The Glass Bottom Float (GBF) is a floating public robot with the mission of making the critical assessment of recreational water quality a transparent and participatory (across species) real time experience. GBF cruises along a beach shore, and offers itself as a resting spot in places it deems clean enough for swimming. Over time it maps paths of least contamination and highest relative pleasure for fish and people.
GBF assesses the current state of the waters with a three-tiered sensing system informed by best practices of recreational water quality assessment science: established metrics (algae, chlorophyll, dissolved oxygen and others), experimental metrics (statistics based near real-time in-situ e-coli, wave motion), and untested metrics (the presence and sounds of fish and crustaceans) are combined and compared with post swimming experience surveys to create a probabilistic, qualitative measure of water quality; the swimming pleasure measure (SPM). The platform is available to the public and to water quality professionals to cross-validate data and opinions. All results are public domain (a beacon blinks a color-coded pleasure measure visible from the shore) and available in full on the Internet. A subset of the data is available for mobile phones to give SPM locative agency, allowing for on-demand inquiry of swimming pleasures and discourse on water quality and our limits of understanding it.
GBF is intended as a conduit into new complexities that become apparent through patience via long-term measurement of resources shared with others and other species. GBF should be an object that W.G Sebald would have stopped and wondered about, had he wandered the shores of Western New York.
This is poetry and then some.
GBF is an engineering challenge!
Most of the hardware we build ourselves. We assess the wave motion with a sonar transducer on a double pendulum such that the sensors always points straight to the lake ground, even in the choppiest of waters. We combine local weather assessed with an on board weather station with 5km range weather patterns measured by NOAA. We make use of state of the art water chemistry assessment built by YSI (http://www.ysi.com/) and combine that with real time hydrophone input: what does it sound like – to ears underwater – when speed boat zips by?
Redundancy is built into the system on several levels. Each sensor runs on a separate process, programmed with python, such that failure in one sensor has no effect on the performance of the others. These sensor processes are monitored by a master process that controls the overall program flow and restarts the measurement sequence should it be terminated unexpectedly. Time stamped results are logged onto an internal SQL database and sent via mobile phone to a duplicate SQL database on a remote server. Update intervals are data dependent. The NOAA updates occur every 60 minutes based on NOAA’s own update rate, the YSI-sonde and local weather station occur every 15 minutes. The sonar transducer is queried every 30 minutes at one second intervals for 300 data points. This delivers a snapshot profile of the wave motion at that time. Together these readings give a rather fine grained description of important direct and indirect water quality properties.
GBF is research!
Our swimming pleasure measure is a metric formed by machines and people. We ask the man, woman, and child on the beach how they like the water and enter this into your knowledge system (mobile access). We combine this together with the hard data from our various sensors, NOAA weather readings, and local micro-climate data. The park manager enters his daily lab based e-coli measurements into the same system. With this data-conglomerate we train a neural network to “hold” this complex and sometimes contradictory knowledge. We are still experimenting with various network topologies. The larger goal is to find a computational approach to capturing people’s “opinions.”
In particular we want to gauge their knowledge, biases, and intuition about their environment, in this case the swimming experience. This soft knowing we relate to the hard knowing of the apparatus in order to derive knowledge of the waters neither people nor machines alone can have. This is a grand challenge in the computational sense of the term. It is related to the kind of computing von Ahn performs in his CAPCHTA systems and coins “human computing.” We are attempting to formulate something akin to “civic computing.” Von Ahn rewards his visitors with a gaming experience. We reward with a responsive resting spot in the water.
GBF is for the people!
We have an established community partnership with Woodlawn Beach State Park in Western New York and are assisting in the park’s efforts to identify and track the various sources of water pollution it is subject to. Our first prototype has been at the beach for several weeks this summer. Park manager Kevin McNallie (Kevin.McNallie@oprhp.state.ny.us) can be contacted with questions if required.
GBF is in your face!
(Well, on your phone and your computer). Part of the project is geared towards finding the best ways to give this rich data new agency. GBF output is available for mobile phones and for Twitter users. We make full use of the “instrumental internet” to keep you from driving to the beach when it is closed and are playing with text generators to go justice to the “expressive internet” and the elusive “co-presence” that micro-blogging can, at its best, achieve.
GBF is a collaborative effort!
Joe Atkinson (University at Buffalo) offers expertise in water quality assessment, Rolf Peifer (AILAB) expertise in embodied robotics, Mark Shepard (University at Buffalo) is the mobile phone master, and Marc Böhlen, a.k.a. RealTechSupport, runs the circus.
The project web page contains additional information and images: http://realtechsupport.org/new_works/gbf.html
Marc Böhlen is an artist-engineer, a maker of systems, situations and devices that critically reflect on the role of automation in the 21st century – in the widest sense possible. He is currently associate professor in the department of Media Study at the University at Buffalo and Visiting Artist at the AILAB of the University of Zürich. Böhlen’s research is tightly coupled to robotics design in methodology and succinctly different from it in scope and critical focus. It is an ongoing effort to diversify machine culture. Signal processing, artificial intelligence and control systems are cultural artifacts inscribed by those who create and use them in similar ways as more traditional media are acknowledged to be.
Recent research is centered on forms of mixed knowing, the combination of ways of knowing from different ontological and perceptual frameworks. The Glass Bottom Float project presented at ETech2009 attempts to combine human and machine knowing to formulate knowledge of water conditions at a public beach neither people nor machines alone can understand. Recent work has been presented at Cynetart (Dresden 2008), Dorkbot (Toronto 2008), and Satellite Voyeurism (Dortmund 2007). Recent and upcoming publications include Robots with Bad Accents; Living with Synthetic Speech (MIT Press 2008), Second Order Ambient Intelligence (JAISE 2009), Ambient Intelligence in the City (Springer 2009), and Micro Public Places (Architectural League New York 2009). See www.realtechsupport.org for details