Sampling our
environment can be a significant challenge using traditional methods. In most
cases, the area to be sampled is large and inaccessible. When natural or
man-made disasters happen, the ability to respond to these areas can be a
challenge. Even when the areas are accessible, it takes time to obtain the samples
and then have them analyzed. The ability
to revisit the sampling sites in many cases is impractical or cost prohibited. The
use of small unmanned systems opens an opportunity to meet these challenges in
a profound way.
The ability to
autonomously take water samples using multi-rotor unmanned aerial systems by
obtaining spatially separated, high frequency water samples of the environment
can make significant impact at a relatively low cost and effort. It can provide scientists, environmentalists,
and governments a quick means to assess environmental areas before or after an
incident, and make decisions that will have an immediate impact. Adequate resources can be provided once the
information is processed, so that resources are not diverted to areas not
impacted by an incident. As time progresses, the sites impacted by an incident
can be revisited to monitor the long ill effects.
Ore and his team
developed an autonomous unmanned aerial vehicle to take water samples of rivers
and lakes. The vehicle is capable of
taking three 20 ml samples per flight. To do so, they developed and integrated
a hardware and software solution, which allows control of the vehicle at low
altitudes, less than one meter above the water.
Their solution is built on a Ascending Technologies Firefly aerial
vehicle. It includes a water sampling mechanism with three chambers for
sampling, a plastic tube hanging below the aircraft, and an attitude estimation
solution utilizing two ultrasonic sensors and an Kalman filter based software
solution that estimates the altitude of the vehicle above the water.
The Kalman low attitude
estimate has three inputs. The first is
the barometric pressure reading from the internal sensor in the autopilot. The other two are the ultrasonic
sensors. As the vehicle lowers to the
predetermined sampling site, the Kalman filter estimates the attitude of the
vehicle once the vehicle is below two meters based on the barometric reading. The estimator will then provide the best
estimate so that when the aircraft is one meter from the water a sample
can be taken. The ultrasonic sensors are pre-filtered before the Kalman
estimation to reduce the risk of false inputs to the estimator. This allows for a quicker convergence on a
good estimation.
The aerial
vehicle can repeat the sampling process at three locations before returning to
it launch site. The samples can then be
analyzed at the site or in a laboratory.
Depending on the distance from the launch site to the sampling location,
this can be done quickly and repeated often.
To put it perspective, it would take longer to put a boat in the water, than
it would take to fly the first set of sampling missions for one vehicle. By the time the boat takes the first sample,
the aerial vehicle can take multiple samples at various locations. This is an efficient use of resources as a
fraction of the cost.
There is additional work to be done with their vehicle. Outdoor trials in varying weather conditions are still needed. The effects of wind, currents, and wave actions will affect the vehicle's ability to take an adequate sample. In some cases, the use of the vehicle may become impractical. However, there are numerous application where it does make sense and where these multi-rotors will have an immediate impact.
Ore, John-Paul, Sebastian Elbaum, Amy
Burgin, Baoliang Zhao, and Carrick Detweiler. "Autonomous Aerial Water
Sampling." Springer Tracts in Advanced Robotics Field and Service
Robotics (2015): 137-51. Web.
