This project started when I, along with my friends, decided to participate in Walmart Sparkathon 2025. We chose the theme : "Transforming retail supply chains: From inventory management to last-mile delivery".
For the first round, we decided to start simple. Instead of simulating whole of the world's most complex logistics, we focused on a few basic components: distribution centers, cross-docks, retail stores, trucks, vans, drones, forklifts, and individual workers.
Using React Three Fiber, we created immersive 3D environments for warehouses and dispatch centers, while building a stylized SVG city map for route visualization. Every vehicle was clickable, showing its cargo, destination, ETA, and the ML decisions affecting its route.
The USP was the real-time ML route optimization. We implemented mock ML APIs (just for the sake of the hackathon) that continuously analyze traffic patterns, predict congestion, and reroute vehicles dynamically. You could literally watch a truck change its path when the algorithm detected a better route.
The idea was that in just 2-3 minutes, judges could see trucks moving through a city, force congestion to trigger ML rerouting, click on facilities to see operations, and watch live dashboards update. We wanted to play smart on the term digital twin but just like every other hackathon, we never made it to the next round!