z

Co-Aerial-Ecologist

UniversityofNebraska-LincolnPIs:CarrickDetweiler,AmyBurgin,Sebas4anElbaum,andMa8WaiteStudents:J.Higgins,J-P.Ore,A.Shankar,N.Sharma,K.Song,A.Taylor

UniversityofCalifornia-BerkeleyPIs:SallyThompsonandMichaelHamiltonStudent:M.Chung

Results

MotivationMonitoringandpredic4ngwaterqualityposesa significantchallengesincesourcesof freshwaterandcontaminantscomeinfromhugeareasoflandandwaterways.Further,thesourceofpollu4oncanchangequicklyduringandaMerrainfallevents.Characterizinglarge-scaleandquickly changingwater systems remains a cri4cal bo8leneck that inhibits understanding oftransport processes and thedevelopmentof effec4vemanagementplans to addresswaterquality issues. Fixed sensors tend to lack the versa4lity to directly detect contaminants ofinterest,areexpensive,andonlymonitorasingleloca4on.Consequently,thereiss4llastrongrelianceuponmanual"grab-sampling"withinhydrologicandaqua4cecologyapplica4ons.Atbest,thisrelianceisexpensive, inconvenientandpresentssafetyriskstopersonnel involved(e.g.when samplesmust be taken at night). Atworst,manual sampling results in datasetsthat cannot answer many ques4ons of interest due to limita4ons of temporal and spa4alresolu4oninthesamplingstrategyortheinaccessibilityofsites(e.g.canyons).

National Robotics InitiativeNational Institute of Food and AgricultureGrant #2013-67021-20947

•  Successfullycollected100sofsampleswithscien4sts.•  Validatedthataerialwatersamplingmechanismdoesnotbiasmeasurementsofcri4caldissolvedgasses(O)andions(P,Cl).

•  Developedrobustal4tudecontrollerfornear-waterflightandcharacterizedits’abilitytocollectsamplesinupto15knotwinds.

•  Developednewsensingmodali4esforthermalstructureandwaterconduc4vitygradientsthatarefastandnon-intrusive.

•  Createdcodeinstrumenta4ontooltoconnectuserinterven4ontoprogramparameters,sowhenauserintervenes,thesystemsuggestsparametersthatareimpac4ngbehavioratthatmoment.

•  Createdtooltoautoma4callygenerateunittestsfromtracefiles,makingregressiontes4ngfasterforsystemdevelopers.

Integration with static sensors

Sub-surfacesampling

Fremontlakes,NE,isoneofthemostpopularrecrea4onsitesinthestate.Sincetoxicalgaebloomsin2004,theyrequireac4vemonitoringandmanagement.

Aerialwatersamplerinfieldtest.

Comparison to traditional sampling

•  Sub-surfacesensing/sampling•  Robustheightcontrol(winds)•  Automatedtestgenera4on•  Robotdebugging•  Invasivespeciesmonitoring•  Characterizeradiopropaga4onoverwaterforUAV/underwaternetwork.

Carrick DetweilerComputer Science and Engineering - University of Nebraska Lincoln

carrick@cse.unl.edu

New Sensing Modalities

Future WorkWaterConduc4vityWaterThermalStructure

•  ThermalStructureobtainedbysamplingtemperaturethroughawatercolumn,enablinggreaterspa4alresolu4onofthermaldatasetswithoutextensivesta4csensorarrays.

•  WaterConducHvitymeasuredquicklyinhighresolu4onwithoutdisturbingsensi4veecosystems,helpingscien4stsiden4fyandunderstandcryp4csalinewatergradients.

Sub-surface Sampling

Robotic Actions System Traces Concrete / Abstract Tests

Automatic Unit Test Generation From Trace Files (ROSBags)

Regression Testing

0 50 100 150 200 2500

10

20

30

40

50

60

XY Distance (meters)

Z D

ista

nce

(met

ers)

Rad

io P

acke

t−Su

cces

s R

ate

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Further characterization of radio Propagation over water.

Robotic Water Sampling and Sensing in the Wild!

Nebraska Intelligent MoBile

Unpersoned Systems