[Dprglist] Line Following Challenge Simulation

[Doug Paradis]

John,
    Interestingly the two committed competitors for the Challenge Line
Following competition this May, Carl and Ron, have both demonstrated
simulations. You may have missed Ron's simulation last meeting. I agree
that simulation is a strong tool for approaching problems, maybe we can get
Carl, Ron, and you to give a presentation this year on simulation to help
others get started. I believe you use Matlab as your primary tool., Carl is
using Javascript Node and Roborealm. Ron I believe uses some type of OpenCV
simulator.

    I hope you throw your hat into the ring and give Carl and Ron a run for
their money.

Regards,
Doug P.

On Mon, Jan 8, 2018 at 4:05 AM, John Swindle <swindle at compuserve.com> wrote:

> The Line Following Challenge is a significant contest. The combination of
> problems is realistic of an asphalt roadway that has been repaired, like
> driving 75 MPH posted speed limit at night from Buna to Mauriceville, TX on
> two-lane with no shoulder. The combination of 1) wide gaps, 2) S-turn gaps,
> 3) color changes, 4) intersections, 5) sharp angle turns, 6) noise from the
> tiles (edges and sparkles), 7) randomly and suddenly changing lane widths
> (a penchant of civil engineers) and 8) having these hazards close to one
> another (another penchant of civil engineers to do stupid things like
> closing a traffic lane in a curve) makes the contest interesting.
>
> I thought a two-step test would solve the problem, but I am not sure. I
> have tried to make a solution that is stateless, not caring what it has
> seen before, but the intersections may have thwarted that goal. The acute
> angle turn interferes with the intersection avoidance. The stains were easy
> to ignore (though requiring two passes, for the white tiles and the black
> tiles), but the stains are quite realistic.
>
> I like Carl's idea of showing a simulation of the control program. I get
> the impression that few of us do simulations. I always do simulations.
> Maybe my terminology doesn't sound right to you. My simulations are given a
> starting condition and then provide simulated loop data to the next
> iteration of the simulation. I bet many of us do this. If you don't, you're
> missing out on a great opportunity to prove your algorithims.
>
> The thing that many of you do that I have not yet done is datalogging or
> tracing the run. I find that to be a debugging tool, and I prefer to be on
> the creation side of the development instead of being on the debugging
> side. But, when using a development tool that has debugging built into it,
> the simulation is naturally debuggable.
>
> Good stuff. Hoped I'd have a simple solution that could be hosted on
> someone's robot, but the S-turn gaps are a problem for my edge-detection.
>
> It's a reasonable problem.
>
> Best to y'all,
> 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/20180108/9d22be6a/attachment.html>