[Dprglist] Fwd: LF intersection angle rule
David Anderson
davida at smu.edu
Sun Dec 10 14:30:26 PST 2017
Carl,
You know old people repeat themselves ;)
The SR04 robot can do line-following of the simpler variety. It is set
up so the line width doesn't matter (as long as it is not wider than the
robot) and the sensors are "shown" what the line looks like and what the
background looks like before it sets off. That way it can run with any
convenient lines that might be found, i.e., stripes in a parking lot,
safety lines in a shop, tar lines in the asphalt, lines on a basketball
or tennis court, etc. When DPRG held its meetings at the Science Place
in Fair Park, there were inlaid lines of the solar system orbits in the
floor of the lobby. You no doubt remember the video I shared of SR04
following the ecliptic around the room at the Science Place as an example.
So I guess I have a bias towards behaviors that are not tuned to a
specialized environment. Now if one rotated the challenge course up on
it's edge and made it 3D, the robot would then be perimeter-following
--- itself a challenge up from simple wall-following --- rather than
line-following. There are lots of environments where that turns out to
be really useful.
You may remember the video I shared of nBot following the perimeter of a
complex environment in the basement of the Heroy building at SMU as an
example.
A more significant example is the video of jBot using conditional
wall-following to escape from dead-end cul de sacs while navigating the
TI campus in Denton, available on jBot's webpage.
Of course if the challenge-level line following solutions turn out to be
applicable in some nifty way when the robot runs in the real world away
from the contest course, then I clearly will have been wrong!!
On that note, I'd encourage you as you get physical to think about what
is required to run on less-than-optimal non-contest surfaces (wheel size
and traction, torque, ground clearance, etc), and build that in from the
beginning. Just my $.02.
The work that you and Ron are doing is first rate and very much food for
thought --- obviously! At this point I haven't been able to do any
roboting at all for about a year, so I'm content to let you guys push
the envelope while I just yell from the sidelines.
onward!
dpa
On 12/10/2017 12:51 PM, Carl Ott wrote:
> David,
>
> you've shared those videos before - fun for speed even if the line is
> rather boring :)...
>
> you mentioned no plans to field a robot - why not?
>
> e.g. You could field the first line-following bipedal walking robot in
> the club, and give Ron and I a 'run' for the money...
>
> Carl
>
> On Sun, Dec 10, 2017 at 11:51 AM, David Anderson <davida at smu.edu
> <mailto:davida at smu.edu>> wrote:
>
> I did see Ron's robot demo, and interesting contrast between his
> approach and Carl's, though Carl's is admittedly a simulation at
> this point.
>
> Carl says that this year he is going to "get physical" so I was
> about to post a link to Olivia Newton-John's "Let's Get Physical"
> as a theme song. But then I went back and read the lyrics, and
> all the thinly and not-so-thinly veiled innuendo from 1980's disco
> seem oddly discordant in today's cultural environment ("I've been
> patient, I've been good. Trying to keep my hands on the table").
>
> So instead I offer the following, in keeping with the
> line-following theme and the observation that all robot contests
> eventually devolve into races. Even if the requirements are,
> say, that only line-following WALKING robots are allowed:
>
> <https://www.youtube.com/watch?v=4XiRxNkZleY>
> <https://www.youtube.com/watch?v=4XiRxNkZleY>
> https://www.youtube.com/watch?v=4XiRxNkZleY
> <https://www.youtube.com/watch?v=4XiRxNkZleY>
>
> cheers!
> dpa
>
>
>
> On 12/10/2017 11:32 AM, Doug Paradis wrote:
>> David,
>> From watching actual robots run the stain elements, I know
>> that it very possible for the robot to follow the outside edge of
>> the stain and recover the line on the opposite side (my robot did
>> this for one). Following this path the robot would be considered
>> not following the line. Again after watching multiple robots
>> navigate the S curves, it is easy to see if the robot is
>> following the line (I.e., swaying with the S curve) versus just
>> cutting across the center line (i.e., not following the line). I
>> agree that bigger robots, those with wheel bases larger than the
>> maximum radius of curve used in the course (i.e., 6 inches),
>> might not be as responsive as the smaller robots. However, I
>> still believe that you can tell when they are not following the line.
>>
>> BTW, did you see Ron's robot demo on Saturday? His robot was
>> doing pretty well. I think the course he was running had a S
>> curve which he handled without issue. The practice course didn't
>> have any stains, if I recall right.
>>
>> Regards,
>> Doug P,
>>
>> On Sun, Dec 10, 2017 at 10:49 AM, David Anderson <davida at smu.edu
>> <mailto:davida at smu.edu>> wrote:
>>
>> Thanks Doug,
>>
>> I'm not sure I follow your examples (pun intended!).
>>
>> For the segmented S curves, the difference between following
>> the arc and following straight line approximations of that
>> arc are pretty small, and likely to be indistinguishable from
>> Ron's robot's path, for example. Similarly, each of the
>> stains on the published "challenge" course is on a straight
>> line segment between the entry point and exit point, so the
>> paths would be identical in both cases.
>>
>> For a robot with a base the size of the tiles, I'm not sure
>> there would be any difference at all, unless the rules
>> require that the "exact center" of the robot must remain on
>> the line, in which case I don't see how either Carl's
>> approach or Ron's would qualify.
>>
>> This is just an intellectual exercise, I'm not planning on
>> fielding such a 'bot.
>>
>> cheers!
>> dpa
>>
>>
>>
>> On 12/10/2017 10:06 AM, Doug Paradis wrote:
>>> David,
>>> Going from entry point to exit point is not considered
>>> enough. You have to follow the line. For example, the
>>> segmented S curves, the robot needs to follow the curve and
>>> make an S type movement. Another example is the stain
>>> elements, the robot needs to follow the line across the
>>> stain not follow the outside edge of the stain.
>>>
>>> Regards,
>>> Doug P.
>>>
>>> On Sun, Dec 10, 2017 at 1:11 AM, David Anderson
>>> <davida at smu.edu <mailto:davida at smu.edu>> wrote:
>>>
>>> Great meeting today.
>>>
>>> Doug, I have a question based on Ron and Carl's
>>> presentations, perhaps it has already been answered, to wit:
>>>
>>> As I understand, the line following course consists of
>>> an assembly of square tiles, each of which has an entry
>>> and exit point.
>>>
>>> Is it sufficient for the robot to just identify the
>>> entry and exit points, and drive directly from one to
>>> the other?
>>>
>>> That is, it would not follow the line per se, though
>>> with sufficiently large robot the differences would
>>> probably be pretty negligible, but it would follow the
>>> course.
>>>
>>> thanks,
>>> dpa
>>>
>>>
>>>
>>> On 12/09/2017 08:18 AM, Doug Paradis wrote:
>>>> John,
>>>> Currently, the intersecting lines are always
>>>> straight and all intersections are 90 degrees. Curves
>>>> lines at an intersection would be a possible addition
>>>> to further challenge courses (interesting idea). The
>>>> link to the course layout is at
>>>> https://www.dprg.org/wp-content/uploads/2017/11/DPRG-Roborama-2011b-Challenge-Level-LF-Course.pdf
>>>> <https://www.dprg.org/wp-content/uploads/2017/11/DPRG-Roborama-2011b-Challenge-Level-LF-Course.pdf>
>>>> If you have any additional questions, let me know. I
>>>> would love to see your work on the tests that you are
>>>> developing.
>>>>
>>>> Regards,
>>>> Doug P.
>>>>
>>>> On Sat, Dec 9, 2017 at 12:36 AM, John Swindle
>>>> <swindle at compuserve.com
>>>> <mailto:swindle at compuserve.com>> wrote:
>>>>
>>>> Doug,
>>>>
>>>> I'd say the crossing in your drawing was about
>>>> either 30 degrees or 150 degrees, and since it
>>>> could come in on either side, I'd have to look for
>>>> both. I am not asking that the rules state which
>>>> side the angle is measured from. My issue is that I
>>>> am developing a two-step test that covers all the
>>>> conditions in the Challenge, but in the steps I
>>>> need to include something that rejects the
>>>> intersections. I am OK with rejecting any line that
>>>> is 70 degrees to 110 degrees on either side. If the
>>>> intersecting line is straight, the test is a bit
>>>> more robust. If each side is 70 to 110 degrees (a
>>>> bent intersecting line), my two-step test might fail.
>>>>
>>>> Doing it "that's just wrong" way,
>>>> John Swindle
>>>>
>>>>
>>>> -----Original Message-----
>>>> From: Doug Paradis <paradug at gmail.com
>>>> <mailto:paradug at gmail.com>>
>>>> To: John Swindle <swindle at compuserve.com
>>>> <mailto:swindle at compuserve.com>>
>>>> Cc: DPRG <dprglist at lists.dprg.org
>>>> <mailto:dprglist at lists.dprg.org>>
>>>> Sent: Fri, Dec 8, 2017 11:52 pm
>>>> Subject: Re: LF intersection angle rule
>>>>
>>>> John,
>>>> In the challenge course, all the intersections
>>>> are 90 degrees. The rule was written to allow
>>>> crossing variations in the future. I'm thinking
>>>> that 70-90 degrees would represent the smallest
>>>> angle of the intersection. I not sure that is
>>>> right, just the way I would interpret the angle.
>>>> if you saw an intersection that was like this:
>>>> | /
>>>> | /
>>>> /
>>>> / |
>>>> / |
>>>> / |
>>>>
>>>> What angle would you say the intersection was?
>>>>
>>>> Regards,
>>>> Doug P.
>>>>
>>>> On Fri, Dec 8, 2017 at 9:03 PM, John Swindle
>>>> <swindle at compuserve.com
>>>> <mailto:swindle at compuserve.com>> wrote:
>>>>
>>>> Doug,
>>>> The Line Following Challenge rules say
>>>> "Intersections may cross with angles of 70 - 90
>>>> degrees." Doesn't that really mean 70 to 110
>>>> degrees? Is the intersecting line straight, or
>>>> can it bend at the intersection?
>>>> Thanks,
>>>> John Swindle
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> _______________________________________________
>>>> DPRGlist mailing list
>>>> DPRGlist at lists.dprg.org <mailto:DPRGlist at lists.dprg.org>
>>>> http://lists.dprg.org/listinfo.cgi/dprglist-dprg.org
>>>> <http://lists.dprg.org/listinfo.cgi/dprglist-dprg.org>
>>> _______________________________________________ DPRGlist
>>> mailing list DPRGlist at lists.dprg.org
>>> <mailto:DPRGlist at lists.dprg.org>
>>> http://lists.dprg.org/listinfo.cgi/dprglist-dprg.org
>>> <http://lists.dprg.org/listinfo.cgi/dprglist-dprg.org>
>>>
>>> _______________________________________________
>>> DPRGlist mailing list
>>> DPRGlist at lists.dprg.org <mailto:DPRGlist at lists.dprg.org>
>>> http://lists.dprg.org/listinfo.cgi/dprglist-dprg.org
>>> <http://lists.dprg.org/listinfo.cgi/dprglist-dprg.org>
>> _______________________________________________ DPRGlist
>> mailing list DPRGlist at lists.dprg.org
>> <mailto:DPRGlist at lists.dprg.org>
>> http://lists.dprg.org/listinfo.cgi/dprglist-dprg.org
>> <http://lists.dprg.org/listinfo.cgi/dprglist-dprg.org>
>>
> _______________________________________________ DPRGlist mailing
> list DPRGlist at lists.dprg.org <mailto:DPRGlist at lists.dprg.org>
> http://lists.dprg.org/listinfo.cgi/dprglist-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/20171210/eb40ed7e/attachment-0001.html>
More information about the DPRGlist
mailing list