EECS 373 Fall 2012 Sponsored Project Opportunities Prabal Dutta Background: There are lots of interesting embedded systems out there that researchers could use but either don't have the skills or time to actually design. However, some of you may have the time and skills, especially after 373, to create such things. We'd like to encourage that by identifying some project opportunities. Most of these may not seem as cool as building a small Segwey or a quadcopter. On the other hand, they are quite useful for addressing a range of research problems. Picking one of these projects will ensure that you will get a lot of attention from the 373 instructional staff as well as a nearly limitless budget (at least by 373 standards) to carry out your ideas. All of these projects can easily lead to publications in top tier research conferences, so if you're interested in graduate school, this would be an excellent way to get some visibility. 1. Smart badge. Design and implement smart badges that can be worn by people and collect human inter-contact data. A functioning system will capture any interactions that last longer than a few seconds between two closely-located (roughly 0-2 meters separation) people, including the distance separating them, using ultrasonic time of flight ranging. The system should be small (less than 2" x 2"), be able to run off a coin cell battery (250 mA-hr) for at least a few days, and cost no more than $25 in modest volume. It should include at least a few MB of NOR flash memory. A successful project could lead to a large-scale (1000-2000 node) deployment. The following paper provides some background on what we're trying to do but the goal is a smaller, less expensive, and more useful badge: http://sing.stanford.edu/pubs/hotemnets10-flu.pdf 2. Energy-Harvesting Energy Meter. Design a cubic-inch AC power meter that utilizes an energy-harvesting front-end. The basic ideas is to: (i) take an extension cord; (ii) split the cord's conductors; (iii) wrap a coil around one of the conductors, thereby making a makeshift transformer; (iv) rectify the output of the coil; (v) use a harvester chip: (e.g https://www.sparkfun.com/products/9946) to convert this to useful energy; (vi) use the energy burst to increment a couneter and transmit a radio packet with the counter. On the receiver end, average power is proportional to count/time, where time is elapsed time between successive counts. The hypothesis is that we can use this to measure energy consumption with more complex and costly electronics. 3. HiJack++. Port the HiJack system (currently running on a TI MSP430 processor and iPhone handset) to use the EnergyMicro TinyGecko ARM Cortex family, and have it conform to the hijack-infinity framework. Create a novel sensor board (e.g. radiation sensor). Useful links: http://www.eecs.umich.edu/~prabal/projects/hijack https://github.com/ab500/hijack-infinity 4. EyeSight: Sighting you eyes with computational glasses. Knowing what one is looking at has tremendous values in retail, real estate, advertising, product design, and other areas. In this project, you will attach two cameras to a pair of glasses, one facing outward (world-facing) and one facing inward (eye-facing). The goal will be to synchronize the image feeds from these two cameras, extract the location of the pupil, and mark where on the world-facing image a person is looking. There will need to be some training to correlate the images from the two cameras. This will require some familiarly with image processing tools (e.g. OpenCV) and low-power imagers.