Drone bases surveillance and delivery is an upcoming trend in sectors like construction, oil/gas, disaster management, last mile delivery, healthcare. This project aims to build pathing functionality for it inside of SingleStore.

What it does

The project includes a wasm function that calculates an optmized path for several drones. It does so by first clustering the geo-coordinates then running simulated annealing for solving traveling salesman problem.

Why SingleStore is useful for this problem

since the logic for calculating the paths resides inside of the DB, the calculation is faster as there is no transfer of data involved. Each drone remains agnostic of other drones. SingleStore is able to act like a hive mind for controlling drones.

How we built it

The algorithm for path optimization is built using Rust and made available to SingleStore in form of a WASM UDF. i have used HTML/JS to simulate drones.

Challenges we ran into

This is a highly federated project. I had to centralize it a bit in order to complete it within timelines.

Accomplishments that we're proud of

The complete business logic for calculating the optimized path is residing inside of SingleStore in form of a wasm UDF. This means that each drone is oblivious of the other drones. SingleStore acts as a hive mind for all the drones.

What we learned

I learned to write Rust and compile it into a web assembly module.

What's next for Drone surveillance/delivery on SingleStore

  1. Performance optimization of the Rust algo.
  2. Integration with Kafka.
Share this project: