From the lab


← Back to all Labs

Experiment

Connected Transportation

The startup we were helping is working with smart transportation and wanted to implement a digital platform to drive out inefficiencies and hidden costs in their customers’ current systems, to show what kind of value they can provide with their solutions.

Tuesday 24 September

Confidential startup in transportation/logistics

Connected Transportation

The startup we were helping is working with smart transportation and want to implement a digital platform to drive out inefficiencies and hidden costs in their customers’ current transportation systems, to be able to show what kind of value they can provide with their solutions.

Proof of concept

The idea is to collect real time data from connected trucks and visualize the data in an interactive dashboard. Using an Autopi (www.autopi.io) to collect, the end product would cover a wide array of data including, but not limited to, current position, route, connectivity (latency, uplink, downlink) and vehicle information such as speed, odometer and fuel levels. With the short time frame of 5 days we had to work with, we of course had to limit the amount of data we would visualize. We settled on putting our focus on what the company felt was the most important data to them: Geolocation, route, and connectivity.

Planning

Our connections at the company provided us with the hardware to use for this experiment. Together we briefly discussed which approach to take when implementing it and quickly came up with the following requirements:

  • Autopi for collecting and publishing real time data.
  • Google IOT Core for managing and configuring multiple Autopi devices.
  • Google Big Query for data persistence
  • Pub/Sub in Google Cloud Platform
  • Node.js
  • GraphQL
  • React
  • Mapbox (Open Source alternative to Google Maps)

Implementation

Since we didn’t actually have any trucks to work with, in order to present data that would somewhat reflect real world scenarios, we used a combination of simulations written in Golang that continuously published data from pre-determined routes, and one Autopi connected to a Prototyp Employee car, driving around the island of Kungsholmen in Stockholm to collect live data.

Both live and historic data were then fetched by a small node backend from GCP and further consumed by a React client. With the oh so fantastic Mapbox, we rendered real time location and connectivity status for each of the simulations and the Auto-pi. The historic data was used to draw every truck’s current route and a heat map showing connectivity was laid on top.

Conclusion

After a couple of days of configuring Google Cloud Platform, wrestling with various Autopi API issues, some deep diving into Golang and faulty hardware, we had arrived at fully working MVP.

Looking back, I am very proud of what we managed to put together in such a short period of time, and I hope the people at the startup in transportation/logistics are happy with the end result.