research
current work (mit / whoi)
my present work is designing systems with autonomous vehicles (ocean robots) to pursue a number of scientific objectives. i am currently working with autonomous surface craft (the SCOUT) and autonomous underwater vehicles (AUVs), primarily the bluefin 21. i work with the MOOS-IvP autonomy architecture developed by p. newman and m. benjamin (see software page)
acoustic communications on autonomous vehicles: robust and relatively high throughput communications between vehicles is essential for multiple node collaboration (for a number of widely different tasks ranging from target tracking to geological surveying).
we are making use of the woods hole micromodem. the maturity of the hardware allows us to investigate ways of increasing the communications performance by intelligent and adaptive message encoding. furthermore, we model the acoustic field (by taking measurements of temperature and salinity to derive a sound velocity profile) in the current operating region. using this model, we seek to implement behaviors that allow the vehicle to seek out regions of high acoustic communication throughput. communications quality does not map onto transmission loss (i.e. loud regions do not necessarily mean better signal quality) largely due to echoes off the seafloor and sea surface.
adaptive oceanography: i am working to develop systems that use multiple autonomous vehicles to accomplish useful oceanographic measurements with minimal input from a human.
- an oceanographer tasks the system on a broad mission (e.g. profile the physical structure of a given bay).
- the system, which may include shore computers as well as autonomous vehicles, must develop a test plan (e.g. take measurements of temperature and salinity at positions [x y]) perhaps taking some a priori knowledge of the scientist into account.
- it must execute the plan, and adapt measurements based on current data to perform the most useful measurements with the least resources (fuel, time, etc.) possible.
{ presentation: MTS/IEEE OCEANS 2008 / Quebec }
cluster autonomy for target tracking: single vehicles, regardless of the sophistication of their sensors, are inherently limited in their ability to detect and track multiple targets. a target could be a whale or a terrorist submarine, for example.
thus, in these situations, a cluster of vehicles working together to localize and track targets over a region of ocean can be useful. while even a single robot can be tricky to program, a new set of challenges emerge when multiple robots must work together (think of preschoolers first interacting with other children).
we have developed and are studying a "zone defense" (think basketball) system of target tracking that involves multiple vehicles collaborating without sharing information beyond their positions, as network bandwidth is extremely limited in the ocean environment.
{ video: 4.27.08 laboratory meeting - supplement to talk [13.0 MB] }
ancillary projects and tools: i have developed and maintain an ocean vehicle viewer (geov) based on google earth, a 3D high resolution mapping program that allows for custom geographic data.
past work
williams college, undergraduate honors research in physics (2006-2007): precision diode laser spectroscopy of thallium towards the measurement of time reversal-violating but parity-conserving (tope) forces. (p. k. majumder)
{ thesis: precision phase shift spectroscopy in thallium [4.5 MB] }
{ poster: precision spectroscopy in thallium }
williams college, undergraduate summer research in physics (2005): investigations of picosecond soliton pulses in an all fiber laser (j. strait).
connecticut agricultural experiment station, summer research assistant in the department of plant pathology and ecology (2004): quantifying aerial dispersion of maize (zea mays) pollen towards a model of temporal and spatial separation necessary for complete isolation of genetically modified (gm) maize. (d. e. aylor).