In the new Virtual KOP there is $75 per student of AWS use. From my understanding AWS is cloud computing and hosting services. In the blog thread there was some discussion of how it can be used for scouting and for data analysis. However, considering that the datasets from scouting are (comparatively) small, what is the problem with doing it locally?
I am interested in how other teams are going to use this. Does this allow anything new that was not possible before?
AWS has some cool services like scanning (OCR) and machine learning. Nothing that wasn’t possible before because you could have used local libraries. But it’s hard!
Also, you can build a chat bot to tell teams at a regional/district event when they are up or what they are ranked or whatever. A chatbot could be fun. Make it easier to look stuff up on your phone.
And there are many other possibilities. I like that FIRST isn’t telling us what to use it for. The year of the Kinnect, we were told what to use it for and, um, didn’t. But a 14 year old used it to create a cool mechanism to look at our CAD via gestures.
So I expect it to be interesting to see what thousands of high schoolers can come up with! Maybe Amazon could give extra credit (extra $ of credit towards services) to teams that come up with cool uses.
Microsoft (and Google and IBM and…) also have cloud services. So this is a good commercial for Amazon while enabling cool tech.
So a couple years ago we had IBM Bluemix in the vKoP, and they had a data science package that included cloud python and R IDEs. And at the time I had the possible idea to use an R package that accessed FMS data and then would be able to setup periodic calculation and generation of scouting analysis. I also had a Google Sheets (Appsheet app) scouting database that could be linked in through other packages (potentially even outputting the results back to the app).
So that is something I think about with cloud access, that you could have it doing the processes without needing a local computer and set it up to do it automatically. It would be good for realtime updated scouting (assuming you have a connection).
It is very interesting the way they decided to structure this program. Each student would get their own account, with their own amount of credit. So it’s not like you can do 10 students * $75 = $750 and then use that all on a team account, which means this isn’t ideal for running web servers (such as a scouting app, team wiki, public website, etc). It also means that the team can’t easily persist their cloud resources as students join and leave.
We got a $500 grant from AWS a few years ago for our scouting app and it was just a single promo code. Just like any other company would, we had a single account, a single bill to with that credit applied, and then we used IAM to create logins for the different team members, mentors, and automated services that needed access to various parts of the account. Seems like this approach would make far more sense than splitting it up by student.
I haven’t looked into it yet. Is the donation really $75/student for however many students? If so, that’s pretty cool. You can get a decent EC2 instance with a good bit of storage for around $30/month. So, $75 is enough to do some decent tinkering for a couple months.
It appears that this approach is really more tailored towards getting several different students familiar with AWS, and preventing the total value of vouchers from actually being redeemed. After all, most sponsors are doing this at least partially because they need more, experienced employees. In Amazon’s (and Dassault’s, and Autodesk’s and etc) case, it’s to get students to go back to familiar tools once they enter the workforce. The ROI might not be fantastic, but it also puts Amazon a more positive light than they find themselves in with informed individuals.
It is each student gets $75 in credits to their name, and it doesnt look like students can pool their credits together.
$75 should be enough to reserve a t2.nano EC2 instance for a year, which should be able to run WordPress if you don’t have a team website already.
More than just fun, a slackbot is the primary way our drive team views scouting data for pre-match planning. We were already running this on a free-tier AWS EC2 node last season.
Tell me more about this Slackbot…
We’ve already been using Airtable for scouting, which is more or less Google Forms/Sheets with a bit nicer UI. The Slackbot pretty much just had one command: @scoutingbot match x. It would grab the alliances for that match from TBA, grab data for the 6 teams out of Airtable, and try to calculate an estimated score for each alliance, along with some basic average stats for each team.
An example output:
------RED: 74.7063492063492---------
Team: 2957
Hatches: 0.75
Cargo: 3.25
Climb: 2.25
Team: 2702
Hatches: 2
Cargo: 2.25
Climb: 5.25
Team: 1885
Hatches: 4.25
Cargo: 4
Climb: 5.25
------BLUE: 67.5---------
Team: 7797
Hatches: 0
Cargo: 0
Climb: 2.25
Team: 4130
Hatches: 2.25
Cargo: 3.75
Climb: 6.75
Team: 4039
Hatches: 3.75
Cargo: 2.5
Climb: 12Code can be found here: https://github.com/iliterobotics/FRC-Scouting-2019. It’s definitely not pretty; this was a last minute idea that was thrown together in a few days, but it was super useful as a concept and something we hope to expand on for this upcoming season.
In my opinion, AWS Educate is not very useful. I got it on my own (not through FRC), and it turned out that all the work for the application really wasn’t worth it (it is good that being through the KOP expedites the application). Some things that limited its use for me:
If you want are more flexible cloud credit for students I would go with the Azure one through the GitHub student pack; it doesn’t have most of these issues.
This is a cool sponsorship that has a lot of potential uses for teams. The $75/student is a little awkward but you can realistically get a ton out from just the free tier even without this credit. Free tier also lets you use IAM as the more typical permission management system so multiple people can interface with the same resources (I haven’t used educate enough to know how its done there).
I suggest people check out what you can do even without this discount: https://aws.amazon.com/free
All of the below are per month, so I expect many teams would have little problem staying in the limits.
750 hours of EC2 t2.micro: This allows you to run a virtual server 24/7 and is probably powerful enough for many use cases. I recommend checking out AWS Elastic Beanstalk which makes deploying a web server relativly easy and doesn’t require any charge with the free EC2 instance. You can use Python, Node.js, Java, Go, and many other languages.
5GB of S3: Simple storage for scouting photos or any other files.
750 hours of Amazon Relational Database: Run an EC2 instance hosting one of many availably flavors of SQL database 24/7.
25GB of DynamoDB (NoSQL): Another database option if you prefer NoSQL.
1 million Lambda Requests: Lambda is a serverless compute platform which is super easy to set up in the web console. This is cool because you can make API endpoints without needing an EC2 instance running because Lambda only runs code on demand. You can develop in basically any major language you want and trigger it with either an API endpoint or other AWS services.
I would say this is a perfect use for Lambda. You don’t need to have a server running constantly so you can use your free EC2 instance for something else. I’ve set up a slack bot before using Lambda and API Gateway. Mind you this one was pretty useless as it was designed to send memes, but you could easily make it useful by querying TBA or scouting data. You can read more about Lambda here.
(Disclosure, I’ve previously interned at Amazon Robotics and take advantage of a lot of these free tier services for a school project)
Right up until you try to use the credits and realize that you need AWS training to make sense of the hundred different products some of which actually compete with each other.
My favorite crazy one is AWS Ground Station, for all your satellite control needs. I also echo the overwhelming amount of options, FRC teams should probably simply stick to the core group of services listed on the website as ‘Featured Services’.
The scouting team would also supplement the raw data with notes when needed. For a first iteration, it was excellent. This is definitely a great way to push info from one place to another; sometimes the drive team would be off dealing with something, and someone on the scouting team would activate the bot for us.
We had a specific channel for scouting bot, so the whole team wasn’t spammed with specifics.
Last year we used AWS for our scouting application. The benefit was that we didn’t have to worry about syncing devices. Our scouting app was a web app and the server stored the data, we could just bring it down to tableau with a couple of different restful enpoints and then analyze it however we wanted. The biggest benefit was collection.
no reserved instance support in aws educate accounts unfortunately.