Smart Drone Delivery Planning

Inspiration

We wanted to build a very general, multi-vehicle path planning simulator, that boasted a unique user experience.

What it does

Allows users to specify, within Apple maps, the location of any number of packages and their corresponding target destinations, as well as any number of drones, each of which can have a custom weight capacity. Once the locations are specified, a path planning algorithm determines how to assign each of the drones to delivery tasks, and a simulation is performed.

How we built it

We used A* with an admissible heuristic initially to solve the path planning algorithm optimally, but ultimately opted to use a more scalable algorithm at the cost of optimality that performs well in practice.

Challenges we ran into

Even with A*, as the number of packages increased beyond just a few, the problem became quickly intractable.

Accomplishments that we're proud of

The algorithm that we designed scales easily with more drones and packages and still provides intelligent coordination between drones.

What we learned

We learned the high cost associated with optimal solutions when dealing with difficult optimization problems.

What's next for Smart Drone Delivery Planning

We want to introduce constraints around the amount of time that a drone can stay in the air before it has to recharge.

Built With

• apple-maps
• swift