Tag Archives: Internet of Things

MirrorMirror: booth-based 3D scanner for online shopping

During the final year (2007-08) of my Physics degree at Imperial College, we studied a module called Research Interfaces (RI). This was a team-based module that focussed on transforming scientific research into commercial business propositions.

This was a highlight of the degree for me: I loved the collaborative nature of it and the entrepreneurial challenge was much more aligned with how I wanted to live my future life.

Our product design: MirrorMirror

Our team designed a product with the working name of MirrorMirror. It was a booth containing a network of cameras with a central computer that would stitch together the images to create a 3D scan model of the user’s body.

This would then be used to generate an avatar that would help them choose clothes that fit and suit them perfectly when shopping online.

Additionally, they could see their body on a screen in real time with different clothing options projected over the image as if they were wearing it (so-called “Augmented Reality”). This reminded us of the magic mirror from the Disney film, Snow White (hence the name MirrorMirror).

There could also be other uses like tracking weight loss for dieters and muscle gain for bodybuilders (if a new scan was made regularly to show the incremental changes) or the visualisation of the results of cosmetic surgery.

Technical Design

We produced several outputs for the class including this Technical Design Review.

In that document, we estimated the cost to build the prototype of £1.45m, a total future manufacturing cost per booth of £13,900, and a price point of £50,000.

This is exceptionally high and I believe it is a result of the fact that we were not actually required by the course to do any prototyping work. If we had, I think we would have focused on looking for a cheaper way to execute the plan.

Our original design required a screen behind a half-silvered mirror. I think in 2018 this would not be required as screens are not of incredible quality and image processing technology has come on exponentially in the last decade.

User experience

We believed that there are many high-end lucrative markets (such as wedding dresses, evening wear and saris) where a quicker and less stressful garment trial process would greatly add to the shopping experience.

Our team also saw the potential for future uses such as generating an accurate avatar of the person that can be used as a little virtual model for the clothes that are being selected. Imagine being email a picture of yourself wearing the latest items from your favourite designer and a link to buy exactly the right size for you?

We envisioned that booths could be installed in shopping centres, allowing customers to create a 3D image of themselves which they could then use to shop online. Additional lucrative applications could also include high-fashion hairdressing.

Our plan of the user journey is mapped in the image below:

User Journey for MirrorMirror

Business Case and Financial Model

You can see the basic financial model we generated here: MirrorMirror Costing.

When I say we, it was actually me that had the responsibility for putting it together and I could have circulated the draft to my team-mates before the deadline so we could have had more eyes on it before submission. We got our lowest grade by far for this part of the module, so I did feel a bit guilty! However, it was apparently the same for all the other teams, so my guilt was slightly assuaged.

After 10 years working in and around startups and scaleups, here are what I see as the big errors and omissions:

  • No time series for the values (everything is static)
  • Lag time between initial burn and revenue
    • A proper cash-flow model would have helped clarify this
  • Significant errors on the business model (i.e. how we could get paid)
    • For example, would we really want to make money on the hardware, or would we prefer to make money on the service provided by the software (i.e. charge money for every image processed – a digital version of the Nespresso model)
  • No R&D tax credits, Government grants, or other potential subsidies included
  • No marketing and sales budget included at all!

It is quite satisfying to look at old work such as this and compare it with what I have learned since then!

Final Pitch

At the end of the 3-month module, we had to deliver a pitch to a packed auditorium and a simulated panel of investors (made up by the professors from the Business and Physics department that ran the course).

You can see our final pitch document here.

This was a really enjoyable part of the course. I delivered it with 2 other teammates and we got everyone in the team up on stage for the Q&A at the end.


We actually won the Elevator Pitch Prize at the end of the module which was a very personally satisfying way to end the project. We all received a good first for the course (>85%) which was very satisfying for all of us.

We entered into the wider university’s Business Challenge entrepreneurship competition, but we didn’t get past the initial screening phase. As a result, we all agreed to disband the project outside of the RI module and did not take it any further.

What didn’t we do?

It is quite telling that we didn’t build a prototype!!!

The reason that we didn’t build anything is that we didn’t have anyone that is super-focused on the tech side i.e. that could be a CTO. I also believe it is because we all saw this as a purely academic exercise and not as a true opportunity to start an entrepreneurial endeavour and make a return with it.

This tinkering on a prototype would have actually helped us see the true costs, challenges around manufacturing, and gaps in the business model. In fact, IDEO’s Design Thinking methodology (diagram below) expressly integrates prototyping as part of the design process. This project was perfect evidence of why that is the case.

I wonder if the Blackett lab requires the students on the RI course to build a prototype as part of the course nowadays?

Design Thinking Source: IDEO Mydhili Bayyapunedi @myd | @Young_Current

Protecting investors against earthquake risk in Silicon Valley

I’ve often wondered what would be the impact on companies in Silicon Valley when the inevitable earthquake hits. Turns out I’m not the only one.

Earthquakes in the Bay Area: a “ticking time bomb”

The Bay Area is subject to major earthquakes roughly every 145 ± 60 years at the current rate. Given that it is 150 years since the Great San Francisco Earthquake of 1868, the next “big one” could happen any day now.

Apparently, about 2 million people live on the Hayward Fault and 7 million are in the surrounding area. A magnitude 7 quake would cause damage in the range of $95 to $190 billion, which would be a disaster for the citizens of the area.

Impact on the tech giants

However, my curiosity centers on what would be the impact on the giant tech corporations that are based in Silicon Valley and the wider Bay Area? Companies like Google, Facebook, Oracle, and Salesforce have their HQs and major footprints in the region, so they will be adversely affected by a natural disaster.

It doesn’t seem like they are particularly well-prepared for such an event, according to this report. Although most of the companies have data centers and operations distributed around the world, an earthquake could still cause potential disruption to the main office and therefore the leadership of the business.

As listed entities, this marks a real risk for their shareholders. Could their share prices or even the whole NASDAQ take a tumble if a major earthquake hits the Bay Area? After the 9/11 terror attacks, the Dow dropped by 14%, so this is not unthinkable.

However, I think the impact goes beyond just their own businesses. The services provided by these tech titans represent critical infrastructure for many European and American businesses, so any disruption could have a huge wider impact.

Early warning: a vital tool to prevent damage

Scientists are getting better at detecting earthquakes early. In Silicon Valley, there will soon be an app called QuakeAlert that can give up to 2-20 seconds of warning of an impending earthquake.

This might not sound like much, but even 2 seconds can be long enough for Internet of Things (IoT)-enabled devices to perform vital preparations such as: opening the doors of fire stations to prevent fire engines getting stuck; isolating certain parts of the electricity, water, and gas networks; slow down trains; and tell elevators to open their doors at the closest floor.

Solution: seismic sensor network to short the NASDAQ

Could it be possible to set up a network of seismic sensors to warn when an earthquake was just about to hit the Bay Area and then send an order to a trading algorithm that could short the NASDAQ?

A similar system could be used to create an early warning for tsunamis. One candidate is the mega-tsunami that geologists once predicted could be created by a volcanic eruption in the Canary Islands which would devastate the northeastern US coast (although further review of the original study showed that this is a worst-case scenario and probably will not happen for another 10,000 years at the earliest).

Startupbootcamp IoT|Connected Devices 2016: startups now selected!

It was a pleasure to be a Mentor at the recent selection event for the new Startupbootcamp IoT|Connected Devices Accelerator that is going to be based in the Rainmaking Loft from September 2016.

They’ve picked 10 awesome startups for the inaugural cohort:

We had such a great time at the event – see some of the best bits below!

Internet of Things: Smart Home Security Systems and Burglar Alarms

Google recently paid $3.2 bn to acquire Nest, the makers of connected smart thermostats and smoke alarms. It is a strategic coup for the company, partly because it brings Nest’s CEO Tony Fadell on board, an engineer with a proven eye for design honed during his time as Apple where he lead the design of the iPod. However, it is also a major move as it positions Google strongly to capitalise on a new frontier: enabling web-enabled devices in the home, more commonly known as the “Internet of Things”.

This is a pretty grand ideal in theory, but what concrete, near term opportunities are there for the company to innovate? Specifically, what are the “low hanging fruit” of the Internet of Things?

Home Security System/Burglar Alarm

A prime example of a pre-digital device that is essentially redundant in its current form is the home security system or burglar alarm. Great though they must have been during an age of tight-knit local communities, the audio signal emitted when an alarm is triggered nowadays is delivered to a largely unconcerned audience. Close neighbours in big cities or even towns are largely unknown to each other, so burglar alarms tend to just add to the cacophony of the urban ether rather than acting as a call to action to apprehend burglars or call the police.

Connected Home Security Systems: the burglar alarms of the future

Features of a smart burglar alarm

A smart burglar alarm would be able to send the signal to the relevant parties by SMS, email or signal to an app on the user’s smartphone, tablet, or other connected device. In addition, GPS trackers on the devices of nominated parties (relatives, friends, and maybe the emergency services) would show the central system of the app when they are near to the house and if they are within a certain response time, they will be also sent an alert by the app so that they can intervene if the householder is too far away to do it themselves.

Nature of pre-smart alarm signal is redundant

Another critical characteristic of pre-smart burglar alarms is that the information carried by the signal is too generic to call for action in a compelling or efficient way. They are binary, with an off state (“silent”) or an on state (LOUD NOISE!!!). This leads to a confusing call to action, as there is too much ambiguity for an actor to investgiate: is there really an intrusion or is it a false alarm? Is the burglar still inside the house? Has somebody already been informed and are they already in the process of dealing with it?

A smart home security system concept

The exponential decrease in the cost of sensors means that a smart burglar alarm could actually convey more specific and hence useful data to the nominated parties, enabling a more effective call to action.

For example, the specific trigger point could be communicated (roof, ground floor windows, front door) so that the alarm points can be investigated quickly and the possibility of a false alarm ruled out in less time. Infra-red cameras could measure if there are people inside the house, counting them and perhaps even identifying them using facial recognition.

The triggering app would allow the user to see who precisely has been signalled, who has acknowledged the signal, who is acting on it, their estimated response time, and their current location.

Competitors and Innovators

Piper’s Home Security System and Mobile app interface

There are some impressive innovations in this field such as Piper and Canary, which are standalone video and sensor units acting as a “mini sentinel” in the home. Piper, which is already available for purchase, also acts as a household device controller and could therefore turn on lights in the home if signalled to do so. They both have the awesome idea of adding video to the equation, meaning that if the motion sensor is triggered, the user could immediately switch to video to see who the intruder is. I imagine the video stream could also be recorded for legal purposes in the event of a burglary.

Home CCTV enabled by Piper

Priced at $239 and $199 respectively for a single basic unit, the issue is that the devices only cover one room each, which makes them an expensive solution for a whole home, although a promising start and a massive leap forward. Canary smashed its request for crowdfunding on IndieGogo so expect to see the first units available later this year.

The miGuard alarm system from Response Electronics uses an integrated mobile phone SIM card to communicate with your phone by GSM/SMS and has a total system cost of £269.95 (about $452). This is a much more attractive price point for a whole-home system, but the technology is not smart enough to capture the full range of possibilities offered by the rapidly decreasing costs of technology and increasing connectivity of web-enabled devices.

miGuard Home Security System – schematic diagram

Other potential players

Of course, this concept is not just a possibility for Nest and the innovators outlined above, as there are other innovative technology companies who are trying to get into the smart home space.

As a Brit and Imperial college alumnus, the most notable example I can think of would be Dyson. My rationale here is that Dyson are one of the great innovators of UK industry and a global pioneer in domestic technology, highlighted by its recent partnership with Imperial College on robotic vision with a view to implementation for autonomous vacuum cleaners.

This is an intriguing partnership, given the promise of Imperial’s recent contributions to the field of Simultaneous Localisation And Mapping (SLAM) in addition to the fact that the unlocked value of Dyson’s disruption of household technology markets runs into the billions.

AlertMe, the British home monitoring controls company could also have a say in the development of this industry on the software side through their Smart Monitoring platform, linking all devices in the home.


A connected home security system is a complex endeavour, given all the possible flows of information and control that are being unleashed by the digital revolution. There is a range of possible ideas already in the market, addressing the various price points that could be considered by consumers.

The advantage that Nest could have if they developed a smart home security device or burglar alarm is that they already have two products on the market that could feed into it, not to mention their experience of successfully designing the necessary user interfaces and hardware for mass consumer uptake. Combined with Google’s expertise with algorithms and handling large data sets, it is a mouthwatering prospect to think what they could do together in this area.

Given the fact that smart burglar alarms will be such an improvement on the pre-digital state-of-the-art, I wouldn’t be surprised if the eggheads at Nest have already been incubating something like this for some time. This guy has even mocked up a great example of how Nest’s existing thermostat interface could be converted into a burglar alarm.

Nest’s Thermostat as a Burglar Alarm

I would not be surprised if there are further acquisitions in this sector in the coming months. These are very exciting times for this emerging technology market.