OVERVIEW

There is a need to provide a platform for better monitoring and sampling in Marine environments. Such a platform should be able to withstand the highly dynamic nature of such an environment as well as cope with its vastness. The platform should be simple and easily scalable. A platform of this type would provide the scientists an invaluable tool in order to further the marine research by monitoring phenomena of biological importance. As part of our research, we are building a fleet of autonomous roboducks (robotic air boats) for in-situ operation (data collection and analysis) in marine environments. The platform would support a variety of sensor suites and at the same time be easy to operate. It can operate in both exploration mode and intelligent mode. It can also collaborate (via communication) with other entities (sensor nodes) in the local neighborhood making intelligent decisions. The roboduck fleet will serve as a test bed for evaluating algorithms including bacterial navigation for marine sensing and adaptive sampling.

APPROACH

Our research focuses on development of an autonomous mobile platform (robotic airboat) for in-situ operation (data collection and analysis) in dynamic marine environments. The platform should support a variety of sensor suites and at the same time be easy to operate. It should be able to collaborate (via communication) with other entities in the local neighborhood making intelligent decisions. The test bed will also serve as a test bed for evaluating algorithms including bacterial navigation and adaptive sampling. There are several advantages of having a mobile entity. These include:

• The exact location of phenomena need not be known apriori.
• o The phenomena being observed need not be static (at a fixed location).
• Can serve as an explorer and data collector (data-mule).
• It can go to a desired location and collect sample(s) as and when needed.

It can collaborate with other statically deployed networks.

• The static network can guide the mobile node to areas of interest where the static nodes cannot go themselves, improving coverage.
• The static network can use the data obtained from the mobile node’s sensors to refine its own neighborhood estimates.
• The mobile node can move the static sensors around as and when needed.
• Can serve to fill the voids between the sparsely located static nodes (in geographically sparse static nodes).
• Can help bridge voids between portions of static network that might become disconnected at times or help bridge several static/mobile networks.
• The mobile node can have the costlier equipment (cost/maintenance/availability and other constraints) which can not be placed on the static buoys (which are significantly more in number).
• It can go to a desired location and collect sample(s) as and when needed.

It can collaborate with other roboducks

• Presence of multiple roboducks in an environment can result in collaboration between the nodes and better performance.
• Can help track phenomena of interest in different regions at the same time.

DESIGN

The Roboduck is a mobile platform developed as part of the marine micro-organism monitoring project under CENS. The platform serves as a test bed for evaluating the algorithms for marine in-situ monitoring.

As part of our research, we are interested in monitoring both the surface and sub surface phenomena and analyze the generated data in real time as this has a lot of pertinence for marine biologists.

The architecture of the roboduck is shown in the Figure.

We chose an airboat for our work. Our choice of the airboat based design was a consequence of this requirement as we did not want the surface water to be disturbed a lot by the boats propulsion system (see figure below).

All the modules and sensor suites have been integrated and connected to the main processor board via the RS-485 bus. This makes plugging in additional modules very easy and convenient without affecting the existing modules.

The various modules of the architecture include:

• Hull design: The first design of the hull of the airboat was based on the Dumas Big Swamp Buggy Kit 31”. The pontoons were later extended so as to provide increased buoyancy so that the boat could handle increased payload.
• Processor: We chose the Intel’s stayton board as our processor as it had low power consumption and we had limited power we could provide for processor board. Later we moved to Intel’s stargate board which had better power management and even lower power requirements. It provided the necessary I/O and network interfaces. The stargate board was also smaller in physical dimensions (3.5” * 2.5”) making it a better choice for the roboduck.
• Additional control on each module was provided using Parallax’s BS2sx basic stamps. These connected each of the modules to the stargate board via the RS-485 bus. Having smaller individual processors to control the modules made it convenient because the higher level controller on the stargate could then be made independent of device specific control. The basic stamps took care of lower level sensor/motor control while the stargate board could issue high-level commands (like collect sample).
• Navigation module: The navigation module consisted of a Garmin 16A gps unit and a PNI V2Xe 2 axis digital compass. Together they accurately provide the current location and heading of the roboduck at a given instant of time. The information received from the gps and compass is used to generate the appropriate guide and steer commands for the boat. These high level commands are sent to the basic stamp on the navigation control board for issuing the appropriate low level commands for controlling the air propeller and rudder control servos.
• Sampling module: We developed a 36 port round valve sampling system. It was designed with a view to collect a large number of samples before having to return and leave the samples on the shore for lab analysis. One of the ports was attached to the fluorometer for in-situ sampling and real time in-situ analysis. The sampler is controlled via a basic stamp which provides the interface between the stargate board and the sampler through the RS-485 bus. We also have a smaller 6 port sampler which utilizes the same interface as above but with reduced sample payload.
• Sensor module: The current sensor suite consists of an array of thermistors for sampling the temperature at different depths and a fluorometer that can measure the concentration of chlorophyll-A. These are connected to a circuit board consisting of a 12bit 8 Channel ADC (Texas Instruments ADC 7828) for connecting the thermistors and a 16bit single channel (Texas Instruments ADC 1100) for connecting the fluorometer. The ADC’s are connected via an I2C bus interface to the basic stamp which controls the low-level operation of the ADC’s and serves as a relay between the RS485 bus (connecting the other modules) and the sensor suite. We are using the Turner Designs Cyclops-7 submersible fluorometer as part of the sensor suite for detection and estimation of chlorophyll A.
• Communication module: The roboduck can communicate with other network entities and base station(s) via the 802.11b wireless connection.

Image of the 6 port sample collector

Circuit board interface for the sensor module

The roboduck can operate in one of the following modes:

• Autonomous navigation: The roboduck can be programmed to navigate autonomously and track phenomena of interest (information about which needs to be provided to the roboduck apriori) without any other information (Example command: “Go and locate phenomena of interest.”) (Algorithm based autonomous navigation and sampling).
• Network guided autonomous control: The roboduck can be programmed to navigate to places of interest as detected/suggested by other nodes within the network (Example command: “Go to this location and sample and send me information.”) (Network guided sampling).
• Autonomous control: The roboduck can be pre-programmed to visit a set of pre-programmed GPS waypoints and it can then visit these in that order autonomously (no manual control) (gps way-point automated sampling).
• Manual and RF control: The navigation of the roboduck can be controlled manually either be sending it navigation commands over the 802.11b wireless connection or through RF transmitter (human assisted sampling).

SYSTEMS / EXPERIMENTS

We carried out several sets of field experiments with the roboduck. We carried out field tests at Shelter Island, NY. It gave us the perfect environment to test the boat by providing the exact conditions the roboducks were designed to operate in. We carried out three sets of experiments with the boat.

• RF control – to understand the boat dynamics, speeds it can move about etc.
• Manual control – to reliably control the boat using a computer
• Autonomous – execute a pre-programmed navigation routine of navigating in a circle and rectangle.

PEOPLE

FACULTY

Prof. Gaurav S. Sukhatme
Prof. David Caron
Prof. Aristides Requicha

STAFF

Carl Oberg
Beth Stauffer

STUDENTS

Amit Dhariwal
Bin Zhang
Eric Shieh