Hackatrain : Building a prototype in a weekend with ING and FINN

Last week, the Hackatrain took place in Berlin. Over the span of two days, 17 teams worked on challenges related to autonomous vehicles whereafter they visited the Tech Open Air festival.

ING and FINN team ready to board the Hackatrain!

On the first of July we hopped on the train with a team of five engineers and designers both from ING and FINN : The Autonomous Lions. Our journey was supported by six mentors and colleagues from both companies. Within a day and a half our team worked out a business plan, ironed out a MVP and demonstrated a prototype. The hackathon ended as usual: with a pitch in front of a jury. We didn't bring any prices home :(, but learned a lot along the way and had a great deal of fun. This is our story!

Early morning for the team, still hacking away. Lion onesies are a must have!

Meet our team

The Autonomous Lions consists of:

  • Santiago Aragon Ramirez as Pitch Wizard and Occasion Dealer
  • Seth van Heeringen as Executive Beverage Officer
  • Coen Mooij as Internet of Toys Fleet Manager
  • Julien Lengrand-Lambert Chief Keep It Simple officer and front-end trainee
  • Maaike Koolbergen as Behaviour Consultant and Lead Design Unicorn

The challenges

For this years’ mobility challenges the focus lied on autonomous vehicles and were divided in two tracks linking to the challenge givers:

  • The TNO and Vodafone challenge focused on autonomous trains and the advances that can be made with 5G
  • The BMW and Parkmobile challenge focused on autonomous cars, robotaxi’s and parking.

Our team got assigned to the BMW and Parkmobile challenge. During our preparations on the train we decided to focus on the challenges of parking spaces in dense cities, and the rise of autonomous taxi fleets in the years to come.

We defined two main problems to solve :

  • In our current world, it might take 10 minutes to reach your destination by car, but then another 10 minutes to find a parking spot available, followed by 5 minutes to walk to your destination. This is inefficient!
  • In a future where robo-taxis are a reality, they have to be able to to drop their passengers safely and according to regulations. This challenge is similar to the first point, but in an even more acute way

Several mentors from other companies were also present at the Hackathon, giving us more perspectives to solve problems. Additionally, Parkmobile, BMW and IBM offered their numerous APIs, while FINN provided IOT kits.

Our solution

Our team really wanted to build something concrete and usable in an MVP state. We tried to concentrate of the following requirements :

  • Concept is achievable today, no crazy expectations (flying cars, fleets of robo-taxis available, 30 years in the future, all electric cars, etc.);
  • Concept is profitable quickly, and cheap to set up, no need for external huge investment to make it worthwhile. It should also sustain itself and add or increase existing revenue streams;
  • Concept is useful from day one and helps the transition from a current situation where most vehicles are private (and not electric yet), to a future with a mix of private and autonomous vehicles.

The concept of SpacePods (SPods) was born. Our solution is based on the idea that the demand for electric chargers in cities is exploding and will keep growing for the years to come, especially with cities such as Amsterdam planning on banishing gasoline cars altogether. We decided to add a simple IOT based sensor on existing charging poles to make them 'smarter', while benefiting from their access to network and power, thus remaining cheap.

In short, the SPods would

  • Allow for customers to book parking spots remotely based on their needs. The customer can see directly in Augmented Reality where he will have to park, and how far from the final destination it is.
  • Detect when the car arrived and lock the spot.
  • Detect when the car leaves, automatically reopen the spot for booking and automatically charge the vehicle for parking usage using FINN's IOT device.

In short, we simulated a crude version of Parkmobile 2.0, where our product accompanies a customer through his whole journey from home, to the his destination and then back home with a single press of a button.

However, SPods would pave the way for an autonomous future. Because of their increased capabilities, we can now be much more aware of the usage of our electric parking spots. In a future where robo-taxis are a reality, we can use empty parking spots to load and offload customers and place them in waiting mode. Moreover, when privately owned autonomous vehicles book a spot which is currently occupied by a robo-taxi, we simply send the latter out of the spot towards another available location, thus maximizing throughput. Because our system is fully automated, charging robo-taxis for usage is also not an issue.

Because our solution is data-driven, we also collect knowledge about real-time red zones for traffic. This allows us to automatically redistribute robo-taxi’s to areas where parking is less of a problem, or send them preemptively to where the demand is higher.

A back-office view of the live redistribution of robo-taxi’s to adapt to the demand and reduce congestion

The beauty of our system is that it provides value from day one, starting from the very first device installed, aggregates data and allows for customers to book remotely. All while readying the current infrastructure for a fully autonomous future.

Setup and prototype

Given the time constraints of a hackathon, it is always a challenge to create a prototype that is functional enough to demonstrate the concept, while taking enough shortcuts to actually have a chance to get a working product.

This is what we implemented :

  • A simple Nodejs server that handles requests for bookings and automatically charges customers;
  • A simple front-end simulating an app or the dashboard of a car. The UI is built in plain HTML and CSS, with Mapbox for the geographical awareness;
Minimal geographically aware app to book parking space and show street view location with overlay, using Mapbox
  • An IOT device that detects the presence of a car. The sensor uses a Raspberry Pi with a Python server for communication, a Doppler sensor to detect arrival and departure of a car, and a FINN sensor to charge for parking usage;
IOT sensor with Doppler, FINN and screen setup. Working on detecting car arrival
  • A toy car to simulate our electric BMW parking :).

The pitch

The pitch for this Hackathon was separated into two phases:

  • A technical assessment, to validate the technical concept and ensure that a prototype has been built;
  • An actual, 3 minutes sharp, pitch to explain the concept and business model followed by 2 minutes of technical questions.

It was great to have an actual technical assessment by the BMW and ParkMobile stakeholders. The core objective of a Hackathon is to hack, and too often this assessment is forgotten and the incentive is placed on the pitch. It was great to see interest in our solution and have a couple minutes of technical questions.

Julien pitching the concept to the technical jury, with the team around ready to answer questions

The time constraint for the pitch was pretty hard, and it caught a lot of teams off-guard. It is hard to summarize the last 2 days into 3 minutes! But we managed to get our point across, and got some relevant questions from the jury.

Santiago pitching our idea and answering business questions from the Jury

Conclusion

Hackathons are always a good way to test your creative juices and technical skills under time pressure. For companies, they are also a way to get some fresh perspectives on problems they face every day.

Our team was created especially for the event, and we didn't know each other beforehand. In the end, we completed each other very well and came up with a concept and prototype we are proud of. We all really enjoyed the event, and learnt from it.

Combining software, UI and hardware in a demo in less than 24 hours is challenging, and the various capabilities we had at our disposal really helped us rock the demo.

ING rocking the world, one hackathon at a time

Hackatrain was a very welcoming, organized down to the last details event that we would recommend to anyone. We did not convince the jury in the end, but definitely grew and had a lot of fun along the way.

Maaike and Julien, from Amsterdam with ❤