Many different companies are actively preparing for smart mobility revolution driven by technological advancements. Even companies that are a part of very traditional industries — GS Caltex (South Korea’s largest oil refiner) and Doosan (multinational conglomerate most famous for its heavy industries product / construction) are also showing active engagement in this growing industry and organized a hackathon at Seoul mobility expo with a very interesting topic. The topic was: Visualize a futuristic gas-station and make a prototype that could facilitate the use of it.
How can gas-stations be futuristic? Currently, they are just places for charging cars. Maybe in a couple years the charging source may change to electricity/hydrogen from oil, but what else can fundamentally change? GS Caltex presented their vision of futuristic gas-stations at CES 2021, and we got inspirations from their pitch video.
Leveraging the advantage coming from abundance of gas-stations in critical locations, GS Caltex wants to turn them into a logistic/mobility hub. A place where people could have easy access to micro mobility services and pick up packages delivered by drones.
So…. What did we make?
We really tried to think in GS Caltex’s perspective. There’s not a single futuristic gas-stations in the world yet. Therefore, we thought making some mundane apps that are more relevant today than in the future might not be something they are looking for. We tried to incorporate a lot of visually futuristic elements and something that has a “wow” factor.
Finally, we came up with the idea of AI human walk-through customer support panel. To enhance the quality of service at futuristic gas-stations, we thought an effective customer support was necessary. Instead of hiring more people, we thought Artificial Intelligence could do the job.
We made two different prototypes. One was 3d rendered visualization of our future gas station. The other was an interactive web service that allowed users to communicate with our AI.
How to make a simple AI walkthrough panel
Step 1: Find a model, rent a chroma key studio, and record phrases and movements.
We decided to use our tallest team member as our AI model. In a couple steps, he will turn into an AI.
Step2: Design a rule-based chatbot and find a Speech-to-Text API.
Based on prepared scenarios, we made a chat bot that recognizes certain keywords. We used Google Cloud’s Speech-to-Text API to convert speech to text and fed this converted text into our chat bot to output different video urls. We used a Socket to send audio stream data to the server for STT.
Step3: Remove green background from the recorded videos
We converted green regions as a masked area, and made the background transparent.
Step4: Add cool backgrounds and transition effects!
To make it more interactive, we added entertaining backgrounds and transition effects based on different scenario.
Step5: Add a Face Recognition Tool to track user movements.
Ultimately, a walk through AI support must follow where the customer goes. Therefore, we added a face recognition tool so that our AI human can appear when the customer arrives and follow him/her around.
You can check the final result here. It only supports conversations in Korean, but we were considering multi-language support as well if we had more time.
3D Rendered Futuristic Gas-Station
While the software engineers in our team prepared the above interactive prototype, our designer produced an amazing 3d rendered visualization below.
Conclusion
Overall, I had so much fun during this hackathon. The prototype we made can not be in service yet, but making it was a great opportunity for us to explore variety of new technologies. This motivated us to add more features and make it more interactive. We were all very satisfied with what we made in three days and fortunately the judges felt the same way as well.
I would like to thank all my team members for this amazing experience!
