This page contains suggestions for various interesting projects and prototypes we'd like to see someone build. They're mainly aimed at Part II Computer Science students at The University of Cambridge Computer Lab.
If you actually try anything listed here, we'd be interested to hear whether it worked. If you're a Part II student and would like to pick one of these up for your project, we'll happily provide support and resources - send email to Richard Watts, rrw@kynesim.co.uk.
There's been some fairly long-standing research at the Computer Lab into video user interfaces. This is becoming rather more interesting with the development of very high-resolution image sensors and displays and associated software correction for highly distorted geometries.
It would be interesting to set up one of these projector/camera setups by a scrabble board, for example; the camera can then sense the state of the game and the projector can project feedback to allow on-line rules checking, network play, etc.
Profilers will generally give you a good idea of where the cycles are going in your code. But that's not the main cost these days - most of your time is likely to be spent in L2 misses. This project aims to develop a memory profiler that measures cache misses and, as an extension, tools to perform EDF DMA scheduling (DMA being really the only thing you can seriously control).
Watching DVDs or TV is now a social activity, but control models for TV and DVD are typically single-user and quite prescriptive. There's been a bit of interest recently in cooperative models for watching TV; it'd be nice to have a cooperative way to control TVs.
Soft processors like Microblaze can have their architectures tweaked on the fly. This potentially allows the pipeline and ISA to be tweaked to a particular application.
This project will aim to develop a feedback-directed synthesis engine for one of the issue logic, cache or instruction set. A cunning implementation will change the processor configuration on the fly to cope with new conditions - just in time reconfiguration would be a nice idea.
Remote controls are annoying - there's typically only one of them and it's hard to tell which remote goes with which telly/DVD player/AV switch/your cat ..
So we'd like to explore different modalities for controlling your TV. Gestural interfaces might work, but depend on picking a set of gestures that people don't find embarassing to perform - this project aims to develop a set of gestures and then use them to control a TV.
So you'd like to use your mobile phone as a TV remote? The obvious way to get data to and from it is over the internet, but how do you do identification, authorisation and how do you work out which TV the user intends to control?
Sweeping assertion: phones will soon all have GPS. The telly will also have GPS and a camera. It's fair to assume that any mobile phone pointing to the telly is authorised and intending to control it (or rather that the user is in whacking distance).
Making this actually work may be harder than it sounds...