Atis research ideas

From DiLab
Revision as of 15:10, 12 October 2011 by Atis (talk | contribs)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search

Energy harvesting from high-voltage power lines

The idea

Use EM field radiation from high-voltage power line as an alternative energy source for powering low-energy sensor motes.

What problem it solves

Energy harvesting in open-air environments with nearby high-voltage lines, especially forests etc. (low sunlight intensity, far from easily accessible power sources, battery maintenance can be difficult)

What is needed

  • Inductor /indukcijas spole/, for harvesting the energy from EM field
  • Supercapacitor, for storing the energy

EDI has already performed research on a related topic - high-voltage line EM field intensity measurements (Aivars, Gatis)


Precise "time of flight" localisation

The idea

Use precise timing on motes and UWB or similar highly time-localised impulses to achieve high precision localisation by measuring the time it takes from radio signal to travel from one mote to another. (The so-called "time of flight localisation".)

This could be more of a "theoretical" direction of research. It could be an interesting research, even if the the initial solution turns out to be too costly for immediate practical applications.

What problem it solves

At the moment, time-of-flight localisation solutions in WSN have 1+ meter error margins. This can be unacceptably high for various applications, e.g.

  • when the network is geographically small (i.e. the distance between motes is in the order of centimeters instead of meters. Think of "smart dust" as an extreme version of this.)
  • for robots and mobile actuators in general
  • for augmented reality /precizēt, kam tieši/

Higher precision of localisation can be achieved by using ultrasound instead of radiowaves. Less precise timing is required in this case because sound travels many orders of magnitude slower. On the other hand, sound based localisation is not suitable for:

  • outer space environments
  • noisy environments
  • /probably can think of other reasons as well/
  • /maybe/ it takes more energy

What is needed

  • A few (e.g. 3 on the network) motes with high precision UWB impulse receivers
  • High precision UWB impulse senders for each mote in the network (because sending is easier than receiving)
  • A mechanism which allows to incorporate data in the UWB impulses (?)
  • Precise clocks and a very precise network synchronisation protocols (the details are not clear yet)
  • Triangulation algorithms etc. (?)

All signal generation & reception + registration should be done in hardware, because adding software components can cause unpredictable delays. (The unpredictability, rather than the length of the delay is the problem.)

EDI has already performed research on related topics - UWB impulse generation (Gatis, Jānis), using UWB for data transfer (?), high precision time measurements (Artjuks).

Note:

  • c = 300'000 km/sec (approximately)
  • For <= 1 m localisation error, minimum sample rate is 300 MHz (radiowaves travel 1 m in 3.33 nanoseconds)
  • For <= 1 mm localisation error, minimum sample rate is 300 GHz (radiowaves travel 1 mm in 3.33 picoseconds)


Position tracking

The idea

Use IMU (and optionally GPS) to track athletes e.g. on ski race track.

What problem it solves

Athlete tracking for fun & profit, using cheap, off-the-shelf equipment. It allows the athlete to analyse his performance in details and compare himself to other athletes - useful both for amateurs and, if high enough precision is reached, also for professionals. It is an improvement over simple chronometry + video recordings, because:

  • much easier to setup and use than multiple video cameras
  • usually allows higher precision
  • automated analysis and comparisons are easier

This has been done before, but most likely not in Latvia.

What is needed

  • a simple or not so simple IMU system, i.e. accelerometer + gyroscope. The exact requirements on hardware for precision are not clear yet
  • RTK GPS (a.k.a CP-GPS) in the ideal case (but almost certainly not for the first experiments, unfortunately - too costly)
  • a custom algorithm to get the most out of gathered data. Most likely the algorithm will be based on Kalman filter


Floating sensors (with Reinholds Zviedris)

The idea

Drop sensor is a waterproof box in a river.

What problem it solves

  • Measuring water pollution
  • Measuring water (i.e. river) speed
  • Measuring water opacity. (This parameter also can be useful for determining the pollution level).

What is needed

A small mote with a radio or an active RFID tag. A small, round, tightly sealed box (with a "tail" so that the mote does not spin around constantly, but is aligned with the flow of the river.) The density of the box should be roughly the same as the density of water. (It should move with the flow, not float above it).

Research challenge: make a box that has variable density. How to achieve this without using an engine or other complicated, energy-wise expensive solution?

A project using floating ECO motes in a waterproof box: PipeProbe: Mapping Spatial Layout of Indoor Water Pipelines

Artificial fish - ? (There are many videos on YouTube etc.) Research challenge: make it small and energy eficient.


Light sensor calibration tool

The idea

Make an universal tool for measuring light with specific wavelength(s).

What problem it solves

The need to accurately measure photosynthetically-active portion of the light. (I.e. measure the light that plants can use.) This is needed for agricultural applications of sensor networks (e.g. SAD).

Using this tool, light sensors could be calibrated correctly. At the moment precise calibration is impossible. The cheap luxometer we have has light wavelength response curve that apparently is completely different from the one we need.

What is needed

  • A number of light sensors. (The more the better. I think, at least six are needed.)
  • A mote / embedded board to which all of the sensors can be attached at once.
  • Light filters for specific wavelength (e.g. strips of colored film could be used)
  • A board / enclosure where to mount the system. Preferably with integrated level tool /līmeņrādis/
  • Software that reads all the sensors (almost) at once and applies different weights to specific wavelength reading. The weight vector used depends on the light we need to measure. E.g. there can be one sensitivity curve for apple trees, another - for cherry trees.


EEG for fun

The idea

DIY EEG (electroencephalogram) tool.

What problem it solves

  • Fun toy for demonstrations etc.
  • Neurofeedback for meditations.

What is needed

  • Computer interface
  • Optionally - some actuators
    • E.g. cat ears (Google / YouTube search for Necomimi to see an example)