Dr. Douglas Nowacek (Electrical and Computer Engineering & Marine Science and Conservation Division)
Duke Nicholas School of the Environment Entrepreneurship Fund
Dr. Douglas Nowacek Lab
National Geographic – Early Career Grant (Technology)
The desired outcome of this project was to produce an open-source, depth-rated, modular tag that reliably and accurately records and stores data. The tag should be a stand-alone device that is encased in a dense plastic and has an attachment mechanism for both the tag body itself and various sensor modules. To eliminate the risk of water intrusion, the entire tag will be encased with wireless inductive charging and WiFi data offload. By the end of summer, a basic IMU and RPox sensor will be integrated into the tag unit.
Ultimately, the tag would be used by researchers who are either unable or unwilling to pay for the high cost of existing tags.
An initial prototype has been developed using an Arduino microcontroller. This prototype includes an accelerometer, magnetometer, gyroscope, pressure, and temperature unit, and a particle sensor unit for pulse oximetry. The team has now translated the Arduino system to a Texas Instrument system which proves to be more robust to handle high sampling rates and is commonly used in the large-scale production of tech tools.
The Texas Instruments based system is currently set up using individual breakout boards for each capability we are looking for on the tag. From this, we have proof that the Tag Operating System (TagOS) indeed can work to drive every function of the Tag. This now is now in the PCB design phase. The PCB we manufacture will include not only all of the sensors, but also the power system, data storage, and inductive charging.
An initial shell for the casing has been 3D printed and tested using several types of suction cups supplied by 1900 Engineering Inc. The suction cups were inserted into the printed tag body and tension tests and shear force tests were performed with the tag suctioned to both pseudo-whale skin and pseudo-shark skin. To model whale skin we used the curvature of a bucket to mold a layer of foam, representing blubber, and a layer of flooring underlayment to represent the dermis and epidermis. To model shark skin we used too different rough grits of wet/dry sandpaper molded against the curvature of a bucket. Although our tests with these models were mostly quantitative, much was learned about the adhesive properties of different suction cup materials on both types of skin.
Overall, we are in a good position to commercialize this as a product within the next 6-12 months.
Hardware (personal focus)
Hardware in the context of this report is referring to the electrical components that run the tag. There are four major assemblies within the tag:
- Power (wireless charging, battery management, voltage control, wireless switch)
- Microcontroller + Non-volatile
- Communication (WiFi -data offload, Long Distance RF – recovery)
- Sensors (IMU, Pressure, GPS etc.)
These assemblies must also consider physical constraints. For example, the pressure sensor must be in an oil-filled section of the housing, the GPS must have signal and therefore have a connection to the top of the tag, the reflective pulse-oximetry sensor must be at the bottom of the tag (near the animal).
The initial prototype device was powered by an Arduino Uno paired with a Sparkfun MAX30105 breakout board for doing reflectance pulse oximetry pilot studies with captive bottlenose dolphins.
Phase II focused on identifying the electrical components that will be used in the final tag while still on a physical breakout board. Due to limitations with Arduino, Texas Instruments’ MSP432
The next iteration of the prototype is being developed with the Texas Instruments MSP432 microcontroller platform. For the prototyping and software development phase, the TI MSP-EXP432P401R Rev 2.1 LaunchKit board is being used. The LaunchKit is wired into a number of breakout boards, including:
- MAX30105 (RPox Sensor)
- MPU-9250 (9-DOF IMU with 3-axis accel, mag, and gyro)
- BMP-280 (Pressure)
- Temperature sensor
All breakout boards use I2C, with the MSP432 LaunchKit acting as the master I2C device.
The focus of Phase III is developing a commercial-grade electronics package. Once the Phase II hardware design was completed with stable software, a PCB (Printed Circuit Board) design began. This is desirable as it is infinitely more reliable and consistent in manufacturing than a protoboard.This process is somewhat iterative as changes continue to occur in software (affecting Phase II/III hardware).
Autodesk EAGLE was used to first complete circuitry schematics (See appendix for circuitry schematics), which in turn allows the physical layout of the PCB to be completed (see right). The layout of the circuitry is relatively complex and takes time.
NOTE: This stage is still in progress and requires a lot of iteration
Figure 1: Power Layout
Figure 2: Sensor Layout
In the past few weeks, we have been working on potential ways to attach the reflectance pulse oximetry (RPOX) printed circuit board (PCB) to the tag body housing. We began by sketching a few attachment designs and narrowing those down to the four designs (an elliptical cylinder design was also added, very similar to the circular cylinder design). Then, to begin modeling the hydrodynamics of the designs, we started using COMSOL Multiphysics to run simulations. For example, we ran a 2D simulation looking at velocity and pressure contours under laminar flow conditions comparing circular and elliptical cylinder designs. After using COMSOL, we decided to switch to using AutoCAD Inventor as it is more efficient for turbulent models which shows a more representative condition of the tag when secured on a fast-moving animal. We then ran a series of CFD tests for all the proposed designs. With asymmetrical designs, we ran different models for specific directions of flow to consider different potential applications of the tag. The direction of flow is shown by the long blue arrow.
The overall tag housing is also in development with several shapes in testing using COMSOL and Inventor. An example of a proposed design is shown below. For this component of the tag, it is important to think about the placement of the tag on the animal. If it is possible to get our desired alignment then we would hope for a shape similar to the examples above. We have discussed these designs with individuals at the Duke Marine Laboratory, who are well-versed in tagging large cetaceans, and they have shared that is often difficult to align the tag perfectly along the spine. For this reason, and also because most megafauna do not swim in perfectly straight lines, we have recently been leaning towards a symmetrical design that will sit low to the body and not be limited by a
The prioritized form of attachment is using suction cups on the underside of the device, and various forms of release mechanisms are being experimented on. Currently, the release mechanism will likely use nickel chromium or magnesium burn wire, as it dissolves in salt water, or tungsten wire which can heat up from the inside circuitry. The final material of the device will also be decided upon in field testing – we are leaning towards a polycarbonate or urethane material, or a fully epoxied device.
Figure 3: CFA Analysis
A basic real-time tag operating system, known as tagOS, is being developed in TI’s Code Composer Studio 7.x, using C and C++.
It is important that power is conserved as much as possible. A series of system modes have been created to optimize power consumption and extend working life of tag.The task flow can be seen in the figure to the right.
The software is robust, even if an unexpected data package comes from one of the sensors. This ensures that data packages sent to the non-volatile memory are not corrupt even under exceptional
circumstances.Within the software, absolute priority has been given to data collection to ensure less-important tasks do not compromise experiments.
Within the software, absolute priority has been given to data collection to ensure less-important tasks do not compromise experiments. It is possible to export the collected data in a .tsv file format for easy analysis in MATLAB, R or Excel.
The current stable working branch of the OpenTag tagOS repository can be found at the Duke Conservation Tech’s FinBit_MSP432 GitHub repository.
NOTE: This software is still an alpha state – however, it is stable and able to control tag as needed.
With results from field testing in Hawaii on bottlenose dolphins and a more recent trial on an Atlantic Sharpnose shark suggesting a different design for the RPox module, the module is in the design and
fabrication phase. This new iteration will potentially be tested on mantas in Indonesia under a National Geographic Early Career Grant and Dr. Andrea Marshall, as well as with a Marine Mammal Foundation Scientist on whale sharks in the Philippines as early as this spring. The Atlantic Sharpnose shark was tested in conjunction with the UNC Shark Census Program.
This project has some good momentum and will continue under the Duke Conservation Tech student group. A working prototype will be finalized during the fall semester, with the aim of testing with the Marine Megafauna Foundation on whale sharks in January 2018. During the fall, testing will take place at the Duke Marine Lab. A pending application for a National Geographic grant may accelerate the project further.
Ultimately, after sufficient testing we hope to have a v1.0 system ready for commercialization by mid-Spring 2018.
Ashley Blawas (BME ’18, MSCL Certificate)
Henrik Cox (ME ’18, Visual Arts Minor)
Carmen Hoyt (EOS & Biology ’18)
Sam Kelly (ME ’18, Environmental Science Minor)
Dave Haas (Ph.D. Student, Marine Science)