Connected Devices Weekly Update/2016-10-27

• PIB Checkin completed Fri, Oct 21 to provide follow up action items from Oct 4 Gate1 review:
  • Provide details project plan, Architecture diagram and AWS data hosting costs. Refer to Gate 1 PIB check-in for more details.
• Kicked off User study #1 for user interviews in Taipei to start Oct 25 to 28.
  • Initial findings: LED (full lights vs metered), Battery operated vs direct power supply.
  • This study covers UI/ID and installation, unboxing experience, user interaction of setup process and value proposition.
• Hardware Milestone: Received on Oct 25 from 3rd party HW vendor, a list of HW key components and detail Hardware Development Schedule with task and ownership to meet our early prototype deadline for Dec 2016.
• Upcoming Tasks:
  • Will share results from User Study #1 in Taipei during week of Nov 7.
  • Will kick off pricing sensitivity study in Taipei in mid to late Nov.
  • SensorWeb Branding Checkin meeting with Branding team early Nov to review proposed branding framework.

• Findings and recommendations from User Research # 9 are here
• Team preparing for next check-in on Nov 3rd
• Gate 1 pitch date set for 21st December
• Engineering team looking into software modules, will meet platform team next week

Updates

• Released Design white paper outline this week (Platform Architecture & integration patterns). Target 11/9 for the first draft.
• Studied the in-market platforms & IOT devices to create an exhaustive list of use cases (devices, services) for a typical platform to support. This will serve as an input to verify completeness of the platform as well as weighing in different technology / architecture options

Platform extraction

• Initial re-usable components of the platform will come as a way of extraction from track1 products. Starting discussions with track1 teams.
• Prioritizing the other common components to resource by 1H2017 based on above inputs.

OWDCRB (www.openwebdevice.org)

• High level Mozilla CD plans shared with the member companies in the board meeting, waiting for feedback on our plans as well as the suggestions for the new charter for OWDCRB.

Weekly Notes:

• https://docs.google.com/document/d/1yo1AtuozukgfkZwzO745LdEsB1fdZrEWoIYr8F9eydo/edit#heading=h.qgtmzupjdx35

• http://doyouunderstand.me/ is ONLINE and will show WER over time for our TED corpus
  • Currently has 1 data point since we started today
• We have begun integrating DeepSpeech into TensorFlow Serving
• Bug fixing, bug fixing, and more bug fixing
  • If you are interested in following the issues, please do so here here
• Below is a image of our network:

Product Concept Validation

• Launched a consumer study Oct. 25th to test two different value propositions with groups of 100 in the target demographic group. Will monitor results and review the questions we're asking for appropriateness.
• Also launched a Facebook campaign and landing page to gauge interest in Abigail and collect email addresses.
• Working closely with an agency on the Abigail video that will feature how a family of four can be helped by Abigail.
• Models of Abigail have been printed in 3D to be used in the video and available for studies.

• Wrote and posted a blog on Abigail.

Work Week Results

• Team had a successful work week. Was very helpful for the team to work together to get Abigail working end-to-end.

• Now focused on finishing up things like logging, refactoring of user-setup script to use master user, adding the ability to create reminders using a keyboard, etc.

Abigail Study with Prototypes

• Team is preparing working prototypes next week for use in consumer homes.
• Team is developing detailed plan for the visits to consumer homes for this testing.
• Study will include consumer interaction with the prototypes and collection of feedback regarding usage and responses to questions asked of the consumers.

Schedule

• Value proposition validation - Oct. 24 - early November
• Prototype development: month of October
• Assembly of devices: week of October 31
• Testing in homes: November 7 - 18
• Data synthesis from pilot and video/landing page studies: November 21 - 29
• Gate 1 pitch: November 30

• Last Friday (Oct 21st) we presented the target audience and stories for our Pilot/MVP to the PIB (deck and video recording)
• Finished Sprint 6 (Demo Video for the Virtual/GPS Beacon discovery) and started Sprint 7 which goals are:
  • Be ready for the Mozfest and have the app up and running to show the new Virtual Beacon experience in Magnet: we have some pending work in iOS
  • In the client side all development work for the Channels subscription and UX specs to be ready at the end of the sprint
  • Server side: Embellish/replace metadata with the 'extra_metadata' returned from the content service and adding the call to actions to the metadata response
  • Research product/solution fit: Facebook/Landing page test for pre-signups worldwide and app downloads from Londoners. We expect to run a campaign next week when the creative, copy, and legal all are ready. Planning on testing 4-5 value propositions in order to tune it with the goal of achieving a 5% response rate from FB ad 10% action rate from the landing page.

Target Audience (MVP)

• Primary audience: Curious Londoners who are passionate about their city, who are interested in discovering new things around them
  • “Bottom up” curious explorers: they are driven by a immediate experiences with reinforcement about their exploratory behavior. A novel, complex, unexpected, or uncertain event results in a sense of wonder and a desire to explore it.
  • “Top down” interest-driven curious people: they seek out information based on search for novel stimuli. This may be based on specific interests or niche interests.
• Second Audience: visitors to the city of London who want to go off the beaten track and discover new things on their own

Stories

• Breadth Use Cases: Focused on bottom up curious Londoners who seek surprise and delight through exploration as Strange, Interesting, Hidden things around the city.
• Depth Use Cases: Focused to top-down curious Londoners who have interest in a specific topic. Initial Experimental topics: Public Art, Buildings, Jack the Ripper
• We already have around 700 POIs!

Technical Feasibility Study

• Team members met with hardware vendors at IoT Expo
• Sample white label hardware arrived yesterday
• Four week feasibility study targeting 24th November (Thanksgiving)
• Have successfully set up hardware and created our first Linux builds
• Next steps are around ZigBee integration and WiFi camera integration
• Also investigating OTA update system
• First impressions: Good so far, some concerns about CPU & RAM specifications for video transcoding

Landing Page Experiment

• First drafts of landing page and survey submitted to UX/UR/Marketing for input
• Some UX input, waiting on UR/Marketing

Early draft of landing page (partial):

Work Week Progress

• Team has made good progress during their work week, building a prototype game board that includes an LED base and second layer of Halls sensors and magnets.
• Spinner also works with the board.
• Remote play works.
• Game play is working end to end, and a demo will be shown at the Gate 0 pitch.
• The team's volunteer contributor was able to attend one day of the work week.

Gate 0 Pitch

• Team collected and incorporated feedback into the Gate 0 pitch.
• Gate 0 pitch will be presented to the PIB today, October 27th.

Prototype testing

• Prototype testing is pending Gate 0 approval but a preliminary plan has been developed should the team be given the go ahead to move forward.

Engineering

The Lighthouse code now enables a user to do the following:

• Point the camera at an object and press and hold a button until you hear a shutter sound, then release the button.
• Listen for a ping and then say the name of the object. After you've identified the object, you'll hear another ping.
• After an object's image has been captured and identified, you should be able to point the camera at the object, click the button, and listen for the shutter sound signaling the image capture. After a second or two, you should hear your voice identifying the object.

User Study with Prototype

• Team is building prototypes next week to be able to test with study participants who are blind.
• Detailed plan for how to conduct the prototype study is being developed. Working on logistics for the shopping trip, use of the prototype, feedback to collect.
• Working with the Santa Clara Valley Blind Center to coordinate shopping trips with participants. Photos will be taken of items they need to buy while in the store, with the participants' descriptions of the items recorded and stored with the images. The participants will return to the SCVBC with their purchases, where the participants will be asked to identify their items using the prototype.

Form Factor Study

• Also planning to test out various form factors with off the shelf cameras that can be clipped on or Velcro or strapped or hand-held as a wizard of oz style lean experiments with blind users. The SCVBC is helping us to schedule this in mid November.

Schedule

• Complete development of prototype: October
• Assemble prototypes: Oct 31 - Nov 4
• Prototype pilot: Nov 7 - 18
• Data synthesis from pilot and video/landing page studies: November 21 - 29
• Gate 1 pitch: November 30

• Bulleted list item
• Bulleted list item

• IoT Hack Hyperabad Event Picture are up
  • https://www.flickr.com/photos/131182177@N06/sets/72157670770038230/with/28078441314/

• From a Vox interview with Marc Andreessen: "The Echo project at Amazon was 1,500 engineers for four years". Puts IoT projects in perspective...
• Air quality monitoring:
  • Almost half of the local authorities in the UK breach air quality guidelines, according to a government report. Pollution levels exceeded the legal limit in four out of 10 UK councils in 2015, according to data collected from local monitoring stations. Across the country, 169 local authorities were found to have gone over annual EU limits on nitrogen dioxide, primarily because of heavy traffic on the roads. To help combat the problem, this month the government made available £3m for councils to bid for funding to improve air quality.
  • The Environmental Pollution Authority in India introduced an app meant to help fight air pollution by reporting incidences of leaf and garbage burning, and dust generated by construction activities. The app allows people to report incidences of air pollution and also has a section for officials to resolve cases and upload evidence about what has been done. The application will generate weekly reports on the status of the cases and action taken, which will be sent to the officials of the Central and state pollution control boards and municipalities.
• IoT Security:
  • Security analysts believe that IoT and Smart Home devices were involved in the DDOS attack which affected important websites on Friday. The attack used a malware client called Mirai, which "scours the Web for IoT (Internet of Things) devices protected by little more than factory-default usernames and passwords," explained cybersecurity expert Brian Krebs, "and then enlists the devices in attacks that hurl junk traffic at an online target until it can no longer accommodate legitimate visitors or users."
• IoT in Germany:
  • Fewer than 10% of adult internet users in Germany said they owned wireless speakers in July 2016, according to a report from Deloitte, which were the most popular items in the internet of things (IoT). Just 4% of those surveyed said they owned a fitness tracker and fewer than one in 10 respondents were interested in smartwatches. 7% of respondents said they would consider purchasing a smart watch, the leading response along with wireless speakers. Other choices pointed to domestic concerns: 6% said they would consider buying a surveillance security system, while another 4% would consider a home appliance, and 3% would consider a smart lighting system or a smart thermostat. Another report quoted by eMarketer said that nearly 50% of consumers in Germany don't know the term “smart home,” and nearly 70% are not able to describe a “connected car,”.
• Speech Recognition:
  • In a paper published Monday, a team of researchers and engineers in Microsoft Artificial Intelligence and Research reported a speech recognition system that makes the same or fewer errors than professional transcriptionists. The researchers reported a word error rate (WER) of 5.9%, down from the 6.3% WER the team reported just last month. The 5.9% error rate is about equal to that of people who were asked to transcribe the same conversation, and it’s the lowest ever recorded against the industry standard Switchboard speech recognition task. “We’ve reached human parity,” said Xuedong Huang, the company’s chief speech scientist. “This is an historic achievement.”