Amazon's IoT Hacking Contest Won By Voice-Controlled Drone (thenewstack.io) 16
An anonymous reader writes: On Thursday, Amazon announced the winners of its first-ever "AWS IoT Mega Contest," a competitive hardware hacking event held in conjunction with Hackster last month which drew nearly a thousand participants. First place went to an RFID, infrared, light and sound sensor system that gathers data about a sleeping baby and to a voice-controlled drone that sends radio signals using a Raspberry Pi board. "IoT is here now," posted an Amazon cloud evangelist, just four months after Amazon released their own Internet of Things platform. "People are building devices, sites, and applications that are sophisticated and useful."
How does that work? (Score:2)
Re: How does that work? (Score:2)
Right. Because if you don't already have the drone assembled how do you know you have all the parts?
Plus Amazon already has an Alexa api and one of the target applications for IoT is Prime air, so isn't this biased?
Gee... What a coincidence... (Score:2)
Of course... (Score:3)
Reading both, I can't help but to feel that Amazon services are crammed in, regardless of value.
In the drone case, at least in the US all the AWS and echo do is shamelessly plug Amazon, without providing value. In the US, private flyers must have line of sight anyway, so going over the internet is not exactly that interesting except for gratuitous plug for Amazon. Besides, barking orders to a drone seems like the least fun and least practical approach to operating it (though also showing the least sensitivity to latency, since the latency is so terrible in such a scheme to start with).
In the baby case, again the internet is not actually that relevant for a family wanting to study their own child, since they are actually living with the baby, the round trip out to Amazon and back is again gratuitous use of AWS. I particularly thought it was interesting when they reluctantly mentioned Amazon SNS, but with a disclaimer that effectively says that would be a bad idea compared to a more robust and cheaper local alerting design (i.e. 'not intended to replace a baby monitor'). In this example, one *could* imagine extending the reach of academic research more conveniently into the homes of study participants, but that might be viewed a little more creepy, hence the focus on the mostly useless incorporation of the internet to this use case.
It all highlights what makes me groan about 'Internet of Things', the very wording suggests they aren't sure exactly what they want to do, but they know we can make small enough chips/radios/antennas/sensors/batteries to put them on 'things', and some way or another we are going to find practical applications with that, come hell or high water. It's a technology-first push, rather than focused on nice and productive ways that are enabled by the technology. Of course we've long lived with such things in the industry (e.g. a huge part of supercomputing is a race who can calculate arbitrary linear algebra problems fastest, rather than talking about the real problems addressed), but the open endedness of 'things' is a new low.
Re: (Score:2)
Well, I also wondered what Amazon's Cloud was doing in the drone example (see this figure [imgix.net] from the cited article). It turns out it does the voice recognition (apparently with Amazon's "Alexa" service).
BTW, the drone article (didn't read the babyphone one) gives a step-by-step instruction how to setup the different programs. Could be useful also to others wanting to use speech recognition for whatever. Although, given the example phrases in the article such as "Alexa talk to Drone”, “Command Laun
Re: (Score:2)
But do we really need that for speech to text? We've had speech to text for a long time for such restricted vocabularies on systems that the Pi would put to shame. The commands sound like a pretty simple command set, so not so much natural language processing going on.
In a world where we have 2-3 watt processors that blow away what 120W processors could do when we first started seeing commercially viable speech to text. it seems an odd time to start acting like the client device is hopelessly incapable of
Re: (Score:2)
Except those directives aren't really adequate for anything but a gimmick. They don't really scale up to an autonomous flying delivery drone, nor would it accurately show how a remote operator would really be operating if not autonomous (it's tons more efficient to do controller based input to such a system that can only understand basic directives).
Per the video, the drone is probably not even following the instructions accurately. For example, the drone drifts, rather than holding position, probably bec
Security? (Score:1)
I'm assuming that none of the entries have any security protection against being taken over for nefarious purposes. IoT, not for me!
WIFI != Intnernet (Score:2)
Stupid Shit No One Needs & Terrible Ideas Hack (Score:2)
I guess this was that famous Stupid Shit No One Needs & Terrible Ideas Hackathon Iv been hearing about.
Important government announcement (Score:1)
There is currently no evidence linking interest in electronics with RADICAL ISLAMIC TERRORISM. As far as we know. Sincerely, the FAA.