I'm still looking for that cheap "electronic skin" that gives me touch sensitivity over the entire surface.
--TE
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.