Encore: The Show Must Go On

Drone light shows are impressive when everything goes to plan. Hundreds of drones fly in sync to create shapes in the sky but if one fails mid-show, it leaves a gap. In a typical pre-programmed show, the other drones keep following their assigned paths rather than moving to fill it.

We wanted to see if we could change that.

Introducing Encore

Encore is a 3D simulation of a decentralised, self-healing drone swarm. Instead of relying on a central controller to respond to every failure, each drone uses local sensing and radio messages to work out what is happening around it and decide how to respond.

When a drone fails, the swarm follows four steps:

  1. Detect: Nearby drones check whether a formation position is empty using physical sensing and missed heartbeats.
  2. Decide: Eligible drones broadcast claims and use shared priority rules to work out who should fill the gap.
  3. Recover: The winning drone moves into the empty position, prioritising the parts of the formation that matter most.
  4. Adapt: Other drones can fill the positions left behind, allowing the formation to reorganise over time.

We also use gossip-based communication to share new formations. An update can start with one drone and spread through its neighbours, without needing a message from a central broadcaster to reach everyone at once.

Putting It to the Test

We built an interactive simulation with 200 autonomous drone agents. You can trigger failures, disrupt radio communication and split the swarm into disconnected groups to see how it responds.

We also compare Encore with a static, pre-programmed baseline under the same conditions, so we can measure whether self-healing actually helps preserve the formation.

Encore can't replace drones that have failed, and it won't handle every situation perfectly. The goal is to show how local decisions and simple communication rules can help a swarm adapt when things don't go to plan.

One drone fails. The show goes on.

Share this project:

Updates

Submission history