About a week back, I went to my first hackathon and this post is about the experience I had there. I did not know much about how it feels to be in the hackathon, working through the night to find a solution. So, when an email came through my mailbox about the #2019FireHack I told myself… that is exactly the kind of thing I'm looking for! I will have something exciting to talk about with my students, and perhaps, I can learn something new and meet new people.

Currently, I am reading through a Ph.D. in Robotics and working as a Tutor for the Computer Science module at the DCAD Foundation Programme. One of my responsibilities is to create and supervise projects that the students develop during the academic year. This year some of my students are working with drones and we are investigating autonomous flights, their landing/taking off and how they can be flown with low visibility as in the case of fog or smoke. That being said, a hackathon focused on assisting emergency services to deploy intelligent drones capable of identifying and mapping wildfire was definitely a match.

The hackathon was organised by the DSTL (UK Defence Science and Technology Laboratory) in partnership with AFRL (US Air Force Research Laboratory), and it was running simultaneously to another hackathon in Ohio at the Wright Brothers Institute. A massive panel with live videos of the U.S. teams kept us informed on how the work was being conducted on the other side of the Atlantic.

During the first challenge, we had two wildfire spots that expanded over time. Additionally, we had two fixed-wing UAVs and two rotors to work with. Our team survived the first round, winning a very tight third place. Moving on to the second and third challenges, it was natural to observe that the locations in which the wildfires appeared had an element of randomness requiring a more creative geometrical approach for navigation. That reminded me of the work of Barnali Das who talked me through the navigational patterns of honey bees and how she used it to control swarms of robots.

I had the pleasure of meeting Barnali at the TAROS18 conference and inspired by her work, I started reading about honey bees’ navigation, but until now I could not see how this could be implemented in my own work. For me, the best thing about working with the AMASE software is that it removed the computer vision element leaving me with just the sensor readings and the AI to work with. It was then that I could see how the honey bees’ navigational patterns could be applied; not just honey bees but loads of different biological inspired navigation patterns could be developed and learned by the Deep Neural Networks to improve detection and mapping in areas of low visibility.

Another positive impact the hackathon had in my own research was the opportunity to network with bright minds such as Sam, a colleague from Southampton University. During the hackathon, Sam had developed an algorithm that merges overlapped mapping data and significantly reduces the number of points required to define the wildfire area. This not only speeds up the mapping process but also reduces fuel consumption since the UAV does not need to review the less significant areas. Because of the hackathon, we are now working on the integration of his algorithm with my current work that uses Deep Neural Networks to control the 6DoF of UAV when flying in unstructured environments.

YouTube demo at https://youtu.be/KWNNtGlBvEY.

Looking back, I feel that there were key elements that came into place to make this hackathon such an extraordinary experience. Starting with the location, the Spark building (Solent University) is breath-taking and walking to the DARPA (UK Defense Advanced Research Projects Agency) meeting inside the pod felt a bit like getting into the Enterprise. The meeting itself was really exciting. We had clear guidelines on how to take our ideas out of the paper and contribute to something that has a massive social impact.

The overall atmosphere of the place was vibrant; we had an entertainment area with VR, a giant tower of red cups and a game which I have no clue what it was about and did not hold my attention long enough for me to comprehend the rules. However, my real challenge was having to refuse to abandon my teammates during the hackathon while my true wish was to run upstairs to code a swarm of DJI TELLOS. You see, TELLO is the cutest UAV in the market and they had a swarm! Naturally, I was literally dying to play with them. One thing I cannot forget to mention is the playlist, which was spot on; the best selection ever! When Katy Perry started singing Fireworks, my heart beat increased and this kept me coding for the rest of the night with a smile on my face.

This brings us to the final question: why write about a hackathon?

The answer is because you should be there too. Wildfires affect us on a global scale, and not too long ago, the Kineta (Jul 2018) wildfires had claimed the lives of dozens. Even places such as Sweden and the Arctic Circle had suffered from the devastating consequences of wildfires. Building algorithms that improve mission planning and are capable of reacting to the dynamics changes have a significant impact on the deployment of future missions to combat wildfire worldwide. To contribute towards these solutions, you don’t need to be a geek or an expert in coding to join a group. Your ideas are as good as any coding skill anyone of the hackathon participants could have. So, listen to Katy… she knows best, and keep an eye on the next hackathon. Don’t miss out :)