[evaal contest] EvAAL - on-site tracks - outdoor markers and score computation

[Joaquín Torres Sospedra]

Please find my opinion about Francesco’s questions inline below.


-- 
Joaquín Torres-Sospedra
Postdoctoral researcher at GEOTEC Research Group <http://www.geotec.uji.es/>
Office UB1519SD @ Espaitec2 Building <http://smart.uji.es/index.php?room=UB1519SD>
Institute of New Imaging Technologies <http://www.init.uji.es/>, Universitat Jaume I <http://ujiapps.uji.es/>, Castellón, Spain
Social media: @jtorres_phd <https://twitter.com/jtorres_phd> - LinkedIn <http://es.linkedin.com/pub/joaqu=C3=ADn-torres-sospedra/55/87b/66a%20/en/> - ResearchGate <http://www.researchgate.net/profile/Joaquin_Torres-Sospedra>
Web: http://www3.uji.es/~jtorres <http://www3.uji.es/~jtorres>
> El 28/9/2015, a las 15:05, Francesco Potortì <Potorti at isti.cnr.it> escribió:
> 
> FLOOR PENALTY
> 
> We use (see <http://evaal.aaloa.org/current-competition/ta2015track1-2>)
> a floor penalty to compute the error on a marker for which there is a
> wrong loor estimation.  I suggest the penalty to be between 10 and 20
> meters and I am soliciting opinions on the exact number we should choose
> for the penalty P.

I suggest a penalty of 15 meters for Tracks 1&2.
When a low value is selected, the penalty does not have any significant effect. 
On the other side, high penalty values tend to penalize accurate systems in excess. 

> MAXIMUM TOLERABLE ERROR
> 
> We set (see <http://evaal.aaloa.org/current-competition/ta2015track1-2>)
> a maximum of 25m on the third quartile of error to be eligible for
> winning the prize offered by ETRI for each track.  I am soliciting
> opinions on this number: do you find it adequate?  Do you think that it
> should be different for the two tracks (1 & 2), given that no GPS can be
> used in track 2?

Sorry, I cannot give my opinion for this point. 

> NUMBER OF OUTDOOR MARKERS
> 
> We have part of the path that is outdoors.  Being EvAAL-ETRI 2015
> focused on indoor localization, only a small number of markers will be
> outdoors.  I see at least three different criteria to help you have an
> idea on the number of outdoor markers:
> 
> 1) we can decide that no more than 10% of markers will be outside
> 
> 2) we can decide that the average distance of markers is X indoors
>   and Y outdoors, where the ratio Y/X is at least 5
> 
> 3) we can decide that we have only N markers for each outdoor part of
>   the path, with N equal to 1 or 2
> 
> I am soliciting opinions on the criterion we should adopt and the exact
> parameters we should set.

Due to the "indoor" nature of the competition, I consider that the third option may be the most appropriated one. 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://evaal.aaloa.org/pipermail/contest/attachments/20150928/eddb9238/attachment.html>