OpenPlant funded a team of eight engineers and scientists to start the LunaFlow project. The aim of the project is to study the behaviour of a strain of bioluminescent organisms, so-called dinoflagellates (Pyrocystis Lunula), in their application to the visualisation of aqueous fluid flows.
Dinoflagellates naturally respond to rapid variations of tension that occur within a flow field by emitting visible blue light. These light emissions are rare sights in the natural world, but not uncommon to the United Kingdom. Dinoflagellates blooms are occasionally reported during summer nights along the British and Welsh coasts and exhibit wondrous glowing patterns as the organisms are pulled and stretched by waves breaking against the shorelines.
The evolutionary reason behind their light emission remains unclear but is often attributed to be a defensive mechanism against predators. The biophysics of the light emissions are also mysterious. They are closely tuned to their circadian cycle and indeed, dinoflagellates require extended periods of darkness to shine. They can be trained to adapt their bioluminescent responses to twelve-hour shifts of light and dark and the team exploited this property to use them in the laboratory.
The dinoflagellates’ natural piezometric responsiveness and relatively small size, c. 20 m, offer an attractive opportunity for their use as a flow tracer. Traditional techniques that measure piezometric properties (albeit pressures, strains and forces) use instrumentation such as Pitot tubes, pressure transducers or force balances. They are recognised to be intrusive (disruptive) to the observed flow and are limited to a local measurement. Artificially lighted particulate flow tracers, on the other hand, are often used to simultaneously measure velocity fields that exhibit complex spatial variations.
Measurement of velocities is achieved using image processing algorithms based on either optical flow techniques, particle tracking velocimetry (PTV) or particle image velocimetry (PIV) techniques. Dinoflagellates offer an opportunity to modify these techniques and measure simultaneous and spatially-distributed piezometric properties. The team set out to assess and demonstrate the feasibility of this application.
The project
The LunaFlow team set itself a two-fold challenge for the competition: to develop (i) a bespoke low-cost incubator to grow the organisms and (ii) a three-dimensional low-cost camera system to study them. The team was composed of three mechanical & civil engineers, two data scientists, two chemists and a biologist. The Biomaker Challenge provided the opportunity for the team to meet and form the group and the support of the OpenPlant organisers provided a technical and financial platform to test and nurture the idea.
Outcomes: Incubator
The incubator design was led by Dr Duncan Scott and was a sheer collective effort of design, build, wiring and testing. The volume required for the incubator was significant, c. 0.2 cubic metres (200 litres). The design was improved through several iterations of modelling and prototyping. Cooling was a major challenge due the incubator volume and the team tested both an air Peltier heat pump system (successfully in a small 20-40L volume, but not so, in a larger 100-200L one) and a mixed water-cooling air-heat-dissipation Peltier heat pump (with improved performance).
The temperature control system was actuated using the Biomaker OpenSmart Arduino board and sensors provided in the competition kits. The kits allowed the team to rapidly develop algorithms and test them for rapid deployment. The Arduino system was set up to communicate by serial communication to a Raspberry Pi Zero in order to feed a remote monitoring system. The control of the remote monitoring system was set up using a ThingSpeak channel that continuously streamed temperature data and sent out email-based emergency alerts. Fully detailed description and instructions for the build are documented on our Hackster page.
Outcomes: Camera system
The camera system design was led by Dr Francesco Ciriello. The primary objective for the system was to create a synchronised multi-camera system with low-cost hardware that was scalable to use with many cameras. Increasing the number of cameras improves the potential of better resolving the three-dimensional structures within the observed flows.
The hardware architecture works with different commercially available hardware boards (tested with Raspberry Pi 3B+ & 4 and NVIDIA Jetson Nano boards). Communication between devices is set up over a private network that runs on a local DHCP server and uses an MQTT protocol as middleware for synchronised acquisition. Three-dimensional reconstructions were implemented using algorithms from the MATLAB Computer Vision Toolbox and set up so that the cameras automatically calibrated their extrinsics using feature-based registration. The team developed a full end-to-end workflow for acquisition and packaged it into a suite of MATLAB apps that can be executed either in MATLAB desktop, online environments or as a
standalone application. The concept worked well, and the team is now looking at how to expand the middleware to use MQTT within ROS-based architectures. Software and examples are released on GitHub.
About the author
Following the Biomaker Challenge, Francesco Ciriello moved from his Postdoctoral position at Cambridge University Engineering Department and joined the MathWorks Education Customer Success team. He now travels the United Kingdom promoting better teaching practices in higher education. The experience from the competition makes him a firm promoter of reverse classroom approaches based on project-based learning and he whole-heartedly recommends it to all students and staff.
Acknowledgements
Special thanks and congratulations to all LunaFlow team members: Duncan Scott, Shivani Maharaj, Karla Cervantes Barron, Alessandra Luna Navarro, Edoardo Gianni, Nicholas Wise and Fernando Guzman Chavez.
Special thanks to Biomaker organisers Jim Haseloff, Alexandra Ting and Dieuwertje van der Does.