2016 list

Complete list of design briefs to be advertised to students for 2016 group design projects.

(text transcluded from individual project descriptions)

Campy Bird

Client: Gerard Hester, Morgan Stanley <Gerard.Hester@morganstanley.com>

Flappy bird is possibly the least intelligent video game ever made. But similar game mechanics could perhaps be used to automatically acquire more useful habits, for example if a whole class full of kids learns to take correct actions together, as part of a class activity. Your task is to design a whole-class education app, in which the players can only finish if everyone learns together how to avoid infection from campylobacter - a bacterium commonly found in the food chain that causes 22,000 hospitalisations and 110 deaths in the UK each year. The whole class has to guide a new-born chicken as it travels from farm to fork, taking rapid decisions based on what they have learned. One wrong step by any pupil, and campy bird will educate the whole class further. Think hard about motivation, or else the result may not be as entertaining as we hope!

Originally suggested by Andrew Grant, Vet School,

Equity Exchange

Client: Matt Johnson, Frontier (mjohnson@frontier.co.uk).

The crowd-funding proposition (e.g. Kickstarter) is simple - give us your money, we’ll make a product. So is Mechanical Turk - do our menial task, we’ll give you money. What about situations where everyone is adding value, and the rewards should be shared (or negotiated) in proportion? Many startups, indie video games, and social enterprises work this way. Because there are no suitable online platforms for negotiating and setting them up, at the moment everyone involved has to live in the same place, whether Silicon Valley or Cambridge, to work it out over endless macchiatos. Your task is to create an online service that uses artificial intelligence approaches to help potential co-founders form a team, and create a fair legal framework, financial structure and investor/market engagement mechanisms. Software startups would expect all of this to sit on top of a Github-style repository for content and code, customised to their individual needs. If there is a market here, you’re free to use the result to bootstrap your own business!

Pebble Jogging App

Client: Dong Fang, Bloomberg <dfang14@bloomberg.net>

We would like you to make a hazard warning app on the smartwatch Pebble. When a user takes their Pebble for a jog, he/she can record GPS location and details of any potential hazard they come across on the road, such as broken pavement, dog dirt, flooding... After the run, they can pair their Pebble to a computer program or mobile app, and upload the data via an interactive map. This allows other Pebble users to download the data, and receive an alarm when they are near the hazard. The relatively constrained set of controls on the Pebble, and the constraint that the jogger will want to work quickly before cooling down, means that you will need to provide a intelligent text entry method.

Listening to a million voices

Client: Chris Smith, Jagex <chris.smith@jagex.com>

Automated sentiment analysis tools are big business, but often inaccurate - consider the difference between the tweets “practically the best game I’ve ever played” and “hardly the best game I’ve ever played”, which can get very similar scores. A key to future success will be far more rapid training, collecting actual human responses to huge data sets. Your task is to create an engaging interface for navigating and coding millions of items from in-game chat and forum posts. Analysts should start by typing a word, with the system responding by showing frequency-weighted distributions of neighbouring words from all other texts. Users should be able to move left and right, filter out texts they are not interested in, and “deep-dive” to check meaning of individual texts. The system should dynamically build a thesaurus, learning which words might have similar meanings from context, and adjusting the sentiment weighting of words and phrases. You’ll need to use a super-fast index algorithm such as the Symmetric Compacted Directed Acyclic Word Graph from Schulz & Mihov. The client will provide a sanitized set of chat data from their own industry-leading games, and would appreciate a dashboard-style visualisation of current status.

The Busking Bus-Stop

Client: Ian Lewis, Cambridge Infrastructure Investment <ijl20@cam.ac.uk>

Cambridge buses are fitted with GPS tracking equipment. This is used to predict the arrival time for the next bus, as displayed on many bus-stops around the city. It is also collected in an archive of all bus journeys. It would be nice if the display provided more information, and did so in a more entertaining way, customised depending on how long the bus will take to arrive. Information snippets might include confidence intervals (in layman's terms) for actual arrival time based on historic data, and comparisons to other journey options. Entertainment might be automatically generated poems, fictional dialogues etc, that adapt intelligently to local context such as weather information, news stories and so on - but at a length appropriate to the remaining time. With luck, we hope to deploy the resulting system - at least on the stop outside the Gates building!

Edible Lego

Client: Vaiva Kalnikaitė, Dovetailed <vaiva@dovetailed.io>

It’s great fun to make elaborate decorations out of chocolate, but hard to make anything really impressive, because big complex moulds are impractical to commission. Your task is to make a 3D chocolate sculpture tool that takes a (somewhat) arbitrary 3D shape, and decomposes it into pieces that can be assembled from standard shaped blocks such as chocolate bars. However, any unusual local details will need to be queued offline to order a 3D-printed mould for that specific part. You will have access to a basic 3D printer for testing purposes. Unfortunately the Lab cannot supply chocolate, but this project could be a good starting point for a revenue-making business model.

Put your phone to work

Client: Dominic Nancekievill, G-Research contact via Beth Gorman <Beth.Gorman@gresearch.co.uk> (later substitute Dominic.Nancekievill@gresearch.co.uk ?)

When people browse social media sites on their phones for hours every day, most of the CPU power goes unused. The old desktop equivalent of this problem was the screensaver, which did little of value until it was co-opted for distributed computing projects such as SETI@home. Your task is to make a platform that can perform useful computation in the background on a large number of mobile phones, while the owners are on social media - or even while they are asleep. It will have to run cross-platform, perhaps using JavaScript, but must also give the appropriate incentives to users - will it drain batteries or incur network charges? If so, what kind of application would customers pay to run on such a platform? Would phone sensors offer any specific value? You need to demonstrate an end-to-end solution including servers, mobile clients and an example application, keeping in mind the security implications if either customers or phone owners try to cheat the system.

Reducing food waste with IoT

Client: Vaiva Kalnikaitė, Dovetailed <vaiva@dovetailed.io>

15 million tonnes of food are thrown away every year in UK alone and almost 50% of this comes from our homes. It seems likely that Internet of Things (IoT) devices, mobiles and wearables could be used to detect and avoid some causes of this problem, for example by sharing surpluses or reducing stockpiles by sharing “buffers” of food between neighbours. Your task is to recover some traditions of sharing food among neighbours, with a community platform that enhances social connections through sharing food. Your client will provide a prototype of one novel IoT device - “Spare Bite” (www.sparebite.com), but you should also consider the ways that other connected devices may contribute - iPads used in the kitchen for online references, or even the infamous Internet Fridge.

The Politics of Wikipedia

Client: Ben Scott <bensc@amazon.co.uk>

Much of the content of Wikipedia is encyclopaedia-like, but Wikipedia editors also respond to current events. The goal of this project is to highlight which parts of Wikipedia are more political (or at least contemporary), by correlating Wikipedia edits with recent traffic on Twitter. You should create a custom application for viewing Wikipedia, in which the content the user sees is customised and updated to follow Twitter traffic. Users should be able to navigate according to recency, popularity, or controversy. They should also have some way to know (through correlation analysis) when different topics or pages appear to be alternative views of the same underlying events.

Simulated Stock Exchange

Client: Piers Thompson <piers.thompson@baml.com>

Almost all stock trading is now done by algorithms. Early profits have reduced, and algo-trading strategies must adapt quickly to the changing economic environment, market movements and technological advancement. The most common way to evaluate a new strategy is to “back-test” it against historical market moves, which has the benefit of being based on real data but doesn’t predict how the market will react to the introduction of a new strategy. The aim of the project is to build an online platform to allow competitive testing of algos on a shared simulated exchange with both historical and live market data. The platform will provide an API to access market data and issue orders to the exchange. The user-developed algos will compete against each other interactively and a reporting service will display the analysis of the relative profitability of each strategy and track the best algorithms on a leaderboard. Students will be provided with an algo pseudo code and examples of sources of market data

Sailing by sound

Client: Lancelot Robson <Lancelot.Robson@metaswitch.com>

People with visual impairments enjoy outdoor activities as much as we all do, and local sailing organisations are pleased to provide opportunities for disabled and visually impaired people to experience sailing. One disappointing limitation is that the standard navigation instruments (location, windspeed, compass heading etc) all have visual readouts, so visually impaired people are not able to participate in this central aspect of sailing. Your task in this project is to create an embedded device (you will have technical assistance with hardware - bdj23@cam.ac.uk) that uses a Raspberry Pi to communicate with navigational instruments using the NMEA2K standard, and provides users with speech output in response to commands given via a small number of waterproof buttons on the outside of the unit.

Intelligent Game Designer

Client: Vince Darley, King Digital Entertainment <Vince.Darley@king.com>

There has been a lot of recent publicity about the success of CST alumnus Demis Hassabis' company Deep Mind, in creating an AI system that can learn to play 1980s arcade games. Of course, there is not much profit in playing computer games - the real business opportunity is designing them! The goal of this project is to create a fully automated AI system that can design original new levels of a casual game such as Candy Crush Saga. There is an art in designing good levels - they should be easy at first, get a lot harder, but be ultimately solvable. You will need to construct a population of simulated players, at all different levels of ability, to assess the quality of the levels generated. By optimising simulated play, each level should be challenging, motivating, while also being aesthetically pleasing.

Fly-past Finance

Radmilo Racic <radmilo.racic@imc.nl>, cc: <Taylan.Toygarlar@imc.nl>

We think the Oculus Rift can be used to navigate complex and correlated financial data using virtual reality. Your goal is to integrate time series data on global futures, stocks, bonds and currencies with trader sentiment and pin point market moving trades. The team will be provided with time series data from Eurostoxx, DAX, CAC, KOSPI, Nikkei, ES, EUR/USD, T-Note, GBL, etc. The VR interaction module should allow users to look and left and right to see the past and future, nod forward and back to zoom in and out for more detail. Other movements, such as leaning to the left or right, could be used to select data sets or analyses.

Surprise the Singularity

Client: Andy Fawkes <andy.fawkes@bisimulations.com>

If Singularity is achieved and a super-intelligent artificial intelligence takes over the world, Cambridge is likely to be the first target. The Singularity is not actively hostile to humanity but it will aim to control all movement of vehicles and people so it can prioritise its transport needs over humans. Unfortunately, important strategic information has been published online, where the Singularity can easily find it. For example, the Computer Lab layout is at https://www.cl.cam.ac.uk/research/dtg/openroommap/, and the University map at https://wiki.cam.ac.uk/university-map/. Your task is to confuse the Singularity by creating distractor maps, navigated in a way that a disembodied mind might not realise are impossible, for example as Moebius strips or non-Euclidean spaces. Don't show the whole map at once, where the edges will spoil the illusion. But do include a simulation of real activity - public transport synced with real-time information from Cambridge buses, simulated self-driving cars, and of course locative social media messages from the (simulated) people in the panicking crowds.

Digital Currency for Public Good

Client: Jan Kis, IMC <Jan.Kis@imc.nl>

The digital currency Bitcoin uses a block chain distributed database to guarantee that any transaction can be traced to a unique original coin. The same technique could be used for anonymous authentication of a chain of custody for other kinds of data. Your task in this project is to create a secure end-to-end system for authenticated reporting and verification of organised criminal activity, corporate safety violations, governmental human rights abuses etc. You need to think about how reports can get into the system safely, and how those receiving a report can use the block chain to prove that the evidence has not been tampered with, all without endangering the whistleblower.

==

{{:}}