(With more to come with the winners being announced)
On Friday 18 March 03:00 PST | 06:00 EST | 10:00 GMT | 15:30 IST | 21:00 AEDT, Hackmakers will announce the winners of the #FormulaAI Hackathon 2022. It will be an exciting moment to conclude the event. Stay tuned at https://www.formulaaihack.com/ to watch the public live stream.
It’s been an immense learning experience for many people (including myself). Here’s a snapshot of some of those learnings when I look back in review. Please note that the content below does not contain any spoilers about winners and solutions delivered.
The hackathon itself felt like a crescendo. With the announcement of the hackathon and the opening of registrations a few weeks before the event kicked-off, people were instantly connected into the event’s discord server. The hive of activity was similar to previous hackathons – people finding their teams early or coming in as a team; people searching out what the challenges were about; participants signing up to access Oracle Cloud to use as part of their solutions; becoming aware of the environment that will be their home for those 4 days.
In the leadup as there were many questions about the challenges that were on offer to execute against.
I had the privilege to work with the challenge owners Ignacio Guillermo Martínez Rincón – (here) for Theme #1, Wojciech Pluta (here) and Bogdan Farca (here) for Theme #2 and Theme #3 to delve deeper into where the specific problems came from. I was super impressed with the level of detail that they were thinking and the fact that these challenges were very tangible.
Prior to the hackathon itself, Stuart Coggins and I conducted this workshop to help give further context.
Many of the participants and teams formed, included past winners and participants of Hackmakers’ hackathons. This time round, the challenge statements released were very different and very specific. This created a challenge unto itself. Here’s a brief description of what was required of the teams. If you want to read up more about the challenges themselves, you can find them via the links below.
Theme #1 – Data Analytics (here) – this challenge wanted teams to create a model that would predict weather at specific time intervals into the future based upon a specific dataset captured from the F1 2021 video game. The output was a dictionary of weather type and a rain prediction percentage. Teams had to submit their notebook for judging.
Theme #2 – Augmented Reality (here) – this challenge wanted teams to build an Augmented Reality application that interacted with the API data given to them. The dataset came from the F1 2021 video game which consisted of different game sessions of a player racing on different tracks where car telemetry and world positions were captured.
Theme #3 – 3D Modelling (here) – this challenge wanted teams to generate a 3D model of the race track using Blender Python. The dataset was similar to Theme #2 however the focus was on the world positions of a player driving around the track.
Once the teams were formed and moving along, the pace of execution varied with the different challenges. The depth of analysis during the first 24-36hrs was intense and most of it was about the data itself.
#ThankYou Nacho, Wojciech, Bogdan and Stuart for supporting the participants in their data discovery to understand the data.
For Theme #1, it was evident that much of the time was spent understanding the data structures – what data existed; what the data meant; what were the features; what were the determining variables required to predict. For some teams, the volume of data was a struggle – about 700MB of CSV data or 7GB+ of JSON data representing the dataset. Different teams used different techniques and platforms to analyse the data to come up with their analysis and predictive model.
For Theme #2 and Theme #3, the data was a little easier BUT still presented some challenges – the track needs to be drawn; data needed to be streamed into the experience; all based upon data of a racing car driving around the track. The world we live in is imperfect and so is the data and there lies the problem. From some of the analysis that I came across, the driving line that the car took was not always following a certain line. There were cases where car was spinning off the track or at at other times, the car drove straight through corners.
Throughout the weekend, Stuart and Wojciech were creating data for the teams to use driving different lines on different tracks. To demonstrate, here’s one of the laps by Stuart Coggins creating data for the teams.
Let us know in the comments what you think about Stuart’s driving …
At these hackathons, we want people to succeed; we want people to execute and deliver. Hence, the support provided by the diverse group of mentors was definitely on display in many of the different facets of the hackathon – helping the teams understand the challenges; helping the teams think about how to best approach the challenges; helping the teams in resolving conflict and maintaining their mental health throughout the process; helping the teams present their solutions.
#ThankYou to the mentors that gave their time and experience for the benefit of the participants and team – now and in the future.
These events are hard and the effort is not to be under-estimated. There is normal rate of attrition where for whatever reason, some teams don’t submit at the end. Hopefully, whatever the goals that drew people to the event, that they are able to still gain from the experience.
For past week or so, the judges have be judging with the final result being announced next week. Stay tuned at https://www.formulaaihack.com/ to watch the judging unfold on the live stream. Good luck to all of the teams.
RedThunder.Blog 