July 23, 2015 at 7:28 pm #1646rejsharpParticipant
Paris Developers SDK Meeting:-
I attended one of the two sessions, here is what happened.
About a dozen Developers attended each session. The session was divided into three parts:
a) a walkthrough of the Scio SDK process, current and future steps: Yaron Dyclan VP Products.
b) demo and general info: Damian Goldring CTO (co-founder)
c) hands on session with prototypes, building test collections: Attendees
The Talk: Yaron went through the whole process of how the SCiO the PC SDK Lab, the Mobile SDK lab, the API kit, the Cloud conversion, and the User mobile Apps fit together.
We will get access to the API and User App shortly. The SDK software needs only final polishing. I was pleasantly surprised to learn that we can use different processing models in the same App (giving extra confidence in the answer).
Another strong message was that the quality and quantity of samples in the reference collections must be carefully and completely done. Later in Damian’s demo we covered this again. As an example: The test dairy collection for fat and protein could be built up entering the data from the packaging. This would be good enough for idea evaluation, but nowhere good enough for the production data. Consumer Physics are sending all their samples to an accredited (and tested) lab. Obviously sample identity is easier than quantification.
The Demo: Damian showed the current User App with three demo features, Dairy estimation, Alcohol in drinks estimation, and pills identification. The App was not the final version but worked great. I am sure you have all seen these before, so let me pick the key new points.
There was a new sampling shade, that blocks out excess ambient light, and holds the SCiO 1cm away from the target ensuring more consistent results. It was magnetic like the holder for pills and other small stuff.
There was a new liquids sampling aid, designed to overcome surface reflection, and inadvertent wetting of the SciO. Imagine a small tube with open cage walls at the bottom to allow the liquid to cover a reflecting baseplate. Scio sits at the top, safe and dry, peeping through a protective glass window which can touch the liquid. Brilliant idea.
These will be delivered when ready, and not be allowed to delay shipments of the SCiO to us.
We then discussed limitations. On the good news side, some substances which have strong resonance under IR such as alcohol can be detected (and estimated I think) at less than 1%. Limitations to getting useful results are with shiny metals (already known), very dark samples that absorb all IR (careful blackness in visible light does not always mean black in IR). Powder mixtures are hard to quantify, mostly because it is impossible to get valid reference samples – the bloody grains segregate if you look at them, let alone stir them.
We moved on to discuss how developers can cooperate together with collection creation and Apps – simply they can use a common ID in the system, but CP could look at creating a joint working protocol if this is popularly desired.
Damian moved on to plans for User collected scans to add to the database. This will start at the end of this year, or early 2016 – most likely with medication samples where errors are less likely. Users will be invited to scan every medication they can lay their hands on, adding the classification info in a comment format. CP will then clean up bad scans and infrequent material scans before adding the good scans to the Database of Matter. That will set the scene for the future. Estimation scans will need a lot more care about data provenance and value measurement accuracy.
Hands on: Well there is not much to say – the attendees got furnished with prototype SCiOs and iPhones if needed, and got busy creating test collections and analysis models. They all seemed to be having fun. Darn – they made us give the hardware back!
Unfortunately we did not have any time to socialise, and the attendees rushed off to catch trains and planes. I wanted to ask why none of them ever appear here in these comments!
We did give short video interviews at the end, and I guess they will be published soon.
Shipments: August would be the big month with shipments starting. Bear in mind that they have 12k units to ship to Kickstarter Backers, and then a couple of thousand more for on-line sales – it will not happen in a day.August 23, 2015 at 3:09 pm #1784UCPHParticipant
Thanks for the summary, wish I’d been there. Looking forward to more meetups in the future. Now that I have my SCiO in hand it makes things much more realistic!August 27, 2015 at 1:20 pm #1893rejsharpParticipant
Thanks for that! Do you have a prototype or a finished production item?
- You must be logged in to reply to this topic.