[Dprglist] Seeing the RoboColumbus cones

[David Anderson]

John,

But I think you _are_ making your point.  That's why I suggested it 
might be useful to examine the traffic cones in the UV spectrum. Maybe 
the plastic itself has a unique signature, irrespective of the color.  
Or maybe not.  Would also need a UV sensitive sensor or camera, which 
might not be trivial, though the astronomers seem to do it all the 
time.  Someone needs to investigate...

My first pass at finding the cones a few years ago just used the robot's 
sonar to attempt to identify the proper shape, like SR04 does in the can 
collecting contest.  In that contest, the SR04 robot does not care what 
color the cans are.  I actually thought it was a bit of a cheat when the 
club decided to paint them all orange in order to facilitate a certain 
type of sensor.  They aren't "soda cans" any more; they are fluorescent 
orange cylinders.

In a previous enlightenment SR04 was able to locate the cans with it's 
sonar, and then sort them according to color (Coke, Sprite, Root Beer, 
Orange Soda) using a bi-color LED and CDS cell on the ends of the 
grippers.  All wasted effort with the "modified" cans. :)

But the cones turned out to be devilishly hard to identify with the 
sonar unless you were already looking directly at it.   The shape tends 
to send the sonar reflections off in different directions otherwise.  As 
with lots of robot techniques, things that look good on paper or around 
the water cooler often don't work in a practical application.

Now the Mars rovers had a tiny mass spectrometer on the end of the robot 
arm.   There's a nifty solution!  Except for the cost, which is shall we 
say above my pay grade.  And that also requires that the sample be 
pre-located and have close proximity.   Not so useful for our 
applications.  Still, a boy can dream.

onward!
dpa










On 12/06/2016 12:20 AM, David Anderson wrote:
> Flame cultivator.  Awesome.  I googled "cotton farm flame cultivator" 
> and came up with this:
>
> https://flameengineering.com/pages/agricultural-flaming-guide
>
> Invented by one Price McLemore in 1938.   Clever.  I want to see one 
> in action!
>
> cheers
> dpa
>
>
> On 12/05/2016 11:28 PM, John Swindle wrote:
>> Dave,
>>
>> Great stuff. Thanks for setting me straight on the already-uniqueness 
>> of the orange cones.
>>
>> But I'm still not making my point (largely because I'm still slowly 
>> formulating it). At the same time, I get your point, which is: Go and 
>> do it! But vision-related stuff is just not interesting for me to pursue.
>>
>> More of my point: It's not to look for, say, an orange cone, but to 
>> look for, say, a plastic thing or a rubber thing or a wood (or 
>> non-wood) thing. Unique in the arena and likely unique in a large 
>> unconstrained space.
>>
>> My interest in this is to build a robot that weeds the yard. I 
>> suspect that it is insufficient to look at leaf shapes, and I suspect 
>> that different plants have other signatures that will be easier to 
>> detect, as with the Earth sciences satellites that see what kind of 
>> trees are in a forest.
>>
>> When I saw a root heater weed killer for sale (basically a soldering 
>> iron on a stick), I got interested in having a gadget wander about 
>> the yard, selectively cooking the roots of undesired plants. A 
>> vinegar spray works also, but the root heater is more precise. (A 
>> flame cultivator is more dramatic though. We had those on the cotton 
>> farm.)
>>
>> You say we often build stuff that is purpose-built to win an 
>> arbitrary contest. In this case, I am looking for something that 
>> serves as a sensor for something beyond color and shape differences. 
>> It would be great if it involved sound! Talk to the weeds.
>>
>> I thought we developed sight for visible light because that's what 
>> gets through the atmosphere. Water has a big impact on the atmosphere 
>> as well.
>>
>> And so a UV sensor would be OK for a robot because it would either be 
>> designed to not get burned up by UV, or we wouldn't care that the 
>> sensor only lasted a few years. We try to anthropomorphize things too 
>> much, building our own limitations into our creations.
>>
>> Good stuff. Back to slowly formulating now.
>>
>> John Swindle
>>
>
>
>
> _______________________________________________
> DPRGlist mailing list
> DPRGlist at lists.dprg.org
> http://lists.dprg.org/listinfo.cgi/dprglist-dprg.org

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.dprg.org/pipermail/dprglist-dprg.org/attachments/20161206/12fcdb82/attachment.htm>