This idea draws from several sources of inspiration some of which are:


1. My previous research in Sensor Networks and lofty goals2. Ray Kurzweil's vision of Virtual Reality in which people want to interact and see the world through their avatar's eyes3. Kulathep's idea of video streaming for tutorials4. Ultimate goal of Ubiquitous computing.5. That Bruce Willis movie on the avatar thingy.6. Microsoft research on combining mobile video streams to enhance quality.


Idea explained:Well, simply put, it operates like torrents. The more you share the more credit you have to ensure fairness and good distribution of video. Anyone can become an avatar for someone else in essence sharing the state of the world at that particular instant with the observer. This should be more effective than actually trying to deploy video sensors to sense the world at any point in time.


Note: This is *NOT* trying to be a social network, rather a sensor network!


Possible name could be MAG (Short for Mobile AGent)


There will have to be external hardware involved. Cheap hardware that *IS* fashionable. Think ipod nano like devices or even smaller with cameras.  Might come in different shapes and sizes but must be flexible about where and how it is worn. On neck, arm, pocket (protector)...etc.


I can get several video streams at the same time and stitch them together. $5


Does augmented reality fit anywhere in here?


Realize the modifie reality browser.


First item:Context awareness.Inferring population informationPlatform?


Reach into the matrix and contact an actual person.


Build the map, allow folks to connect their phones.


Applications: Private security (Indoor security)Next generation maps.


Name: Metaverse & SenseMe.


This seems too passive, is there a way to get the user more engaged creating content?


Initial focus is on indoor location and real time tracking(Look at 2019 vision and Nokia prototypes), bigger vision is for outdoor tracking in real time.


How and why do people use location based services? Is this a location based service?


If it is strictly "non-social", that is, you only use it to gain information about your surroundings, and sat certain locations, people can uncloak themselves for a particular period of time and receive any messages...then?


Roadmap of features:Build tourism version with following capabilities: ability to track a single user, multiple users, search and index places, people, things, 3d construction of surfaces.


Build security version with:Ability to search, track and follow particular entity, history and patterns of movement.


Tagline: searching the real world literally? Should be positioned as the easy to love, struggling challenger. Remember AAPL's ad about disruptors.


First Focus is on the indoor locomotion capabilities...messaging becomes secondary. Think tourist guide model.


Seems very well suited for Indoor discovery and tracking.


Scenario: chuka is riding to work on the train and Daniel asks for chuka's location. When Chuka gets prompted, he is able to share both global info about where he is(gps and current methods) and local information(actual reality about the train).


For mapping would automatic echolocation work? It does not have to be 100% precise.


Another great pitch would be that the networks are getting overloaded, we establish 'metaverses' within their own Internets and all computation remains A local affair. Capture local semantics that would be otherwise missed completely. Support connecting isolated worlds.


Going for a much bigger play than just tracking humans...eventually will offer services for intelligent / persistent sensing environments.


Measures of interactiveness by monitoring human cluster formation


In light of the recent changes in computing paradigms currently occurring (shift from desktop centered to mobile) I would like to push this boundary even further, to embedding computing everywhere deep within the fabric of our physical world. That means we deliver on demand sensors, actuators and access to the data streams of these sensors. Enabling interaction with our physical world, querying and causing "effects".


As a practical product, we provide instore customer walk in analytics. Another cool idea would be brenda's idea of tracking papers for signature.


Name: Jerks? (Unit of dy/dx of acceleration?)


Collaboration amongst local nodes that can sense the target, then report generation, then estimation of distance from sink node.


Trying to digitize more of the physical world to make it indexable, queryable(searchable), easier to manage, actionable. Real time data streams. To provide indoor realtime mapping services.


Market to government!


Another opportunity, when outdoor reality browser is rolled out, allow scenarios such as this:
A person in office A (who has tied his credit card to his account) buys a coffee from the nearest starbucks (found using spatial memory of his surroundings or through 'available' service). He interacts with the virtual cashier(an agent) buys his coffee and then thats it. It will get delivered to him where-ever he may be. Even if he is travelling/on the move.


Animate certain aspects of the system. In the event that the person cannot be found, try using direct message (which may be text, email or call).


Can a quick mash-up occur using the above vision? Also, checkout Skyhook's service.


In summary, you create a 3-d world view representing some part of the observable world of a person. Avatar interactions take place and services bought within can be provided (in real life) hence the name reality browser. It mirrors the real world and tries to keep as much information as possible. This information is gathered from several real world assets/sensors.


Kinect like avatar like meeting place?


Not Kinect, rather, a virtual world where u interact with agents like Stanford vision.


Everyone has a unique world around themselves. It should be composed of services which reflects the real world commercial (and other?) services around them.


In a museum, request help, and location automatically shared with attendants who proceed to find you.


Formal description:
The creation of an application platform  (Sale of actuators: event handling and delegates) that allows the delegation of tasks(e.g order a pizza) to a representation of the user (a digital user/avatar/surrogate  ). This digital user also allows means to notify user of real world events. New services can be added by exploring service interfaces). It also makes extensive use of abstractions(Event locker...). We are creatures of habit.


Revenue opp: App platform, Make new suggestions when executing routine tasks.


Two opportunitites: Security related tracking and virtual agents.


Verdict: Indoor mapping for now.
Review: SLAM Algorithm and "Custom Approach"


Possible name: Quads.
Break down into bite sized projects. First should be sharing world.


If a restriction on "world shares" is placed to only places you are directly connected to (like you can only share your current location) what does it entail? Can we capture detailed info of the world at such a small scale? What is the usefulness of such data?


Changing the physical world in what ways?


Rather than creating a browser how about a "tell me if you see a red car scenario for sensor networks?"


Maybe the reality browser can be restricted to certain environments for now. Rather than targeting just anyplace lets try restricting it to colleges for now. ==> Draft components./Outlines required today.
For college data try to search and find exact location of anyone on campus.


Do we have to track human beings? Start by tracking the damn train in real time.


The mac donald scenario. Discover available services on the fly. Start with the caltrain experiment.


Group oriented services? React can make use of group info.


Wanna integrate the internet (and associated services) into the very fabric of our daily lives.


Possible services: Local chat, navigation services. Start with museum specific services, incorporating location data.


Rather than an app centric world, why not service centric?
Give the local businesses and your pal a chance to have their own "app".


Combine location specific services with service platforms to disrupt app centric world.


Agents. Design sample agent tasks. To reason about world?


Pitch: We are re-imagining how people will experience the internet of tomorrow. Rethinking how all this info from internet of things is displayed, interacted with.


Users are not the customer rather they are the product. Also, Kinect platform __must__ be used as a distribution channel! That is imagine getting analytics about their life using kinect.


Integrate data from publicly available sensor


Always weigh the consequences of making/not making certain decisions *FAIRLY & COMPLETELY*


Phases:
Build out sense-react metaphor....the environment becomes the presentation layer.
Build functional environment prototype without sensing capabilities
Design and implement sensing framework for gathering external weather elements (temperature, humidity) for a particular region.
Look at using rf strength for estimating __foot__ traffic
Real time traffic information for a particular region
Research internet connected power devices and internet connected rfid devices.(View downloaded papers)
Research and implement more advanced actuator controls for devices
Finally instant search for devices.
Cost: $1 a month.


List sensing capabilities we will aim to have initially.


KISS: Keep it simple stupid.
Pitch can be focused around automation through the use of your puppets. Why will people use it? Because it actually saves you time. One more thing you dont have to pay attention to. Problem, how does the search reality factor in? Maybe the sense-react framework handles this.


We see 'sense-and-react' scenarios everywhere in the real world similar to map-reduce for breaking problems down.


Create a standard interface for sense-react which people can extend and use to compete with each other in our eco-system.


Best way to create the sense react (map-reduce like) representation of tasks is to build samples and break down representations.


Seperation of all the various different components is very important if we are going to scale (Remember, dont over engineer...yet)


Layers: Presentation Layer/Eyes (Primarily Unity 3D and associated tools), Processing Layer/Brains (Node.js, Twisted, Tornado) ===> General Web App Architecture
In addition to these two layers, there is the kernel layer which supports the "driver" architecture (data suppliers) and the application apis (for 3rd party developers to extend the platform) ==> External Applications.


We envision a centralized sensor-actuator control application platform dealing with several of the issues around security, control(of devices) and access to sensor feeds.Think of the entreprise and being able to track the state of documents, people etc.


Your devices should work for you (think of the fridge, in which it auto generates shopping lists), since it is now 'smarter'.


The trends seem to be towards more data. Amazon, google Facebook have in common understanding of user intent and properties through the use of social networking data and search data all for commerce (be it the sale of ads or product). A general trend that seems to be happening is the explosion of data everywhere. Key would be enabling citizen scientists, that is autonomous experimentation. Uncover insights, without a need to understand complex mathematical models behind such efforts.

Facebook defines who you are based on 'who' your connections are.
Google defines who you are based on 'what' you search on
Foursquare defines who you are based on 'where' you visit most

If I were arrested, who would be the first person I call for help?

Possible name: Quad?

"Query world"
Goal: Expose services offered by resturants through web services.
The sense apis are designed around sense-and-react services. At the fundamental level, everything is an entity which can be talked to via web services. All entities can be "sensed" and could expose methods to "react". An entity can give rise to other entities or sensed data. "React" occurs when you cause an actual state change on the entity you sense (for instance resturant order as opposed to getting streetview info from google maps) causing it to save all or part of your info to "trigger" a response.

Next: Create sense taxonomy for entities then BNF parse trees.

sense environment | [gets availabile services/service groups]
services/service group -> [Specific list of services like Navigation, Food..actual company listing]
company listing -> Support APIs for customization, How should data logic be handled?

Data logic case study: A resturant might have several a diverse menu with subcategories, how will they expose this service. Solution, use a tree like representation where similar services are grouped under the same node. Use the sense and react metaphor to explain what happens to each node(appropriate name?). Movement up or down the tree should not have any side effects, meaning users can jump to any already exposed part. When it comes to react, we should

What can sensor data be used for (apart from the obvious visualization)? For instance using Paul Suffo's logic, is there a way to provide a real world service like autonomous taxis in exchange for street view mapping (subsidied by cost of taxi trip).

Thought: The web is made up of various content (text, media etc). Physical enviroment can be represented with data from different types of sensors, what is the link if any?

Exercise: Write out trends currently playing out(wrt mobile and desktop): data control, ui, size, weight, power, connectivity, mobility, storage, specialization, threats, weaknesses.

          Desktop                                        Mobile
UI:      Detailed,Noisy                              Glanceable,Quick
Size:     Bulky,Immobile                              Small,Portable
Power:   High                                              Low
Storage  Near infinite                                   Very limited
Speci     General purpose                             App oriented
Connec   Internet only                                 Sensors,and others

The phone seeks to be completely freed from the desktop, and become its own application powerhouse. Your phone becomes your identity, payment system, and portal to the world of internet services,

Problems existing right now:
-> Severely constrained bandwidth in mobile networks
-> With phones holding more personal information, malware attacks become more rampant.
-> Overwhelming number of apps leading to too much noise.

An application that uses a persons' personal data available on the phone in combination with other external data sources to provide tangible service. Hardware business?

Evenings, passive activities win. Activities that only require consumption and no in-depth thinking. It is always time for Facebook(inter personal relationships).
Utility applications that cater to a specific need (e.g yelp for discovering places of interest).

Forward looking: How can we incorporate elements of robotics, in mobile application.
-> Can we use elements of swarms(swarm robotics) to do something?

Human behavior monitoring and profiling given sensor data?
Answer the question 'what do you hope to gain from this data?'

Messaging back in time application. 'Forget scenarios'