GSoC-Adding Sense to Beagle, My GSoC problem statement

Hello

I want to make a voice controlled embedded device working as a
personal assistant to the user. Internet is a vast hub of information
but sitting in front of computer and getting information is useful to
only some extent,information is more useful when it can be used in day
to day life. An embedded personal assistant using internet as its
knowledge base to answer our queries is what we are proposing. There
are a lots of services available on internet which provides exact
information to the user queries rather then a bunch of
search results, using these services, this embedded device can answer
to many of the user queries as a real human being. These service
includes answer engines, which provides answers not the results,
querying to it will give A two piece embedded device responding to
user you a result or no result, means 0 or 1. These answer queries
using internet as its knowledge base engines can be used in an
embedded device, making them more useful for users, other services
like Google maps, Google business etc provides similar results. The
beauty of this project is that the usefulness of these service
increases a manifold by embedding them into hardware. Its like
providing the information exactly when it is needed
instead of making the getting information as a separate task. The
device being a personal assistant can do a lots of things other then
providing information and solving user query, it can give suggestions,
it can assist in managing schedules, meeting etc. It can work as a
social network device, can update twitter, facebook etc status, read
out tweets, by
user voice query, making this task more obvious, then to go and type
what are you doing.

The Problem Statement

Internet has grown unbound, since its emergence.And is growing
continuously. The need is to use this information in a way it
become more useful and easy to access. Since not all the information
can be consumed sitting in front of your computer and making
queries. We need to address this problem of making an intelligent
device which can present to you this information when and where it is
required. The problem statements hence can be framed in following
words,

Problem Statement 1
“Since a bunch of online services provides solution to day to day
problems like finding directions, finding cheap goods, commuting,
finding right vendors, and all that in real time, using these
information, right at the point we need, i.e. in market, at workplace,
on way, there is a need of a device which can take our queries to the
internet,makes a request to these service and get back to us with the
required result ,that too without having to write,or access computers”

Problem Statement 2

“Need an embedded intelligent device,which can tap the enormous
information available on internet,present in a way that can replace a
human personal assistant in a super challenging way, the device should
be, intuitive, energy efficient, easy to adapt, up gradable, extend-
able and can be mass produced”

Problem Statement 3
Despite of large number of internet services available to solve day to
day problems, these services are not of much use for visually
impaired, let there be a device that can make internet queries with
voice and get back with results in sound. Devices are need which can
make use of map, directions, shopping services available on internet
directly accessible to visually impaired without having to use
computers”

Combining all these problem statements, we require a device which can
replace a personal assistant, use internet as its knowledge base and
do wonderful things with the power of Beagle :slight_smile:

I will use temperature sensor , GPS , GSM ,DTFM,speech recognition in
my device so I would like to add it on the 'adding sense to beagle' .
I would like to know whether anyone has ever started such a project so
that i can acquire a basic idea of the path i need to trace to reach
my goal .

These statements are the motive behind my idea and hence require
mentors for the discussion on my proposal and guidance to unleash the
power of beagle and solve the above stated problems.

waiting for great response :wink:
thanks

Dinesh,

Interesting problem statements. I would love to see the actual plan
and concrete deliverables for it.

Frans

My basic plan

The device has two parts, small handheld Beagleboard and a bluetooth based earpiece containing microphone and earphone. The following can be seen as a work flow of device working:

  1. User has bluetooth earpiece on her head and handheld in her pocket. Both device powered on, user makes a query using voice, to tell the device that user is making a query it uses a keyword based on type of query something like ‘weather’ followed by the query. The
    bluetooth receiver, get the query, the software at the receiving end converts it into text. On the basis of query type, device makes a request to the internet service it is configured to use.
  2. On receiving the result, device converts it into voice signals.
  3. These voice signals are sent back to the bluetooth earphones as a response to the user query.
  4. In response to the query, there can be images, videos and large text too in some cases, which can be seen on the handheld device.

thanks

Dinesh,

This sounds like it should be an excellent application for a cell
phone platform, since state-of-the-art cellular phones support GPS,
bluetooth, communication, and network connectivity. Perhaps this could
be developed under Android.

Since the speech interaction will require large-vocabulary dialog
capabilities, it would make most sense to transmit the speech query
and related meta-data, such as location, to a network-based speech
server where the speech recognition and dialog management would take
place. The speech server could integrate with the internet query
software also running on the server to determine an answer to the
query, and then provide the desired multimedia output to the phone.

Lorin